BaTiO3/AlN Field-Plate in Power Diodes

• The BaTiO3/AlN field-plate concept uses BaTiO3’s high permittivity for effective electric-field redistribution while AlN offers superior thermal conductivity and dielectric robustness.
• Research shows that strategic stacking improves breakdown voltage and reduces localized hotspots, as demonstrated in vertical β-Ga2O3 Schottky diodes with optimized dielectric configurations.
• Key challenges include precise interface engineering, maintaining favorable band alignments, and adapting the concept to Al-rich nitride devices without compromising device reliability.

A BaTiO$_3$/AlN field-plate is a stack-dielectric edge-termination concept in which BaTiO$_3$ provides very high permittivity for electric-field redistribution, while AlN functions as an interfacial insulator chosen for thermal conductivity, dielectric robustness, and band-offset advantages. In the arXiv literature, the term does not denote a single canonical device family. The closest early nitride analogue is a lateral Pt/BaTiO$_3$/Al$_{0.58}$Ga$_{0.42}$N heterojunction diode, where BaTiO$_3$ under the anode and access region acts as a field-management dielectric, although the device is not a literal BaTiO$_3$/AlN field plate (Razzak et al., 2019). A later vertical GaN-on-GaN PN diode used an Al$_2$O$_3$/BaTiO$_3$ high-$_3$0 region together with guard rings, spin-on-glass, and a metal field plate, again without AlN (Xu et al., 2023). An explicit BaTiO$_3$1/AlN field-plate was then developed for vertical $_3$2-Ga$_3$3O$_3$4 Schottky barrier diodes as an electro-thermal co-design strategy that addresses both high-field crowding and self-heating (Audri et al., 15 Aug 2025).

1. Scope and nomenclature

The expression “BaTiO$_3$5/AlN field-plate” can refer either to a literal dielectric stack under a metal field plate or, more loosely, to a broader class of high-$_3$6-assisted field-management structures. The literature represented here spans both meanings. In Al-rich nitride devices, BaTiO$_3$7 first appears as an extreme-permittivity dielectric beneath an anode in a lateral heterojunction diode, where it suppresses anode-edge field crowding and leakage-assisted breakdown; functionally this is a field-plate analogue rather than a conventional overhanging metal field plate (Razzak et al., 2019). In vertical GaN power diodes, BaTiO$_3$8 appears as part of a local Al$_3$9O$_3$0/BaTiO$_3$1 high-$_3$2 stack positioned near the anode and field-plate edge inside a multi-element termination that also includes guard rings and SoG (Xu et al., 2023). In vertical $_3$3-Ga$_3$4O$_3$5 Schottky diodes, the term becomes literal: the field plate is formed with a dielectric stack ordered as metal/BaTiO$_3$6/AlN/$_3$7-Ga$_3$8O$_3$9 (Audri et al., 15 Aug 2025).

A recurrent misconception is that the nitride literature already demonstrates a BaTiO$_{0.58}$0/AlN field-plate on AlN. The cited nitride papers do not do so. One uses Al$_{0.58}$1Ga$_{0.58}$2N on an AlN template rather than an AlN active region, and the other uses Al$_{0.58}$3O$_{0.58}$4, not AlN, as the interfacial dielectric. The explicit BaTiO$_{0.58}$5/AlN implementation is instead reported for $_{0.58}$6-Ga$_{0.58}$7O$_{0.58}$8 Schottky diodes, where the motivation is explicitly electro-thermal rather than purely electrostatic (Razzak et al., 2019).

2. Electrostatic principle of high-$_{0.58}$9 field management

The field-plate problem in lateral and vertical power devices is usually not set by the average field across the drift or access region, but by a localized maximum at a metal edge, junction corner, or termination discontinuity. BaTiO$_{0.42}$0 is used because a very large dielectric constant changes the boundary conditions governing the normal component of electric displacement, expressed as $_{0.42}$1, and at an interface as $_{0.42}$2. In the Al$_{0.42}$3Ga$_{0.42}$4N lateral heterojunction diode, the paper describes BaTiO$_{0.42}$5 as an extreme dielectric constant material and estimates the deposited films to have $_{0.42}$6; its introductory electrostatic comparison between $_{0.42}$7 and $_{0.42}$8 shows that the high-$_{0.42}$9 case yields a much flatter lateral field profile (Razzak et al., 2019).

