Greatly enhanced tunneling electroresistance in ferroelectric tunnel junctions with a double barrier design

Ferroelectric tunnel junctions (FTJs) are nanoscale metal/ferroelectric/metal devices that exhibit tunneling electroresistance (TER)—a conductance difference between opposite ferroelectric polarization states—making them attractive for non-volatile memory. Prior strategies to enhance TER in single-barrier (SB) FTJs have focused on modulating effective barrier height/width via screening, defects, interfacial polar layers, phase transitions, and domain-wall states. Here, the hypothesis is that employing a double-barrier (DB) structure—two ferroelectric barriers in series—can greatly amplify TER due to the multiplicative dependence of overall transmission on the individual barrier transmissions in sequential or resonant tunneling regimes. Simple transport models (sequential, incoherent, and coherent resonant tunneling) predict the ON/OFF (TER) ratio for a DB structure can exceed that of an SB structure by orders of magnitude across broad parameter ranges. This motivates designing and evaluating a DB-FTJ comprising two identical Pt/BaTiO3/LaAlO3/Pt units in series.

The work builds on extensive research into FTJs for non-volatile memory. Reported TER-enhancement mechanisms in SB-FTJs include electrode screening effects, domain-wall conduction, interfacial defects, ferroelectrically induced phase transitions, polar interlayers, and large interfacial polarizations. Mesoscopic transport theory on double-barrier systems highlights phenomena absent in single barriers, such as resonant tunneling and single-electron tunneling; importantly, the composite transmission depends nonlinearly on the individual barrier transmissions, suggesting a route to enhanced ON/OFF ratios when adapted to FTJs. Prior studies also show polar oxide interlayers (e.g., LaAlO3) can improve TER in M/F/M junctions, motivating their inclusion here.

Devices modeled: (1) DB-FTJ: Pt/BaTiO3 (BTO, 5.5 u.c.)/LaAlO3 (LAO, 1 u.c.)/Pt/BTO (5.5 u.c.)/LAO (1 u.c.)/Pt, effectively two identical SB units in series; (2) SB-FTJ with BTO thickness 5.5 u.c.; (3) SB-FTJ with BTO thickness 11.5 u.c. LAO is inserted at the right side of each single barrier as a polar interlayer. Interface terminations: O in TiO2 planes positioned atop Pt atoms at Pt/BTO and O in AlO2 plane atop Pt at Pt/LAO; BTO/LAO follows AO–BO2 stacking appropriate for perovskites. - Structural relaxation: SIESTA DFT with GGA-PBE and DZP basis; 5×5×1 k-grid; in-plane lattice fixed to tetragonal bulk BTO (a = 3.991 Å); out-of-plane lattice and all atomic positions optimized until forces < 0.01 eV/Å. Polarization direction set via initial Ti–O displacements; resulting BTO shows tetragonal c/a > 1 with z-directed polarization. - Transport: NEGF-DFT using Nanodcal. Self-consistent k-grid 15×15×1; transmission computed on 300×300 k-grid over transverse 2D Brillouin zone. Equilibrium conductance G computed as integral of transmission at EF over k||; TER defined as (G+ − G−)/min(G+, G−) × 100%. Real-space charge density obtained by integrating LDOS from EF − 0.1 eV to EF. To assess band-gap underestimation, DFT+U tests with U = 5 eV and 8 eV on Ti d orbitals were performed. - Analysis: k-resolved transmission maps, layer- and orbital-resolved PDOS, Ti–O layer displacements, electrostatic potential profiles, and orbital/k-resolved DOS for selected atoms were examined to identify conduction channels and interfacial metallization.

