28 July 2025 in Semiconductor Science and Technology
As CMOS technology continues to advance and scale down, conventional post-metallization annealing (PMA), conducted to enhance yield and device performance, needs to evolve due to its excessively high thermal stress and long duration. In this context, this study proposes rapid deuterium annealing (RDA) as an alternative to conventional PMA for minimizing the thermal budget. Compared to conventional PMA, the RDA process shortened the annealing time from 30 min to 3 min, thereby effectively reducing the thermal budget. High-k metal gate MOSFETs are fabricated for process verification, and the impact of RDA is investigated through device characterization. After the RDA process, device performance and stress immunity have been enhanced through passivation of interface traps.
4 March 2026 in ACS Applied Electronic Materials
20 November 2025 in Semiconductor Science and Technology
Deuterium annealing has been investigated under various temperature, pressure, and time conditions to enhance the performance and reliability of semiconductor devices. However, the use of inconsistent test devices across studies has made it difficult to determine the optimal process conditions. In this work, two recently reported annealing conditions (Process A: 250 °C for 3 min and Process B: 300 °C for 1 h) are directly compared using identical silicon-based n-type MOSFETs. Process B, with a longer annealing time, achieved a 72.15% reduction in interface trap density ( D IT ), while Process A resulted in a reduction of 61.01%. However, in terms of efficiency with respect to annealing time, a short duration exhibits superior performance, since rapid improvement occurs at the initial stage and gradually approaches saturation thereafter. These results demonstrate that longer annealing enhances trap passivation but also involves trade-offs associated with extended duration. This comparative study provides useful insights for optimizing deuterium annealing in mass-produced silicon devices.
26 October 2026 in Applied Physics Letters
Retention degradation in three-dimensional (3D) dynamic random-access memory (DRAM) becomes severe as access transistors adopt nanosheet structures. Using technology computer aided design simulations, the retention mechanisms of a double-gate nanosheet 1T1C DRAM are investigated under all four dynamic/static “1” and “0” conditions, revealing that the band-to-band tunneling-induced floating body effect (FBE) and quasi-static (QS) leakage dominate retention degradation in the 1 and 0 states, respectively. To exploit the independent controllability of the back-gate (BG), a state-dependent BG pulse-controlled scheme is proposed, in which a positive pulse is applied to suppress the FBE in the 1 states, while a negative pulse is used to suppress QS leakage in the 0 states, thereby enhancing retention under all conditions.
19 November 2025 in IEEE Access
Recently, silicon nanosheet (NS) field-effect transistors (FETs) with a gate-all-around (GAA) structure have been introduced into semiconductor industry for 3-nm-node logic technology. Compared to the planar or tri-gate structures, the GAA structure provides strong gate controllability, which suppresses short-channel effects (SCEs) that increase the off-state current (IOFF). However, considering future development trends in which the devices are scaled down to the sub-2-nm node, SCEs become inevitable even in NS FETs. In this context, this paper investigates a novel inner spacer-less (ISL) NS FET for reducing IOFF. While maintaining the device footprint, the ISL NS FET extends the physical gate length (LG) by removing the inner spacer from the conventional structure. The ISL NS FET exhibits enhanced gate capacitance, which suppresses drain-induced barrier lowering (DIBL) and reduces IOFF in the NS regions. This reduction in IOFF is expected to improve the static power consumption of the chip. All simulations of the fabrication process and electrical characteristics are conducted through three-dimensional (3-D) technology computer-aided design (TCAD).
8 December 2025 in IEEE Transactions on Electron Devices
As artificial intelligence (AI) models have evolved aggressively with the advancement of vector processing hardware and the invention of the transformer architecture, dynamic random access memory (DRAM) technologies have entered a new phase to meet the growing demand for higher memory bandwidth in mass data processing. While various strategies have been proposed to enhance system-level bandwidth, such as high-bandwidth memory (HBM) and the double-data-rate (DDR) protocol, several challenges remain, including thermal issues and noise instabilities. Hence, improvements from a device-level perspective are essential. In this article, a DRAM cell transistor with a wrap-around word-line (WAWL) structure is introduced. The proposed structure enlarges the effective channel width as well as the bitline (BL) and storage node (SN) contact areas ( ${A}_{\text {BL}}$ and ${A}_{\text {SN}}\text {)}$ . As a result, the proposed DRAM cell transistor with the WAWL structure exhibits significant enhancements in on-state current ( ${I}_{\text {ON}}\text {)}$ /off-state current ( ${I}_{\text {OFF}}\text {)}$ ratio and subthreshold swing (SS). Consequently, both the write time ( ${t}_{\text {WRITE}}\text {)}$ and read time ( ${t}_{\text {READ}}\text {)}$ are significantly reduced.