The authors modeled SiN film etching with hydrofluorocarbon (CHxFy/Ar/O2) plasma considering physical (ion bombardment) and chemical reactions in detail, including the reactivity of radicals (C, F, O, N, and H), the area ratio of Si dangling bonds, the outflux of N and H, the dependence of the H/N ratio on the polymer layer, and generation of by-products (HCN, C2N2, NH, HF, OH, and CH, in addition to CO, CF2, SiF2, and SiF4) as ion assistance process parameters for the first time. The model was consistent with the measured C-F polymer layer thickness, etch rate, and selectivity dependence on process variation for SiN, SiO2, and Si film etching. To analyze the three-dimensional (3D) damage distribution affected by the etched profile, the authors developed an advanced 3D voxel model that can predict the time-evolution of the etched profile and damage distribution. The model includes some new concepts for gas transportation in the pattern using a fluid model and the property of voxels called “smart voxels,” which contain details of the history of the etching situation. Using this 3D model, the authors demonstrated metal–oxide–semiconductor field-effect transistor SiN side-wall etching that consisted of the main-etch step with CF4/Ar/O2 plasma and an over-etch step with CH3F/Ar/O2 plasma under the assumption of a realistic process and pattern size. A large amount of Si damage induced by irradiated hydrogen occurred in the source/drain region, a Si recess depth of 5 nm was generated, and the dislocated Si was distributed in a 10 nm deeper region than the Si recess, which was consistent with experimental data for a capacitively coupled plasma. An especially large amount of Si damage was also found at the bottom edge region of the metal–oxide–semiconductor field-effect transistors. Furthermore, our simulation results for bulk fin-type field-effect transistor side-wall etching showed that the Si fin (source/drain region) was directly damaged by high energy hydrogen and had local variations in the damage distribution, which may lead to a shift in the threshold voltage and the off-state leakage current. Therefore, side-wall etching and ion implantation processes must be carefully designed by considering the Si damage distribution to achieve low damage and high transistor performance for complementary metal–oxide–semiconductor devices.
The influence of the amount of hydrogen (H) in hydrogenated silicon nitride films (SixNy:Hz) on the etching properties and etching mechanism are unclear for hydrofluorocarbon plasma etching. Therefore, the authors have investigated the effect of H in SixNy:Hz films on the surface reactions during CH2F2/Ar/O2 plasma etching by experimental and numerical simulation techniques. The experimental etch yield (EY) and polymer layer thickness (TC−F) values for SixNy:Hz films with different H concentrations of 2.6% (low-SiN), 16.8% (mod-SiN), and 21.9% (high-SiN) show different trends with the CH2F2/(CH2F2 + O2) flow rate ratio. To understand the mechanism of the different etching properties, the authors estimated the chemical reaction probabilities of the H outflux between F, O, N, C, and Si dangling bonds using first principles calculations and the results of Fourier transform infrared spectroscopy. Based on the estimated reaction probabilities, the authors modeled the surface reactions of SixNy:Hz films under the assumption that the H outflux mainly scavenges incident F radicals (the main etchant species). The authors also consider that the reaction between H and N from outfluxes decreases the desorption reactions of C2N2 and HCN, resulting in a larger TC−F value. Comparing the simulation results of the trends in the whole flow rate ratio range and the absolute values of EY and TC−F with experimental data, the surface model can successfully explain the mechanism. Furthermore, the authors demonstrated time-dependent etched profile and damage distribution for fin-type field-effect transistor SixNy:Hz side-wall etching using the three-dimensional voxel-slab model with the above surface reactions to obtain knowledge about the effect of H on the etched profile and damage distribution. The results show that the etched profile and damage distribution on the Si fin structure are very different for low-SiN and high-SiN because of the different EY and TC−F values induced by different H outfluxes. These results indicate that it is important to carefully control both the etching process and amount of H in the SixNy:Hz film to achieve high-performance advanced complementary metal oxide semiconductor devices.
