V/MgO/Fe Junctions: Interface Engineering

• V/MgO/Fe junctions are epitaxial oxide/metal heterostructures that use a MgO barrier to couple electrodes, enabling coherent spin-dependent tunneling and precise anisotropy modulation.
• Misfit engineering via vanadium alloying in the Fe electrode reduces dislocation density, leading to enhanced tunnel magnetoresistance and suppressed low-frequency 1/f noise.
• The platform supports versatile functionalities, including superconductivity-assisted magnetization reorientation and oxygen migration–driven resistive switching for advanced device applications.

V/MgO/Fe junctions are epitaxial oxide/metal heterostructures in which a MgO barrier or interface couples vanadium, iron, or Fe–V alloy electrodes and produces a set of interfacial phenomena spanning coherent spin-dependent tunneling, low-frequency noise suppression, competing in-plane and perpendicular anisotropies, superconductivity-assisted magnetization reorientation, and electrically driven resistive switching. Within this family, the literature includes single-barrier Fe$_{1-x}$V$_x$/MgO/Fe magnetic tunnel junctions, V/MgO/Fe units embedded in double-barrier stacks, and the closely related inverted Fe/MgO/V/Fe geometry used to isolate oxygen motion at the MgO/V boundary (Herranz et al., 2010, González-Ruano et al., 2021, Martínez et al., 2018, 1402.2517).

1. Epitaxial realizations and structural motifs

The reported junctions are grown epitaxially on single-crystal MgO substrates by molecular-beam epitaxy. In the Fe$_{1-x}$V$_x$/MgO/Fe magnetic tunnel junctions studied by Herranz et al., the substrate is MgO(001), outgassed at 875 K, followed by a 7.5 nm MgO seed layer, a 50 nm bottom electrode of Fe$_{1-x}$V$_x$ or Fe, a 2.0 nm MgO barrier corresponding to 9.5 $\pm$ 0.5 monolayers, a top electrode of 18 nm Fe or Fe$_{1-x}$V$_x$, and a 20 nm Co/20 nm Au cap (Herranz et al., 2010). In the superconducting and spin-reorientation studies, the representative stack is MgO(001)/MgO(10 nm)/V(40 nm)/MgO(2 nm)/Fe(10 nm)/MgO(2 nm)/Fe(10 nm)/Co(20 nm), with all layers grown epitaxially and annealed in situ at 450 $^\circ$C for 20 min in the 2021 work, and room-temperature MBE with post-barrier in-situ flattening in the 2018 work (González-Ruano et al., 2021, Martínez et al., 2018).

Representative stack	Structural feature	Reported phenomenon
Fe$_x$0V$_x$1/MgO/Fe	Reduced FeV/MgO misfit	Suppressed 1/f noise and enhanced TMR
V/MgO/Fe/MgO/Fe/Co	Competing anisotropies in Fe(10 nm)	Broken-symmetry spin reorientation
V/MgO/Fe(001)	Superconducting V below $_x$2	Enhanced effective PMA
Fe/MgO/V/Fe	MgO/V boundary with movable O ions	Resistive switching via oxygen migration

The epitaxial relations are central. For Fe/MgO-based tunnel junctions, the registry is Fe(001) $_x$3 MgO(001) and Fe[110] $_x$4 MgO[100], while the V/MgO/Fe structures are described as cube-on-cube, with $_x$5 and $_x$6 (Herranz et al., 2010, González-Ruano et al., 2021). RHEED and low-energy electron diffraction show 1$_x$71 surface patterns in the 2021 study, and cross-sectional TEM and X-ray reflectivity indicate interface roughness below 0.5 nm and low dislocation densities below $_x$8 (González-Ruano et al., 2021). In the 2018 double-barrier structures, RHEED confirmed one-monolayer roughness or better, and ex-situ AFM showed root-mean-square roughness $_x$9 nm (Martínez et al., 2018).

These structural motifs define the operating regime of the junctions. MgO functions simultaneously as a tunnel barrier, a symmetry filter, and a source of interfacial anisotropy, while V can act either as a nonmagnetic bcc template, a superconductor below $_{1-x}$0–5 K, or an alloying element that tunes the Fe/MgO lattice misfit.

