Thermal atomic layer etching (ALE) for SiO2 films with self-limiting behavior on the surface modification step was developed using sequential exposure to HF and NH3 gases followed by infrared (IR) annealing. X-ray photoelectron spectroscopy analysis showed that an (NH4)2SiF6-based surface-modified layer was formed on the SiO2 surface after gas exposures and that this layer was removed using IR annealing. The etch per cycle (EPC) of the ALE process saturated at 0.9 nm/cycle as the gas exposure times increased. With this self-limiting behavior, SiO2 was etched with high selectivity to poly-Si and Si3N4. The dependence of the EPC on the partial pressures of HF and NH3 was found to be in good agreement with the Langmuir adsorption model. This indicated that the HF and NH­3 molecules were in equilibrium between adsorption and desorption during the exposure, which resulted in the self-limiting formation of the modified layer. In addition to the process with an HF gas flow, it was demonstrated that an H2/SF6 plasma can replace the HF gas exposure step to supply the SiO2 surfaces with HF molecules. The EPC saturated at 2.7 nm/cycle, while no measurable thickness change was observed for poly-Si and Si3N4 films.

Atomic layer etching (ALE) is usually classified into ion-driven anisotropic etching or thermally driven isotropic etching. In this work, we present a thermal ALE process for Si3N4 with high selectivity to SiO2 and poly-Si. This ALE process consists of exposure to a CH2F2/O2/Ar downstream plasma to form an (NH4)2SiF6-based surface-modified layer, followed by infrared (IR) annealing to remove the modified layer. CH2F2-based chemistry was adopted to achieve high selectivity to SiO2 and poly-Si. This chemistry was expected to reduce the number density of F atoms (radicals), which contributes to decreasing the etching rate of SiO2 and poly-Si films. X-ray photoelectron spectroscopy analysis confirmed the formation of an (NH4)2SiF6-based modified layer on the surface of the Si3N4 after exposure to the plasma and subsequent removal of the modified layer using IR annealing. An in situ ellipsometry measurement revealed that the etch per cycle of the ALE process saturated with respect to the radical exposure time at 0.9 nm/cycle, demonstrating the self-limiting nature of this etching process. In addition, no etching was observed on SiO2 and poly-Si films, successfully demonstrating the high selectivity of this ALE process. This high selectivity to SiO2 and poly-Si is attributed to the fact that the spontaneous etching rates of these films are negligibly small and that there is no surface reaction to etch these films during the IR annealing step.

Fabrication of 3D devices (Fin-FETs, GAA, 3D-NAND, etc) requires not only atomic level accuracy and higher aspect ratio patterning but also high-selectivity conformal (or lateral) etching. In this paper, advanced metrology by utilizing high energy back-scattered electrons, especially in the case of HAR holes, and single reaction control for atomic-scale nanofabrication are discussed. Selective rapid thermal cyclic ALE of Si 3 N 4 was, then, demonstrated.

We developed a method for generating a mathematical model of equipment. The model can be used in many model-based applications of prognostics and health management. The method processes sensor data obtained from target equipment to generate a model that contains sensors, latent variables, and approximate equations. First, latent variables are generated by analyzing correlation coefficients. Next, the method divides the variables (latent variables and sensors) into several groups by applying a hierarchical clustering method. Finally, it generates approximate equations of variables within each group. The generated equations can work as features to help users detect potential failures or estimate remaining useful life. The results of experiments using data obtained from electric generators shows the effectiveness of the features. We also discuss the differences between generating features by using a neural network and the proposed method.

We developed a method for extracting sub-systems for predictive maintenance. The method processes sensor data obtained from target equipment to extract several sub-systems that contain correlated variables. First, the latent variables are generated by analyzing correlation coefficients. Next, the method divides the variables (latent variables and sensors) into different groups by applying a hierarchical clustering method. Each group represents a sub-system that has the variables strongly correlated with each other. The application of sub-systems in predictive maintenance helps users to detect anomalies earlier and more accurately than the conventional methods, leading to better maintenance and productivity. The effectiveness of this method was evaluated by using sensor data obtained from compact electrical generators. We also describe possible ways of visualizing the sub-systems.