We developed a numerical simulation method for the depth profiles of plasma-induced physical damage to SiO2 and Si layers during fluorocarbon plasma etching. In the proposed method, the surface layer is assumed to consist of two layers: a C–F polymer layer and a reactive layer. Physical and chemical reactions in the reactive layer divided into several thin slabs and in the deposited C–F polymer layer, which depend on etching parameters, such as etching time, gas flow rate, gas pressure, and ion energy (V pp), are considered in detail. We used our simulation method to calculate the SiO2 etch rate, the thickness of the C–F polymer layer (T C–F), and the selectivity of SiO2 to Si during C4F8/O2/Ar plasma etching. We confirmed that the calculated absolute values and their behavior are consistent with experimental data. We also successfully predicted depth profiles of physical damage to the Si and SiO2 layers introducing our re-gridding method. We found that much Si damage is generated in the pre- and early stages of the overetching step of SiO2/Si layer etching despite the high selectivity. These simulation results suggest that the T C–F value and the overetching time must be carefully controlled by process parameters to reduce damage during fluorocarbon plasma etching. The results have also provided us with useful knowledge for controlling the etching process.
The authors quantitatively investigated the effects of open area ratio and pattern structure on fluctuations in critical dimension (ΔCD) and Si recess depth (ΔdR). To model these effects, under the assumption that three factors—mask open area ratio at the wafer level (global), chip level (semi-local), and local level (local)—affect ΔCD and ΔdR, they performed experiments using wafers ranging from 0.60 to 0.91 of the global range (RG) and the semi-local range (RS) treated by the HBr/O2 plasma etching process, where photoresist mask patterns on the poly-Si film with solid angles (ΩL) ranging from 0.2π to 0.9π were located. As a result, the authors found that ΔCD had positive and linear correlation with the RG value, which was consistent with the trend of the integrated intensity of the etched by-product (SiBrx) estimated by optical emission spectroscopy data and with that of taper angles of observed etched profiles. They also clarified that ΔCD was affected by the amount of SiBrx generated within several times of the mean free path area for the semi-local dependence and that the ΩL value within a 2 μm area, not the pattern space, had a good correlation with fluctuations in ΔCD as a control indicator. Using this experimental knowledge, the authors developed a quasi-three-dimensional Si gate etching simulation procedure that demonstrates the ΔCD value and the etched profile characteristics. Furthermore, our simulation procedure found that ΔdR caused by ion bombardment in the Si substrate, as well as ΔCD, deeply depended on the (RG + RS)ΩL factor. Taking account of the relationship between dR and the ion energy reduced by the SiBrx deposition depth on the Si substrate, the authors found that dR was dependent on the factor, which was consistent with experimental data. These results show that for improving device performance, it is crucial to consider the effect of (RG + RS)ΩL on ΔCD and ΔdR in controlling plasma etching parameters, such as ion energy, gas flux, and etching time.
This work describes the modeling of the surface reactions involved in atomic layer etching (ALE) of SiO2 and Si3N4 with a deposition step using C4F8/O2/Ar plasma and an Ar plasma etch step. In the etching step, the surface was assumed to consist of two layers: a C-F polymer layer and a reactive layer. The effects of residual F from the deposition step and F originating from the C-F polymer layer during the etch step and the influences of the O and N outfluxes generated from the reactive layer were considered, in terms of their effects on the etch rates of the SiO2 and Si3N4 films. Using a three-dimensional voxel-slab model that included the surface reactions described above, an analysis was performed based on the differences between the etching properties of continuous wave (CW) etching and ALE in the cases of blanket wafers and self-aligned contact layers from the viewpoints of numerical simulations. As a result of these analyses, it was found that the use of monoenergetic ion energy improves surface layer thickness controllability for both the polymer layer and the reactive layer and that quantitative control of time variations in both the C-F polymer layer thickness and the ion penetration depth is necessary for high selectivity of SiO2 over Si3N4 (SiO2/Si3N4) and for low plasma-induced damage on the Si3N4 film. Furthermore, in the authors’ simulations, a relatively high SiO2 etch rate was obtained for a modified quasi-ALE (43 nm/min) while maintaining high SiO2/Si3N4 selectivity (more than 100) after optimization of the C-F polymer layer thickness, the ion energy, and the ALE cycle time; this represents a solution in terms of the important issue of the very low etch rate of ALE. These simulation results indicate that accurate prediction of the surface reaction, further quantitative control of the plasma parameters, and optimization of the pattern layout design are necessary to realize higher ALE process performance for practical use in mass production.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.