The Inlaid Al₂O₃ Tunnel Switch for Ultrathin Ferroelectric Films

Abstract

Ferroelectricity is an intriguing property of some materials that belong to certain non-centrosymmetric crystal groups. Ferroelectric (FE) materials show a residual surface ionic charge, called remanent polarization (P r ), even under the absence of an external electric field. The sign of Pr can be reversed by reversing the external electric field. This is a very attractive characteristic for nonvolatile semiconductor memory devices, where the bistable charge states correspond to digital 0 and 1 data. [1,2] In addition, they are also viable functional materials for neuromorphic synapse circuits, where more analogue-type data processing is necessary.[3] Recently, their high electromechanical coefficients have provided micro-and nano-electromechanical systems with a new opportunity.[4] The great functional opportunities of FE materials have led to renewed interest from industry and theoreticians in determining the limitations in the vertical and lateral sizes of FE materials. Although first-principle calculations [5] in combination with experimental observations [6] have shown that it is possible to scale down the thickness to a few unit cells while maintaining the functionality of FE materials, the real situation has not been so promising. In most semiconductordevice applications, FE materials should be integrated with the standard complementary metal oxide semiconductor field effect transistor so that they are formed during the back-end of the line process. This means that FE materials should have metal (or conducting oxide) electrodes and a polycrystalline thin-film structure, which is a very different situation from what the theoretical calculations have assumed. Due to this harsh environment, FE materials suffer from interfacial effects, including either the intrinsic finite electrostatic screening length of the metal electrodes, [7][8][9][10] and a finite intrinsic dead layer, or the processing issues of an intervening defect, phase impurities, and residual stresses. [11][12][13][14] Even when all extrinsic effects are minimized by technological improvements, intrinsic by-electrode effects generate a considerably large depolarizing field to destabilize the single-domain pattern within a single memory cell into 180 8/90 8 stripe domains, which reduces the equilibrium system energy. [6,15,16] These effects become increasingly severe as the FE layer-thickness decreases, rendering ultrathin FE film applications virtually impractical. Furthermore, the length of accessing the write/read pulses of the memory devices should be longer than the actual domain switching time for operation safety. This arouses various reliability issues, such as imprint and fatigue, due to by-electrode charge injection during field overstressing. [17,18] It has been confirmed experimentally that there are normally non-ferroelectric interfacial layers at the interfaces between (metal) electrodes and FE layer that have detrimental effects on the FE functionality of thin films. On the other hand, it is possible that this interfacial layer ca...

Sub‐Picosecond Processes of Ferroelectric Domain Switching from Field and Temperature Experiments

“…[33] After applying these corrections we calculated the dependence of lnJ sw on E c −1 for various values of T, as shown in Figure 5, where J sw = I sw /S. …”

Tristate Memory Using Ferroelectric-Insulator-Semiconductor Heterojunctions for 50% Increased Data Storage

“…The P -V loops generally show a rather distorted shape due to the involvement of a semiconducting ZnO layer, and a large coercive voltage due to the adoption of a Al 2 O 3 tunnel switch layer. [ 11 ] A further notable fi nding is that the P -V loops in the positive V region show a hump, whereas those in the negative region do not. This result can be seen even more clearly from the I sw versus voltage curves.…”

“…The negative coercive voltage and integrated charges are similar to those from the PZT/ Al 2 O 3 capacitor. [ 11 ] For a suffi ciently low negative bias, compensating carriers (electrons) are injected into the Al 2 O 3 /ZnO and fi nally into the PZT/Al 2 O 3 interfaces by tunneling through the thin Al 2 O 3 layer and compensate for the polarizationbound charges. [ 11 ] Under a positive bias, the ZnO layer could be carrier-depleted and should work as an insulating layer.…”

“…[ 11 ] For a suffi ciently low negative bias, compensating carriers (electrons) are injected into the Al 2 O 3 /ZnO and fi nally into the PZT/Al 2 O 3 interfaces by tunneling through the thin Al 2 O 3 layer and compensate for the polarizationbound charges. [ 11 ] Under a positive bias, the ZnO layer could be carrier-depleted and should work as an insulating layer. The correct understanding of the charge transport behavior under a positive bias requires a quantitative simulation of the electricfi eld distribution, which is shown in the related discussions.…”

“…Recently, Jiang and co-workers reported that inserting a thin Al 2 O 3 layer between Pb(Zr,Ti)O 3 (PZT) and a Pt electrode resulted in unexpected improvement of the ferroelectric properties of the metal-ferroelectric-insulator-metal (MFIM) structure. [11][12][13] They showed that the thin Al 2 O 3 layer works as a tunnel switch, which means that it is open during polarization switching and closed during nonswitching or after switching is complete. This closed tunnel switch improves the retention of compensating charge for the ferroelectric-bound charge.…”