2. Misfit engineering, dislocations, and the Fe–V alloy route

In Fe$_{1-x}$1V$_{1-x}$2/MgO/Fe junctions, vanadium alloying is used to reduce the mismatch between the bottom electrode and MgO. The relevant in-plane lattice constants are $_{1-x}$3 nm and $_{1-x}$4 nm, while Vegard’s law implies that the Fe$_{1-x}$5V$_{1-x}$6 lattice parameter increases nearly linearly with $_{1-x}$7. The lattice misfit is written as

$_{1-x}$8

As $_{1-x}$9 increases from 0 to 0.25, $_x$0 drops from about 3.9% toward about 2.5%, and the critical thickness for plastic relaxation rises from about 5 ML at $_x$1 to about 10 ML at $_x$2 (Herranz et al., 2010).

Because the MgO barrier thickness is 9.5 ML, this places the barrier near the crossover between coherent elastic accommodation and dislocation-mediated relaxation. The paper attributes the observed transport changes to the fact that V doping up to about $_x$3–0.2 substantially reduces the density of misfit-accommodating dislocations in the barrier. Dislocation nucleation was quantified from RHEED measurements of the MgO in-plane lattice parameter during growth, with the sudden change in spacing between the (220) and $_x$4 rods used to locate the critical thickness. Under complete relaxation, the average spacing between dislocations along [100] is estimated as

$_x$5

so reduced misfit increases $_x$6 and decreases the dislocation density $_x$7; high-resolution TEM confirmed the defect reduction for $_x$8 up to about 0.2 (Herranz et al., 2010).

The significance of this alloy route is twofold. First, it provides a structurally controlled method for changing defect statistics without altering the MgO barrier thickness from the regime where coherent tunneling remains relevant. Second, it shows that the position of V matters: the strongest improvements were reported when the bottom electrode, rather than the top electrode, was alloyed. This suggests that misfit control is most consequential at the interface where the MgO barrier nucleates and relaxes.

3. Magnetotransport and low-frequency noise

The principal transport observables in Fe$_x$9V$_{1-x}$0/MgO/Fe junctions are the tunnel magnetoresistance and the low-frequency $_{1-x}$1 noise. The TMR is defined in the standard form

$_{1-x}$2

with $_{1-x}$3 and $_{1-x}$4 the parallel and antiparallel resistances, respectively. At room temperature, a pure Fe/MgO/Fe junction showed TMR $_{1-x}$5. When the bottom electrode was Fe$_{1-x}$6V$_{1-x}$7, the TMR increased with $_{1-x}$8, reached about 207% at $_{1-x}$9, and then decreased for $_x$0 (Herranz et al., 2010).

The non-monotonic dependence was explicitly assigned to a competition between two effects: reduced barrier strain and fewer dislocations, which enhance coherent $_x$1-channel tunneling through the MgO symmetry filter, and alloying-induced reduction of the spin polarization in Fe$_x$2V$_x$3, which depresses TMR at larger $_x$4. This is an important corrective to a common oversimplification: V incorporation does not improve TMR indefinitely, and the alloy content has an optimum rather than a monotonic benefit.

The low-frequency noise follows the Hooge form

$_x$5

where $_x$6 is the voltage-noise power spectral density, $_x$7 the DC bias, $_x$8 the junction area, $_x$9 the frequency, and $\pm$0 the Hooge parameter. In the parallel state, the reference Fe/MgO/Fe junction had $\pm$1–$\pm$2 in units normalized to $\pm$3. Doping the bottom Fe electrode with 8–16% V reduced $\pm$4 by almost two orders of magnitude, down to $\pm$5–$\pm$6, and a similar suppression was observed in the antiparallel state $\pm$7 (Herranz et al., 2010).

Herranz et al. attributed the simultaneous enhancement of TMR and suppression of both nonmagnetic and magnetic $\pm$8 noise to strongly reduced misfit and dislocation density. The interpretation is that fewer tunneling-active defects remain available for charge trapping and detrapping. For device design, the paper singled out $\pm$9–0.12 in the bottom Fe electrode as the optimum range, with the additional recommendation that the MgO thickness be kept near, but not much above, the critical thickness for plastic relaxation (Herranz et al., 2010).

4. Competing anisotropies and broken-symmetry reorientation

In V/MgO/Fe-based double-barrier junctions, the Fe(10 nm) layer above the lower MgO barrier occupies a regime where shape anisotropy and interfacial perpendicular magnetic anisotropy compete. Martínez et al. described the full stack as V(40 nm)/MgO(2 nm)/Fe(10 nm)/MgO(2 nm)/Fe(10 nm)/Co(20 nm), with the lower MgO barrier grown on V and the upper MgO barrier grown on Fe. The V(001)$_{1-x}$0MgO(001) mismatch was given as 2%, while MgO(001)$_{1-x}$1Fe(001) was 4%; correspondingly, the lower barrier relaxes less and has somewhat higher crystalline quality than the upper one, and transport data indicated that the upper barrier is about four times more transparent than the bottom barrier (Martínez et al., 2018).