For breakdown analysis, the relevant quantity is often the average breakdown field,

$_3$0

because it measures how effectively the device uses the available lateral distance before local field peaking triggers failure. In the BaTiO$_3$1-assisted Al$_3$2Ga$_3$3N diode, a device with $_3$4 and $_3$5 reached $_3$6, whereas a control Pt/Al$_3$7Ga$_3$8N Schottky diode with $_3$9 broke down at $_3$0 (Razzak et al., 2019). The physical interpretation offered is twofold: reduced peak electric field at the anode edge and suppression of gate/anode leakage-related breakdown.

The explicit BaTiO$_3$1/AlN stack in $_3$2-Ga$_3$3O$_3$4 refines this idea. There the stack does not simply lower the field everywhere; it redistributes the field so that the peak is shifted out of $_3$5-Ga$_3$6O$_3$7 and into AlN. Under $_3$8 reverse bias, the planar BaTiO$_3$9-only field-plate shows a peak field on the order of $_2$0 in $_2$1-Ga$_2$2O$_2$3, whereas the BaTiO$_2$4/AlN stack produces about $_2$5 in AlN. This matters because the paper cites a predicted AlN critical breakdown field of $_2$6 and an experimentally extracted AlN breakdown field of about $_2$7, both substantially above the $_2$8-Ga$_2$9O$_3$0 field scale of interest (Audri et al., 15 Aug 2025). A plausible implication is that the most effective dielectric stack is not necessarily the one that minimizes the absolute peak field, but the one that relocates the peak into the most robust layer.

3. Al-rich nitride precursor: the Pt/BaTiO$_3$1/Al$_3$2Ga$_3$3N heterojunction diode

The 2019 Al-rich nitride device provides the clearest precursor for BaTiO$_3$4-enabled field-plate thinking in ultra-wide-bandgap nitrides. The active semiconductor is Al$_3$5Ga$_3$6N grown by low-pressure MOCVD on a $_3$7 AlN-(0001)/sapphire template with RMS roughness $_3$8 by AFM. The epitaxial stack comprises a $_3$9 undoped $_3$0-Al$_3$1Ga$_3$2N buffer and a $_3$3 Si-doped $_3$4-Al$_3$5Ga$_3$6N layer with doping concentration $_3$7. Ohmic contacts were realized through selective-area MBE regrowth with a $_3$8 SiO$_3$9 hard mask, a regrown $_3$00 heavily Si-doped Al$_3$01Ga$_3$02N layer, and a $_3$03 heavily Si-doped reverse Al-composition graded AlGaN cap doped to $_3$04. The ohmic metal stack was Ti/Al/Ni/Au $_3$05; isolation used ICP-RIE; BaTiO$_3$06 was deposited by RF sputtering at $_3$07 in oxygen ambient; and the anode metal was Pt/Au $_3$08 (Razzak et al., 2019).

The critical structural distinction is that in the heterojunction device BaTiO$_3$09 remains under the anode and access region, whereas in the control Schottky devices it is etched away from the anode and access regions. This geometry makes the dielectric an active electrostatic termination element rather than a passive insulator. Hall data gave a sheet resistance of $_3$10, mobility of $_3$11, and sheet carrier density of $_3$12; C–V integration on a $_3$13 diode yielded $_3$14, with extracted top-region doping matching the intended $_3$15 (Razzak et al., 2019).

