Abstract As semiconductor device structures become more complex and sophisticated, the formation of finer and deeper patterns is required. To achieve a higher yield for mass production as the number of process steps increases and process variables become more diverse, process optimization requires extensive engineering effort to meet the target process requirements, such as uniformity. In this study, we propose an efficient process design framework that can efficiently search for optimal process conditions by combining deep learning (DL) with plasma simulations. To establish the DL model, a dataset was created using a two-dimensional (2D) hybrid plasma equipment model code for an argon inductively coupled plasma system under a given process window. The DL model was implemented and trained using the dataset to learn the functional relationship between the process conditions and their consequential plasma states, which was characterized by 2D field data. The performance of the DL model was confirmed by comparison of the output with the ground truth, validating its high consistency. Moreover, the DL results provide a reasonable interpretation of the fundamental features of plasmas and show a good correlation with the experimental observations in terms of the measured etch rate characteristics. Using the designed DL, an extensive exploration of process variables was conducted to find the optimal processing condition using the multi-objective particle swarm optimization algorithm for the given objective functions of high etch rate and its uniform distribution. The obtained optimal candidates were evaluated and compared to other process conditions experimentally, demonstrating a fairly enhanced etch rate and uniformity at the same time. The proposed computational framework substantially reduced trial-and-error repetitions in tailoring process conditions from a practical perspective. Moreover, it will serve as an effective tool to narrow the processing window, particularly in the early stages of development for advanced equipment and processes.

Abstract With the advent of complex and sophisticated architectures in semiconductor device manufacturing, atomic-resolution accuracy and precision are commonly required for industrial plasma processing. This demands a comprehensive understanding of the plasma–material interactions—particularly for forming fine high-aspect ratio (HAR) feature patterns with sufficiently high yield in wafer-level processes. In particular, because the shape distortion in HAR pattern etching is attributed to the deviation of the energetic ion trajectory, the detailed ion–surface interactions need to be thoroughly investigated. In this study, molecular dynamics (MD) simulations were utilized to obtain a fundamental understanding of the collisional nature of accelerated Ar ions on the fluorinated Si surface that may appear on the sidewall of the HAR etched hole. High-fidelity data for ion–surface interaction features representing the energy and angle distributions (EADs) of sputtered atoms for varying degrees of surface F coverage and ion incident angles were obtained via extensive MD simulations. A deep learning-based reduced-order modeling (DL-ROM) framework was developed for efficiently predicting the characteristics of the ion–surface interactions. In the ROM framework, a conditional variational autoencoder (AE) was implemented to obtain regularized latent representations of the distributional data with the condition of the governing factors of the physical system. The proposed ROM framework accurately reproduced the MD simulation results and significantly outperformed various DL-ROMs, such as AE, sparse AE, contractive AE, denoising AE, and variational AE. From the inferred features of the sputtering yield and EADs of sputtered/scattered species, significant insights can be obtained regarding the ion interactions with the fluorinated surface. As the ion incident angle deviated from the glancing-angle range (incident angle >80°), diffuse reflection behavior was observed, which can substantially affect the ion transport in the HAR patterns. Moreover, it was hypothesized that a shift in sputtering characteristics occurs as the surface F coverage varies, based on the inferred EADs. This conjecture was confirmed through detailed MD simulations that demonstrated the fundamental relationship between surface atomic conformations and their sputtering behavior. Combined with additional atomistic-scale investigations, this framework can provide an efficient way to reveal various fundamental plasma–material interactions which are highly demanded for the future development of semiconductor device manufacturing.