Because the structure of semiconductor devices has changed to miniaturized three-dimensional (3D) structures, etching technologies that enable 3D atomic-scale control have received increasing interest. Therefore, atomic level/layer etching (ALEt), which is a counterpart of atomic layer deposition (ALD), has attracted much attention due to its potential in atomic-scale 3D nanofabrication. ALEt typically involves a cyclic repetition of self-limiting formation and desorption of surface modified layers. Thus, geometric loading effects are mitigated when using ALEt because of the self-limiting behavior. The authors have developed a novel isotropic ALEt process for SiN using cyclic repetitions of hydrofluorocarbon-based plasma exposure and thermal annealing [1]. The first step of the etching cycle for SiN is exposing the sample surface at room temperature to hydrofluorocarbon-based radicals that form ammonium hexafluorosilicate ((NH4)2SiF6) on SiN. The second step is thermal annealing above 100°C to sublimate the ammonium hexafluorosilicate. SiN film with an atomically thin layer can be etched away by repeating the cycle of forming and by removing the ammonium hexafluorosilicate. This cyclic etching shows self-limiting behavior because the formation of the ammonium hexafluorosilicate saturates as the plasma exposure time increases. This technology is applicable to other nitride films besides SiN. The authors have demonstrated selective isotropic ALEt of TiN [2], which is widely used as a gate electrode and barrier metal in semiconductor devices. In this paper, we present our results with the isotropic ALEt process for SiN and TiN. This method features a variety of promising characteristics, such as high selectivity with respect to SiO2and poly Si, atomic-level control of the etching amount, and isotropic etched profiles. The authors developed a novel etching apparatus as an ALEt tool for 300-mm wafers. This apparatus utilizes an infrared (IR) lamp for thermal annealing to achieve rapid temperature cycles. Particular attention is devoted to the surface reaction mechanism in the novel isotropic ALEt process for SiN. The results of in situ X-ray photoelectron spectroscopy (XPS) and thermal desorption spectroscopy (TDS) are described; they reveal the formation and decomposition of ammonium hexafluorosilicate on SiN during the cycle of plasma exposure and annealing. The potential industrial applications of these two processes are also described. [1] K. Shinoda et al., Appl. Phys. Express 9, 106201 (2016). [2] K. Shinoda et al., AVS Symposium and Exhibition, PS+TF-WeM10 (2016).

Variations in sensor data collected from equipment have been widely analyzed by using anomaly detection methods for predictive maintenance. Our experience shows that correlations between sensors effectively predict failures because the correlations usually reflect the status of equipment with higher sensitivity. In this paper, we present a method that exploits correlations between sensors for pre-processing and enables anomalies to be detected using both sensor data and correlations. The method was evaluated by applying it to compact electric generators, and the results showed it detected anomalies more accurately than when only sensor data were used. This method is expected to predict failures earlier and reduce the cost of downtime and maintenance.

An atomic layer etching process for silicon nitride (Si3N4) has been developed in which ammonium fluorosilicate [(NH4)2SiF6] is formed and desorbed using infrared annealing. The cycle of forming and removing ammonium fluorosilicate was repeated, demonstrating that the Si3N4 etching depth was accurately controlled with high selectivity to SiO2 by changing the number of cycle. An X-ray photoelectron spectroscopy peak, which had been previously assigned as N–H bond of an ammonium salt, was observed after radical exposure, indicating that the ammonium fluorosilicate-based modified layer had formed. This peak disappeared after infrared annealing for 10 s, demonstrating desorption of the modified layer. In thermal desorption spectroscopy, NH3, HF, and SiFx were detected, providing further evidence for the formation of the ammonium fluorosilicate-based modified layer. In addition, this layer has a multilayer structure, protecting the Si3N4 from exposure to reactive radicals.

The demand for precisely controlled etching is increasing as semiconductor device geometries continue to shrink. To fulfill this demand, cyclic atomic level/layer etching will become one of the key technologies in semiconductor device manufacturing at nanometer dimensions. This review describes recent trends in semiconductor devices and some of the latest results on cyclic atomic-level etching. In particular, it focuses on two types of cyclic etching that use different heating procedures: infrared irradiation for isotropic etching and Ar+ ion bombardment for anisotropic etching. It describes how an inductively-coupled-plasma down-flow etching apparatus with infrared lamps can be used for isotropic cyclic etching. The isotropic cyclic etching of SiN involves the formation and thermal desorption of ammonium hexafluorosilicate-based surface modified layers. This method features high selectivity with respect to SiO2, atomic-level control of the amount of SiN etching, and isotropic etched features. On the other hand, the anisotropic cyclic etching with Ar+ ion bombardment uses a microwave electron-cyclotron-resonance plasma etching apparatus. The anisotropic process for poly Si is composed of cyclic repetitions of chlorine adsorption and Ar+ ion bombardment. The anisotropic process for SiN is composed of cyclic repetitions involving an adsorption step using hydrofluorocarbon chemistry and a desorption step using Ar+ ion bombardment. Potential applications of these isotropic/anisotropic cyclic etching processes are described.

We investigated a silicon mold fabrication that uses a hard mask stack by using poly(methyl methacrylate)-block-poly(methacrylate polyhedral oligomeric silsesquioxane) (PMMA-b-PMAPOSS) as the block copolymer (BCP) to assemble nano-patterns for a nano-imprint lithography process during bit-patterned media manufacturing. We developed a dry development process comprised of a single step by taking both the selectivity and anisotropy into consideration, which enables us to create hole patterns by using an array of PMMA spheres embedded in a PMAPOSS matrix. The availability of this process was evaluated from the experimental results that showed that hole patterns at several areal densities were successfully obtained by adjusting the process time under a fixed etching condition. The capability of the pattern transfer to a hard mask from the hole patterns of residual PMAPOSS could be improved by changing the hard mask material from SiO2 to amorphous carbon based on the results from an X-ray photoelectron spectroscopy (XPS) surface analysis. Silicon molds with areal densities of up to 2.8 Tbit/in.2 were successfully fabricated by using an optimized process condition and the hard mask stack.