Experimentally, the dielectric-assisted structure traded forward conduction for reverse blocking. For $_3$16 anode-cathode spacing, the Schottky diode turn-on was $_3$17 and the BaTiO$_3$18 heterojunction diode turn-on was $_3$19; the differential on-resistance increased from $_3$20 to $_3$21. Under near-identical sub-$_3$22 spacing, however, the breakdown voltage improved from $_3$23 at $_3$24 in the control to $_3$25 at $_3$26 in the BaTiO$_3$27 device, corresponding to about a $_3$28 increase. The reported heterojunction average breakdown field exceeded $_3$29, reaching $_3$30, while control devices with similar dimensions were around $_3$31 more generally and $_3$32 in the explicitly compared example. The authors further state that $_3$33 was, at the time, the highest experimental breakdown field reported for any semiconductor material/device and compared it with a predicted critical field of about $_3$34 for Al$_3$35Ga$_3$36N (Razzak et al., 2019).

For BaTiO$_3$37/AlN field-plate research, the importance of this paper is conceptual rather than literal. It demonstrates that an extreme-$_3$38 dielectric under a metal edge can make the sustainable average field approach the intrinsic critical field more closely in an ultra-wide-bandgap nitride. The same paper explicitly identifies further optimization through BaTiO$_3$39 thickness optimization, growth-condition optimization, and potential integration of field plate structures. It also makes clear that the transfer to AlN is not experimentally validated: the active layer is Al$_3$40Ga$_3$41N, not AlN, and the BaTiO$_3$42/AlN interface itself is not studied.

4. Vertical GaN architectures using BaTiO$_3$43 with a field plate

The 2023 vertical GaN-on-GaN PN power diode shows how BaTiO$_3$44 enters a more conventional field-plate environment. The device is built on a bulk NEAT GaN substrate and combines five termination elements: nitrogen-implant isolation or guard-ring formation, multiple guard rings, SoG dielectric shaping, a metal field plate, and a local high-$_3$45 dielectric stack of Al$_3$46O$_3$47/BaTiO$_3$48 (Xu et al., 2023). The fabricated epitaxial structure comprises about $_3$49 $_3$50-GaN with targeted Si concentration $_3$51, $_3$52 $_3$53-GaN with net donor concentration $_3$54, a $_3$55 drift layer with average net donor concentration $_3$56, $_3$57 $_3$58-GaN with expected hole concentration $_3$59, and a $_3$60 $_3$61-GaN cap for ohmic contact. The active region is circular with $_3$62 diameter.

The termination was optimized sequentially in TCAD. The final simulated guard-ring design used six guard rings, each $_3$63 wide and separated by $_3$64. A first $_3$65 SoG layer was added around and stepping onto the anode area, followed by a Ti/Al field plate of thickness $_3$66 that covered the complete anode area and extended $_3$67 toward the isolation area in the design discussion. Near the anode and field-plate edge, a $_3$68 Al$_3$69O$_3$70 layer deposited by ALD and a $_3$71 BaTiO$_3$72 layer deposited by sputtering at $_3$73 were introduced as a local high-$_3$74 stack; BTO around the anode area was then etched off using ICP/RIE with BCl$_3$75. A final $_3$76 SoG top passivation completed the structure (Xu et al., 2023).

The paper’s key field-management claim is that the hotspot migrates as termination sophistication increases. With guard rings only, the peak field moves from the anode edge to the edge of the last guard ring. With guard rings, SoG, and field plate, the fields at the guard-ring edges become more uniform but the peak shifts to the field-plate edge. The purpose of the Al$_3$77O$_3$78/BaTiO$_3$79 region is then to mitigate that residual concentration. In simulation, breakdown increased from $_3$80 without guard rings to $_3$81 with optimized guard rings, and then to $_3$82 after adding the high-$_3$83 dielectric stack. The paper explicitly attributes a $_3$84 increase of breakdown voltage to the dielectric layers based on the simulation. Experimentally, the fabricated diode reached $_3$85 breakdown voltage at current density $_3$86, with $_3$87, BFOM $_3$88, leakage current density $_3$89 at $_3$90, and turn-on voltage $_3$91 at $_3$92 (Xu et al., 2023).