Abstract As semiconductor device structures become more complex and sophisticated, the formation of finer and deeper patterns is required. To achieve a higher yield for mass production as the number of process steps increases and process variables become more diverse, process optimization requires extensive engineering effort to meet the target process requirements, such as uniformity. In this study, we propose an efficient process design framework that can efficiently search for optimal process conditions by combining deep learning (DL) with plasma simulations. To establish the DL model, a dataset was created using a two-dimensional (2D) hybrid plasma equipment model code for an argon inductively coupled plasma system under a given process window. The DL model was implemented and trained using the dataset to learn the functional relationship between the process conditions and their consequential plasma states, which was characterized by 2D field data. The performance of the DL model was confirmed by comparison of the output with the ground truth, validating its high consistency. Moreover, the DL results provide a reasonable interpretation of the fundamental features of plasmas and show a good correlation with the experimental observations in terms of the measured etch rate characteristics. Using the designed DL, an extensive exploration of process variables was conducted to find the optimal processing condition using the multi-objective particle swarm optimization algorithm for the given objective functions of high etch rate and its uniform distribution. The obtained optimal candidates were evaluated and compared to other process conditions experimentally, demonstrating a fairly enhanced etch rate and uniformity at the same time. The proposed computational framework substantially reduced trial-and-error repetitions in tailoring process conditions from a practical perspective. Moreover, it will serve as an effective tool to narrow the processing window, particularly in the early stages of development for advanced equipment and processes.

Abstract With the advent of complex and sophisticated architectures in semiconductor device manufacturing, atomic-resolution accuracy and precision are commonly required for industrial plasma processing. This demands a comprehensive understanding of the plasma–material interactions—particularly for forming fine high-aspect ratio (HAR) feature patterns with sufficiently high yield in wafer-level processes. In particular, because the shape distortion in HAR pattern etching is attributed to the deviation of the energetic ion trajectory, the detailed ion–surface interactions need to be thoroughly investigated. In this study, molecular dynamics (MD) simulations were utilized to obtain a fundamental understanding of the collisional nature of accelerated Ar ions on the fluorinated Si surface that may appear on the sidewall of the HAR etched hole. High-fidelity data for ion–surface interaction features representing the energy and angle distributions (EADs) of sputtered atoms for varying degrees of surface F coverage and ion incident angles were obtained via extensive MD simulations. A deep learning-based reduced-order modeling (DL-ROM) framework was developed for efficiently predicting the characteristics of the ion–surface interactions. In the ROM framework, a conditional variational autoencoder (AE) was implemented to obtain regularized latent representations of the distributional data with the condition of the governing factors of the physical system. The proposed ROM framework accurately reproduced the MD simulation results and significantly outperformed various DL-ROMs, such as AE, sparse AE, contractive AE, denoising AE, and variational AE. From the inferred features of the sputtering yield and EADs of sputtered/scattered species, significant insights can be obtained regarding the ion interactions with the fluorinated surface. As the ion incident angle deviated from the glancing-angle range (incident angle >80°), diffuse reflection behavior was observed, which can substantially affect the ion transport in the HAR patterns. Moreover, it was hypothesized that a shift in sputtering characteristics occurs as the surface F coverage varies, based on the inferred EADs. This conjecture was confirmed through detailed MD simulations that demonstrated the fundamental relationship between surface atomic conformations and their sputtering behavior. Combined with additional atomistic-scale investigations, this framework can provide an efficient way to reveal various fundamental plasma–material interactions which are highly demanded for the future development of semiconductor device manufacturing.

Hydrogen plasma treatment (HPT) is commonly employed to enhance the interface quality of high-k metal-oxide gate stacks in semiconductor, but the plasma process is highly sensitive to variations in process parameters. The modulation of processing conditions to meet the increasing demand for atomic-precision in device feature is severely challenging without a proper understanding of the plasma-facing surface. This study aimed to provide atomistic insights on HPT to improve the quality of high-k metal-oxide/silicon interfaces through a computational approach via molecular dynamics and density functional theory. Hydrogen collision, adsorption, penetration, diffusion and desorption were simulated on the processed surface. Depth-wise distribution of hydrogen, structural evolution of target surface and the resultant charge traps at the interface were captured in the computational model for varying processing conditions, incident atomic energies and substrate temperatures. Notably, a physical correlation between the processing parameters and the interface quality was found; the dangling bond density is directly- (at low E) or inversely proportional (at high E) to the substrate temperature, causing variations in trap defects and stability of the gate stacks. A concrete understanding of the HPT offers a guideline to optimize the process window and development of advanced equipment for practical use.