The relevant magnetic parameters were reported as $_{1-x}$2, $_{1-x}$3, and cubic anisotropy $_{1-x}$4, while the demagnetizing anisotropy is

$_{1-x}$5

At each MgO/Fe interface, the surface anisotropy was written as

$_{1-x}$6

and the total anisotropy energy density as

$_{1-x}$7

Simulations indicated that the onset of a field-driven in-plane to out-of-plane flip in a 10 nm Fe layer requires $_{1-x}$8 on the order of $_{1-x}$9–$_x$0, and that strong hysteresis asymmetry appears when $_x$1 (Martínez et al., 2018).

Experimentally, above about 50 K the soft Fe layer is in-plane at zero field. Below about 50 K, however, a brief excursion to $_x$2 kOe drives the magnetization from an in-plane minimum into a new out-of-plane minimum, and the layer remains non-volatile at zero field with $_x$3. The reported switching fields were $_x$4 kOe for the in-plane to $_x$5 transition and $_x$6–1.2 kOe for the return through the in-plane state toward $_x$7 (Martínez et al., 2018).

The broken symmetry of this reorientation is one of the defining features of the system. Micromagnetic simulations in MuMax3 on a $_x$8 block reproduced the one-way flip and non-volatile perpendicular state when the bottom and top interfaces were assigned similar first-order surface anisotropies, $_x$9, but a substantial negative second-order term only at the top interface, $^\circ$0, with $^\circ$1 (Martínez et al., 2018). A plausible implication is that asymmetry between the two MgO/Fe interfaces, rather than PMA alone, is required to explain the observed remanent states and hysteresis polarity.

5. Superconductivity-assisted modification of perpendicular magnetic anisotropy

When the V layer becomes superconducting, the V/MgO/Fe interface acquires an additional anisotropy contribution mediated by spin-orbit coupling and superconducting proximity. In the 2021 study, the anisotropy per unit volume of the Fe layer was written as

$^\circ$2

with $^\circ$3 nm, $^\circ$4, and $^\circ$5 per MgO/Fe interface, giving $^\circ$6 and therefore $^\circ$7, or slightly in-plane easy above 80 K. The bottom Fe/V interface was described as having negligible interfacial anisotropy (González-Ruano et al., 2021).

The additional superconducting contribution was formalized as

$^\circ$8

with $^\circ$9 turning on below $_x$00. The microscopic interpretation given in the paper is that Rashba SOC at the V/MgO/Fe interface converts singlet Cooper pairs into triplets most efficiently when the magnetization is in-plane, so an out-of-plane orientation reduces pair breaking and lowers the proximity-induced free-energy cost. In the Bogoliubov–de Gennes modeling, the free-energy difference

$_x$01

decreases below $_x$02, and the field needed to rotate the magnetization out of plane,

$_x$03

is correspondingly reduced (González-Ruano et al., 2021).

The experimental signature is strongly size dependent. For 10$_x$0410 $_x$05 junctions, $_x$06 dropped from about 1.5 kOe at 5 K to about 0.2 kOe at 0.3 K. For 20$_x$0720 and 30$_x$0830 $_x$09, the reduction was only about 10–20%, while for 40$_x$1040 $_x$11 it was essentially unchanged. The zero-field relative angle $_x$12 between soft and hard layers increased from about $_x$13 at 5 K to about $_x$14 at 0.3 K in 10$_x$1510 $_x$16 junctions, fell to about $_x$17 reorientation for 30$_x$1830 $_x$19, and vanished in 40$_x$2040 $_x$21 devices (González-Ruano et al., 2021).

The paper also addressed an alternative explanation based on vortex stray fields and Meissner screening. Magnetostatic and time-dependent Ginzburg–Landau simulations indicated that realistic Meissner contributions below 7% of the applied field should slightly increase $_x$22 in large junctions, and therefore cannot explain the large decrease observed in the smallest samples. The conclusion was that stray-field coupling is a minor contributor compared with the SOC-assisted proximity mechanism (González-Ruano et al., 2021). This directly resolves a potential misconception: the low-temperature anisotropy change is not adequately accounted for by superconducting flux effects alone.