This architecture is directly relevant to BaTiO$_3$93-enabled field management, but it remains distinct from a BaTiO$_3$94/AlN field-plate. AlN is not present anywhere in the reported device. The interfacial dielectric is Al$_3$95O$_3$96, not AlN, and the paper does not provide BTO permittivity, dielectric-breakdown values, interface-trap data, leakage through the stack, or time-dependent reliability. The significance of the work is therefore architectural: it demonstrates that a local thin interfacial dielectric plus BTO high-$_3$97 region can be co-optimized with guard rings, SoG, and a metal field plate in a multi-kV vertical GaN diode.

5. Explicit BaTiO$_3$98/AlN field-plate in vertical $_3$99-Ga$_3$00O$_3$01 Schottky diodes

The 2025 $_3$02-Ga$_3$03O$_3$04 study makes the BaTiO$_3$05/AlN field-plate explicit and defines its logic as electro-thermal co-design. The baseline device is a $_3$06-diameter vertical Pt/$_3$07-Ga$_3$08O$_3$09 Schottky diode with a $_3$10 thick (001) $_3$11-Ga$_3$12O$_3$13 drift layer doped to $_3$14, a highly doped substrate at $_3$15, a Pt Schottky contact corresponding to $_3$16 barrier height, and Ti/Au back ohmic contact. The baseline planar field plate uses $_3$17 BaTiO$_3$18 with field-plate length $_3$19. The stack version replaces that single dielectric with $_3$20 BaTiO$_3$21 on top of $_3$22 AlN at the semiconductor interface, and $_3$23 is varied from $_3$24 to $_3$25. A deeper termination variant further adds mesa etching and a sidewall field plate with $_3$26, top dielectric width $_3$27, and etch depth $_3$28 varied from $_3$29 to $_3$30 (Audri et al., 15 Aug 2025).

The material division of labor is stated directly. BaTiO$_3$31 is chosen for very high permittivity, reported as $_3$32, and therefore for electric-field smoothing. Its limitations are equally explicit: thermal conductivity $_3$33 and a conduction-band offset to $_3$34-Ga$_3$35O$_3$36 of only about $_3$37. AlN is inserted at the interface because the simulations take its thermal conductivity as $_3$38, its predicted critical breakdown field as $_3$39, and its conduction-band offset to $_3$40-Ga$_3$41O$_3$42 as $_3$43 to $_3$44. The paper also supplements the simulations with vertical metal/AlN/$_3$45-Ga$_3$46O$_3$47 MIS diodes, from which an AlN breakdown field of about $_3$48 is extracted, compared with about $_3$49 for Al$_3$50O$_3$51 (Audri et al., 15 Aug 2025).

Electrostatically, the stack does not simply attenuate the edge field; it displaces it into AlN, where the dielectric can tolerate the stress better than $_3$52-Ga$_3$53O$_3$54. Thermally, the improvement is even more pronounced. Using a 2-D self-consistent Silvaco ATLAS electro-thermal simulation framework with a Joule-heat model, the paper shows that the planar BaTiO$_3$55-only field plate produces a hotspot at the Schottky contact edge near the dielectric. Replacing the single dielectric with the BaTiO$_3$56/AlN stack reduces the peak Joule heat power by about an order of magnitude, reported as about $_3$57 at $_3$58 forward bias. When $_3$59 is swept from $_3$60 to $_3$61, the stack yields approximately $_3$62–$_3$63 lower Joule power density than the BaTiO$_3$64-only configuration. The deep-etch sidewall implementation reduces peak Joule heat near the anode edge by more than an order of magnitude relative to the planar BaTiO$_3$65/AlN field-plate and leaves even the trench-corner peak $_3$66–$_3$67 below the planar stack case. Under $_3$68 reverse bias, the extracted anode-edge field falls from the planar stack value of about $_3$69 to $_3$70, $_3$71, $_3$72, and $_3$73 for etch depths of $_3$74, $_3$75, $_3$76, and $_3$77, respectively (Audri et al., 15 Aug 2025).