Abstract Atomic layer deposition (ALD) is extensively used to fabricate doped dielectrics due to its ability to deposit conformal films with atomic-scale thickness control. Al-doped TiO 2 (ATO) is a promising high- k dielectric for dynamic random access memory (DRAM) applications, offering a high dielectric constant with a remarkable leakage-lowering effect by Al acceptor doping. However, ATO fabrication via conventional supercycle-based ALD suffers from severe crystallinity loss during the growth of TiO 2 upon Al doping owing to the dopant-induced lattice disorder. In addition, Al doping cannot reduce any inherent O vacancies (V O ) of TiO 2 , although the original purpose of doping was to address the n-type nature caused by V O . To resolve these limitations, we propose a single-step, in-situ Ar/O 2 post-doping plasma (PDP) process immediately after the Al dopant incorporation. Using the PDP process, simultaneous atomic-scale dopant migration-mediated crystallization and V O annihilation were successfully initiated. Thus, the surface concentration of the dopant decreased, reducing the dopant-induced lattice distortion, while promoting the highly crystallized seed layer-like surface. Consequently, strong rutile-phase recovery was accompanied by enhanced lattice-matched growth. In addition, the PDP process significantly lowers the V O -to-lattice oxygen ratio by facilitating the recombination between reactive O species and V O , increasing the corresponding 0.4 eV of conduction band offset (CBO). Despite the common trade-off between the dielectric constant and leakage, the Pt/PDP-ATO/Ru capacitor exhibited a simultaneous 30% increase in dielectric constant and up to a 1.6-order reduction in leakage current density.

• An efficient design framework for foam-like structures is newly presented. • The stochastic nature of Voronoi structures is considered in the proposed framework. • The optimized Voronoi structures show 95.25% improvement in efficiency compared to the cases where the proposed framework is not implemented. • The proposed methodology is a powerful tool for efficient design of Voronoi structures for a specific target objective. Recently, many studies have increasingly focused on developing bio-inspired structures, leveraging their lightweight and high-energy absorption properties, which are crucial across many engineering fields. Structural optimization aiming for bio-inspired structures having superior energy absorption capability, however, has been considered a challenging problem. One of these challenges is that nonlinear material behaviors induced by external forces, such as buckling and self-contact of constituting ligaments, intervene in the energy absorption process. Such nonlinearities not only make the relationship between design changes and energy absorption nonlinear, but also exacerbate the difficulties of design, given the complexity of the ligament configurations. To address this, a novel design optimization method for bio-inspired cellular structures with high energy absorption is proposed. First, Voronoi tessellation is used to capture configurations of bio-inspired material, parameterized by geometric variables. Then, Bayesian optimization with Kriging efficiently updates the design, exploring the complex design space through high-fidelity nonlinear finite element analysis. The proposed design method is efficient in structural optimization as it combines a strategy to reduce the number of samples required for surrogate modeling of structural response and optimal search, but it also generates multiple design outcomes with similar advantages due to the intrinsic variance of the Voronoi structures.

In this study, we conducted Density Functional Theory (DFT) simulations to investigate the impact of surface halogenation on the dissociative adsorption of SiH4 on crystalline Si(001) surface. First, we examined the adsorption energetics of surface halogenation using chlorine (Cl2) molecules. The dissociative adsorption of Cl2 exhibited a high reactivity with the undercoordinated surfaces without barrier. Molecular Dynamics (MD) simulations using Neural Network Potentials (NNP) has been developed and adopted to study surface diffusion of adsorbed Cl atoms and the surface concentration of Cl on the Si(001) surface at higher temperature. Second, we conducted subsequent silane (SiH4) adsorption and the first hydrogen dissociation on the bare and Cl-passivated surfaces. Our potential energy surface search revealed that the surface halogenation on the silicon surface suppresses the primary adsorption reaction of SiH4. Overall, this study provides an atomistic description of the dissociative adsorption mechanisms and associated energies of Cl2 and subsequent SiH4 reactions for modeling the epitaxial deposition process and may serve as foundational knowledge for developing next-generation Si epitaxy techniques.