- DB-FTJ electronic/transport behavior: - k-resolved transmission shows a high-transmission cross near Γ for the right-polarized state; left-polarized state exhibits uniformly low transmission. - Equilibrium conductances: left state G ≈ 9.978 × 10^-12 (2e^2/h); right state G ≈ 2.205 × 10^-5 (2e^2/h). - TER ≈ 2.210 × 10^8% (ratio ≈ 2.210 × 10^6), at least three orders of magnitude larger than comparable SB-FTJs. - High-conductance state RA ≈ 0.093 kΩ·µm^2 (ultra-low, CMOS-friendly). - Origin: In right polarization, positive FE bound charge plus (LaO)+ at each barrier’s right interface lowers electrostatic potential, bringing CBM below EF locally (interfacial metallization), effectively narrowing each barrier. Conduction mainly via Ti d_yz and d_xz orbitals aligned with transport; localized d_xy contributes weakly. In left polarization, EF lies largely in the gap with only weak localized d_xy near the left interface, yielding low conductance. - DFT+U (U = 5 or 8 eV) retains interfacial metallization in the right-polarized state, supporting robustness of the mechanism. - SB-FTJ (BTO 5.5 u.c.): - Conductances: right G ≈ 7.301 × 10^-5; left G ≈ 1.373 × 10^-7 (2e^2/h). - TER ≈ 5.306 × 10^3. - High-state RA ≈ 0.028 kΩ·µm^2 (same order as DB-FTJ’s high state). - SB-FTJ (BTO 11.5 u.c.): - Conductances: right G ≈ 1.283 × 10^-10; left G ≈ 1.510 × 10^-14 (2e^2/h). - TER ≈ 8.496 × 10^5% (still ~3 orders smaller than DB-FTJ). - High-state RA ≈ 16.002 MΩ·µm^2 (too large for practical detection). - Multiple resistance states in DB-FTJ: - Independent control of the two FE barriers yields four states (P↑↓, P↓↓, P↑↑, P↓↑). For head-to-head and tail-to-tail states: conductances ≈ 5.492 × 10^-8 and 1.974 × 10^-7 (2e^2/h), RA ≈ 37.407 kΩ·µm^2 and 10.407 kΩ·µm^2, respectively. - The four DB-FTJ resistance states span a wide RA range and are well-separated compared to SB-FTJs. - Role of quantum well states: The middle metal layer forms QW states that can enhance transmission for both polarities, but the dominant TER enhancement arises from the double-barrier effect (overall T ≈ Ta·Tb).

The results validate the central hypothesis that a double-barrier design substantially amplifies TER in FTJs. By placing two identical ferroelectric barriers in series with a thin metallic spacer, the overall transmission in opposite polarization states differs much more than in single-barrier devices due to the multiplicative dependence on individual barrier transmissions, consistent with mesoscopic transport theory. Microscopically, the polar (LaO)+ interlayer and ferroelectric bound charges shape the electrostatic potential to induce interfacial metallization in the right-polarized state, effectively narrowing the barrier and opening strong transmission channels through Ti d_yz/d_xz orbitals, while the opposite polarization keeps EF within the gap, suppressing conduction. The DB-FTJ simultaneously achieves a giant TER and an ultra-low RA in the ON state, enhancing readout margins and compatibility with CMOS. Moreover, independent polarization control across the two barriers enables four distinct, well-separated resistance states, suggesting multi-level memory capability. While quantum well resonances in the spacer contribute to transmission, they do not undermine the TER amplification, which primarily stems from the double-barrier architecture. The approach outperforms simply thickening a single barrier, which dramatically raises RA and hampers detectability despite moderate TER increases.

A Pt/BaTiO3/LaAlO3/Pt/BaTiO3/LaAlO3/Pt double-barrier FTJ, modeled via first-principles NEGF-DFT, exhibits an ultra-low ON-state RA (~0.093 kΩ·µm^2) and a giant TER (~2.210 × 10^8%), outperforming single-barrier counterparts by at least three orders of magnitude in TER. The enhancement arises from the intrinsic double-barrier transmission effect and interfacial potential engineering that induces local metallization in one polarization state. The DB architecture also enables four programmable resistance states via independent polarization control, promising multi-level non-volatile memory applications. Future work could explore defect/disorder engineering and interlayer/interface optimization to further boost TER, and experimental validation of the proposed DB-FTJ structures.

- First-principles calculations (GGA-PBE) underestimate band gaps; DFT+U tests (U = 5, 8 eV on Ti d) indicate the key interfacial metallization persists, but quantitative values may vary. - The study is computational; experimental realization may introduce defects/disorder and interface roughness that can enhance or suppress TER. - Very thick single-barrier designs yield impractically large RA, limiting detectability; practical device optimization must balance barrier thickness and RA. - Coherent versus incoherent transport regimes and phase effects could vary with device fabrication and temperature, potentially modifying quantitative TER gains.