Below $_x$23, an additional small out-of-plane reorientation was also observed at $_x$24 Oe under $_x$25 mV in 10$_x$2610 and 20$_x$2720 $_x$28 junctions, with 600 mV corresponding to $_x$29 across the V/MgO/Fe interface. Above $_x$30, $_x$31 was unchanged by bias polarity or magnitude (González-Ruano et al., 2021). This suggests that the control space is intrinsically multivariable, involving superconducting phase, Rashba SOC, lateral size, and electric field.

6. Oxygen migration and resistive switching at the MgO/V interface

A closely related inverted geometry, Fe/MgO/V/Fe, has been used to analyze resistive switching at the MgO/V boundary from first principles. The ab initio study modeled epitaxial Fe(001)/MgO(001)/V/Fe(001) in bcc alignment with $_x$32 and $_x$33. Fe atoms sit atop O atoms at the Fe/MgO interface with $_x$34, and two MgO/V registries were considered: V on top of O and V on top of Mg (1402.2517).

The essential mechanism is oxygen-ion displacement perpendicular to the interface. The total-energy calculations exhibited two minima, a ground state at $_x$35 and a metastable state at $_x$36. In a large 215-atom cell with V atop O and one O moved out of nine, the metastable state occurred at $_x$37 with barrier $_x$38 meV and metastable depth $_x$39 meV. With a V vacancy in front of the migrating O, the barrier was about 17 meV at $_x$40; for V atop Mg in the small cell, the barrier was about 66.5 meV with metastable depth about 4 meV (1402.2517). The paper defined

$_x$41

Full ionic relaxation markedly reduces the barrier. In the small-cell V-on-O case, allowing relaxation lowered $_x$42 from 0.37 eV to about 0.05 eV, which the authors described as demonstrating the importance of local screening. For field-driven switching, the condition was stated as

$_x$43

Using $_x$44 and $_x$45 meV gives a critical field $_x$46, whereas a typical bias of about 1 V across about 1 nm yields $_x$47, sufficient to activate O migration (1402.2517).

The electronic consequences depend on interface registry. For V atop O, the majority-spin V $_x$48 density of states at the Fermi level increased from about 0.05 to about 0.15 states/eV after oxygen migration, and the zero-bias conductance rose from $_x$49 to $_x$50, giving $_x$51. For V atop Mg, the same displacement decreased the DOS from about 0.20 to about 0.08 states/eV and reduced the conductance from $_x$52 to $_x$53, so $_x$54 (1402.2517).

Charge transfer accompanies the conductance change. The charge-density difference $_x$55 shows Friedel oscillations centered on the O impurity after it enters the V layer, and Mulliken analysis indicates interface-plane charge rearrangements such as $_x$56 in the first V layer and $_x$57 in the adjacent MgO layer for the V-on-O case (1402.2517). Within the scope of the reported calculations, this establishes the MgO/V interface as a memristive element whose polarity and ON/OFF contrast are controlled by local registry and oxygen coordination.

7. Functional scope and design implications

Across the reported realizations, V/MgO/Fe junctions are defined less by a single functionality than by a common interfacial control principle. In Fe$_x$58V$_x$59/MgO/Fe, modest V alloying of the bottom electrode reduces misfit, suppresses dislocation-mediated $_x$60 noise by nearly two decades, and increases room-temperature TMR from about 185% to about 207%, but larger $_x$61 degrades TMR because spin polarization is reduced and chemical disorder becomes more important (Herranz et al., 2010). In V/MgO/Fe/MgO/Fe/Co, the V-supported lower MgO barrier provides a high-quality MgO/Fe interface that contributes strong first-order PMA, while asymmetry between the two MgO/Fe interfaces can generate a metastable perpendicular state and a broken-symmetry switching path (Martínez et al., 2018).

Below the superconducting transition of V, the same interface family acquires a new control channel: the effective PMA can be enhanced through SOC-mediated ferromagnet–superconductor interaction, with the magnitude of the effect set by lateral junction size and modulated by electric field (González-Ruano et al., 2021). In the inverted Fe/MgO/V/Fe geometry, oxygen motion across MgO/V yields metastable resistive states accessible at electric fields compatible with typical device biases (1402.2517).

Taken together, these results delimit the present understanding of the platform. The robust conclusions are that epitaxy, lattice mismatch, interface registry, and interfacial SOC are not secondary details but primary variables. A plausible implication is that V/MgO/Fe junctions should be viewed as a broader interface-engineering class rather than a single device archetype: depending on whether V is introduced as an alloying component, a superconducting electrode, or an oxygen-accepting interfacial layer, the same MgO-based architecture can be tuned toward low-noise spin filtering, multistate anisotropy control, or resistive switching.