The characteristics of poly(methyl methacrylate) (PMMA) etching of self-assembled poly(styrene-block-methyl methacrylate) (PS-b-PMMA) thin film for forming a polystyrene (PS) mask were investigated. In this investigation, first, the etching selectivity of PMMA to PS under argon-and oxygen-plasma processes was evaluated. Higher selectivity was obtained in the case of argon plasma (3.9) compared to that of oxygen plasma (1.7). Second, to investigate the argon process in detail, the time dependence of etching depth was evaluated. It was found that PMMA etching rate decreases by more than half after etching to a depth of around 15 nm. To investigate the mechanism of this decrease in PMMA etching rate, the surface composition of PMMA was measured by X-ray photoelectron spectroscopy (XPS). The XPS result revealed that the reduction of etching rate is caused by a depletion of oxygen by argon ions, and the depleted oxygen attaches to the PMMA film in air exposure. In accordance with these results, to compensate the decrease in oxygen concentration, oxygen was added to the argon plasma at a composition of 1%. As a result of this oxygen addition, constant PMMA etching rate was confirmed, even beyond etching depth of 50 nm. It is thus concluded from these results that a PS lamellar mask pattern with a pitch from 25.5 to 70 nm could be successfully formed by using selective PMMA etching.

The effect of oxygen addition to an argon plasma on the etching selectivity of poly(methyl methacrylate) (PMMA) to polystyrene (PS) (hereafter "PMMA/PS etching selectivity") was investigated. The PMMA/PS etching selectivity was evaluated by using inductively coupled plasmas composed of argon and oxygen. The etching selectivity in the case of argon plasma was estimated to be 3.9, which is higher than that of oxygen plasma, which is 1.7. The time dependence of etching depth shows that the etching rate of PMMA is reduced to less than one half of its initial value after the etching depth exceeds 15 nm. X-ray photoelectron spectroscopy of the PMMA surface revealed that the reduction of etching rate is caused by a depletion of oxygen concentration by argon-ion bombardment. To compensate the oxygen-concentration depletion, 1% oxygen was added to the argon plasma. As a result, the reduction of PMMA etching rate was suppressed, and constant etching rate was obtained even when etching depth exceeded 50 nm. The mixed argon-oxygen plasma was used to fabricate a PS mask pattern with a full pitch in the range of 25.5 to 77 nm. (C) The Authors. Published by SPIE under a Creative Commons Attribution 3.0 Unported License. Distribution or reproduction of this work in whole or in part requires full attribution of the original publication, including its DOI.

This study presents a method to reduce hole-diameter variation and defect ratio in patterning of a self-assembled block copolymer (BCP) for imprint-mold fabrication. The BCP material used is PMMA-b-poly(methyl acrylate) polyhedral oligomeric silsesquioxane (PMAPOSS) in which PMMA spheres with 18.3-nm-pitch are aligned in the hexagonal close-packed positions in the PMAPOSS matrix. When the self-assembled BCP film was etched in the conventional dry-development process, the hole-diameter variation and the amount of hole defects (defect ratio), defined as “no-opening defects” or “connecting holes,” increased. Variation of PMMA sphere diameter and/or position in the perpendicular direction to the substrate plane was assumed to be the main cause of the increase in hole-diameter variation and defect ratio after BCP development. To optimize the etching conditions for BCP development, a new model representing the relationship between defect ratio and relative standard deviation of PMMA sphere diameter and/or position under various etching conditions (selectivity and anisotropy) was developed. The model indicates that to reduce defect ratio, higher etching selectivity of PMMA to PMAPOSS and etching anisotropy are required. It also indicates that optimizing etching anisotropy is more effective than optimizing etching selectivity. On the basis of these modeling results, the dry-development conditions were optimized; namely, selectivity (etching ratio of PMMA to PMAPOSS) was increased from 4 to 6 and anisotropy (etching ratio of anisotropy to isotropy) was increased from 1.6 to 2.1. As a result, the defect ratio decreased from 31% to 5%. A 1.7 tera-dot/in.2 imprint-mold was fabricated by applying the directed self-assembly process.

The gate critical dimension (CD) variation of ultra-large-scale integrated circuit (ULSI) devices should be reduced to improve the production yield. An examination of the formulation of a gate-CD model for the transistor area, including the static random access memory (SRAM), was conducted taking the topographical and layout effects into account. It was found that the formulation of a gate CD for transistor areas with a root-mean-square error (RMSE) of less than 1 nm was efficient. The coefficients of the shallow-trench-isolation (STI) step height and polycrystalline-silicon (poly-Si) thickness were found to be inversely proportional to the distance between the gate electrodes. It was found that this dependence is related to the reactive-ion-etching (RIE) lag in the etching process.