With the rapid advancement of a artificial intelligence computing, the development of next-generation memory devices characterizes by high integration and low power consumption is required. Among various candidates, resistive random-access memory (ReR...
With the rapid advancement of a artificial intelligence computing, the development of next-generation memory devices characterizes by high integration and low power consumption is required. Among various candidates, resistive random-access memory (ReRAM) is considered advantageous for implementation in crossbar array (CBA) structures due to its high-speed operation, non-volatile nature, and simple devie architecture. However, CBAs suffer from sneak current, where unintended current paths cause read/write errors. To address this, external selection devices such as transistors or diodes, selector have been introduces to restrict current paths, but these are accompanied by limitations such as increased device area, process complexity, and cost.
In 2014, a device exhibiting self-rectifying characteristics, in which leakage current is suppressed by a Schottky barrier formed between the top electrode and the switching layer, was proposed by Yoon et al[1]. The switching mechanism, driven by trapping and de-trapping of electrons through traps formed at the hafnium oxide (HfO2) interface during the sputtering deposition of a Tantalum pentoxide (Ta2O5) layer on a HfO2 thin film, was identified.
Despite advantages such as leakage current prevention, high integration, and low power consumption, such charge-trap-based self-rectifying ReRAM are limited by poor device-to-device and cycle-to-cycle uniformity depending on the formation of trap sites. Therefore, improving switching uniformity and operational reliability is essential.
Recently, in 2024, the improvement of LRS variation in charge-trap-based self-rectifying devices by inserting Au nanodots into a Pt/Ta2O5/HfO2/TiN structure to restrict current paths through electric filed concentrations was reported by Park et al[2]. However, this approach is limited by challenges in controlling the distribution of nanodots and the resulting process steps. Additionally the improvement of variation and reliability in self-rectifying devices using aluminum (Al)-doped HfO2 as a switching layer was reported by Kim et al[3]. By doping Al into the HfO2 layer, the depth of deep trap sites is modulated, allowing electrons to be trapped stably. Nevertheless, constraints remain regarding practical application due to the addition of a Ta2O5 layer and the use of platinum (Pt) as a top electrode.
In this study, the improvement of operational performance and stability is aimed for by introducing an Al-doped HfO2 switching layer and optimizing the electrode-switching layer interface. To identify electrode materials with enhanced complementary metal-oxide-semiconductor (CMOS) compatibility that can replace Pt, rectification characteristics according to the work function of the top electrode materials were evaluated. Furthermore, the electrical characteristics of the device were assessed based on the doping concentration and the state of the electrode-oxide interface. Through the analysis of the electrical and material properties of the fabricated devices, methods for performance enhancement are explored, and the rectification and switching mechanisms are investigated. Based on these results, an optimal device structure with high reliability suitable for high-density integration is proposed.