The thermal interpretation is reinforced by a Landauer thermal-boundary-conductance analysis. The calculated room-temperature TBC is $_3$78 for AlN/$_3$79-Ga$_3$80O$_3$81 and $_3$82 for BaTiO$_3$83/$_3$84-Ga$_3$85O$_3$86, with the same trend persisting to higher temperatures. The paper attributes the difference to more favorable phonon transmission from $_3$87-Ga$_3$88O$_3$89 into AlN than into BaTiO$_3$90, including a frequency range around $_3$91–$_3$92 where BaTiO$_3$93 lacks corresponding phonon modes. In this formulation, the BaTiO$_3$94/AlN field-plate is not merely a high-$_3$95 stack; it is a deliberately split dielectric in which BaTiO$_3$96 handles equipotential shaping and AlN handles interfacial heat extraction and dielectric stress.

6. Tradeoffs, limitations, and implications for future device design

Across the three device families, the central tradeoff is consistent. A higher-permittivity dielectric improves edge-field redistribution and raises the usable average field or breakdown voltage, but it can impose penalties elsewhere. The Al-rich nitride study explicitly notes that a high-permittivity gate/anode dielectric increases effective gate-drain or anode-cathode capacitance and lengthens the depletion region, which may reduce peak cutoff frequency relative to a low-$_3$97 barrier or passivation; it therefore identifies dielectric thickness and lateral extent as optimization variables rather than fixed choices (Razzak et al., 2019). The vertical GaN study likewise demonstrates that BTO is most effective when inserted only where the field plate produces the dominant hotspot, not as a uniform dielectric everywhere (Xu et al., 2023). The $_3$98-Ga$_3$99O$_{0.58}$00 study extends that optimization criterion by showing that thermal conductivity, thermal boundary conductance, and dielectric breakdown field must be considered alongside permittivity; in that sense, the stack is an electro-thermal edge-termination solution rather than a purely electrostatic one (Audri et al., 15 Aug 2025).

The literature also has clear evidentiary limits. The Al$_{0.58}$01Ga$_{0.58}$02N paper does not report the BaTiO$_{0.58}$03 thickness in the provided text and does not experimentally validate a BaTiO$_{0.58}$04/AlN interface. The vertical GaN paper does not mention AlN at all and does not report BTO crystallographic phase, measured dielectric constant, interface-trap density, polarization or ferroelectric hysteresis, time-dependent reliability, or thermal cycling stability (Razzak et al., 2019). The $_{0.58}$05-Ga$_{0.58}$06O$_{0.58}$07 paper provides the most explicit BaTiO$_{0.58}$08/AlN formulation, but its main results are simulation-driven, supplemented experimentally only by vertical MIS measurements of AlN dielectric robustness rather than a full fabricated BaTiO$_{0.58}$09/AlN field-plated power diode (Audri et al., 15 Aug 2025).

For AlN-based active devices, the present record supports only a qualified conclusion. The nitride work strongly suggests that an extreme-$_{0.58}$10 dielectric under a metal edge can move performance closer to the intrinsic field limits of ultra-wide-bandgap nitrides, and the $_{0.58}$11-Ga$_{0.58}$12O$_{0.58}$13 work shows why AlN is an attractive interfacial layer when heat removal and dielectric robustness matter simultaneously. This suggests that a BaTiO$_{0.58}$14/AlN field-plate on AlN or Al-rich nitrides is conceptually well motivated. However, direct transfer is not automatic. The Al-rich nitride paper explicitly notes that dielectric/semiconductor band alignment, polarization effects, interface states, fixed charge, material quality, ohmic-contact difficulty, strain, crystallinity, thermal budget, and ferroelectric behavior may all change when moving from Al$_{0.58}$15Ga$_{0.58}$16N to AlN (Razzak et al., 2019). As a result, the current literature establishes the BaTiO$_{0.58}$17/AlN field-plate most clearly as a validated electro-thermal concept in vertical $_{0.58}$18-Ga$_{0.58}$19O$_{0.58}$20 Schottky diodes, a closely related but nonliteral field-plate analogue in Al-rich nitride lateral diodes, and a BaTiO$_{0.58}$21-assisted high-$_{0.58}$22 termination strategy in vertical GaN power diodes.