Atomic-scale cellular model and profile simulation of poly-Si gate etching in high-density chlorine-based plasmas: Effects of passivation layer formation on evolution of feature profiles

Abstract

Atomic-scale cellular model has been developed to simulate the feature profile evolution during poly-Si gate etching in high-density Cl2 and Cl2∕O2 plasmas, with emphasis being placed on the formation of passivation layers on feature surfaces. The model took into account the behavior of Cl+ ions, Cl and O neutrals, and etch products and byproducts of SiClx and SiClxOy in microstructural features. The transport of ions and neutrals in microstructures and in substrates was analyzed by the two-dimensional Monte Carlo calculation with three velocity components. The surface chemistry included ion-enhanced etching, chemical etching, and passivation layer formation through surface oxidation and deposition of etch products and byproducts. The computational domain was taken to consist of two-dimensional square cells or lattices of atomic size, and the evolving interfaces were represented by removing Si atoms from and/or allocating them at the cells concerned. Calculations were performed for different line-and-space pattern features of down to 30nm space width, with an incoming ion energy, ion flux, and neutral reactant-to-ion flux ratio of Ei=50eV, Γi0=1.0×1016cm−2s−1, and Γn0∕Γi0=10. Numerical results reproduced the evolution of feature profiles, critical dimensions, and their microscopic uniformity (or aspect-ratio dependence) on nanometer scale, depending on substrate temperature, incoming flux of oxygen and etch byproducts, and sticking probability of etch products and byproducts on feature surfaces: the lateral etching on sidewalls is suppressed by surface oxidation thereon. The oxidation also reduces the etch rate on bottom surfaces, leading to a transition from regular to inverse reactive ion etching (RIE) lag with increasing flux of oxygen; in practice, the RIE lag remains almost unchanged for narrow space features owing to reduced oxygen fluxes thereinto, thus leading to regular and inverse RIE lags coexistent in a series of different pattern features. The deposition or redeposition of etch products (desorbed from feature surfaces) onto sidewalls results in the sidewall tapering, which is more significant for narrower space features; in contrast, the deposition of byproducts (coming from the plasma) onto sidewalls results in the tapering, which is more significant for wider features. Synergistic effects between the deposition of etch products/byproducts and surface oxidation enhance the passivation layer formation on feature surfaces, which in turn increases the sidewall tapering and the degree of regular and inverse RIE lags depending on feature width. The present model also enabled the authors to simulate the surface reaction multilayers and passivation layers on atomic scale, along with their chemical constituents and surface roughness.

Surface roughening and rippling during plasma etching of silicon: Numerical investigations and a comparison with experiments

“…The ASCeM-3D methodology has been described in part in our previous papers, 46,47 together with the surface chemistry and kinetics concerned, which is basically an extension of the ASCeM-2D. [40][41][42][43][44][45] In more detail, the simulation domain is a square W ¼ 50 nm on a side with a depth of 630 nm, consisting of a number of small cubic cells of atomic size L ¼ q Si À1/3 ¼ 2.7 Å (185 Â 185 Â 2333 % 8 Â 10 7 cells in total), where q Si ¼ 5.0 Â 10 22 cm À3 is the atomic density of Si substrates. The substrates initially occupy a lower 620-nm-deep layer therein (or the substrate surfaces are initially flat, being located 10 nm downward from the top of the domain).…”

“…The etch/sputter products are assumed to stick or redeposit on all feature surfaces (blank, chlorinated, oxidized, and deposited surfaces) with a probability S q ¼ 0.05; 42 otherwise, they are reemitted thermally or reflected randomly with a probability (1 À S q ) from the surface into vacuum, which then move further toward another surfaces of the feature or go out of the simulation domain. Similarly, the incoming etch/sputter byproducts are also assumed to stick or deposit on all feature surfaces with a probability S p ¼ 0.05, 42 and otherwise, they are reemitted thermally with a probability (1 À S p ). The removal of deposited surfaces is taken to be caused by etching and/or sputtering through energetic Cl þ ion bombardment, according to the same chemistry and kinetics as mentioned above; then, the deposited Si, SiCl x , SiO y , and SiCl x O y are assumed to be removed or desorbed from the surface, thermally or isotropically with the cosine law, similarly as for the desorption of etch/sputter products and the surface reemission or reflection of neutrals.…”

“…The particles are assumed to move straight from the top of the simulation domain onto substrate surfaces and then into microstructures thereon, without collisions with other particles in vacuum, where their transport is calculated every movement of step L, taking into account geometrical shadowing effects of the structure; then, the particles are assumed to reach the surface if there is a Si atom in any of the 26 cells neighboring the cell which the particle concerned is in. The local surface normal and thus the local angle h of incidence on microstructural feature surfaces is calculated by using the extended four-point technique 40,42,46,48 for 5 Â 5 Â 5 neighboring cells (125 cells in total) at around the substrate surface cell that the particle reaches, as shown in Fig. 1; this is one of the key procedures in the cell-based simulations such as ASCeM, because the surface chemistry and kinetics calculations rely crucially on the local incidence angle h, as mentioned below.…”

“…21 We have developed a semi-empirical, atomic-scale cellular model (ASCeM) for simulating plasma-surface interactions and the resulting feature profile evolution during plasma etching of Si, based on the Monte Carlo algorithm with the cell removal method for interface evolution; [40][41][42][43][44][45][46][47] in effect, a two-dimensional ASCeM model (ASCeM-2D) reproduced the evolution of nanometer-scale profile anomalies on sidewalls and bottom surfaces of the feature, together with pattern-size or aspect-ratio dependences of them, during Si trench etching in Cl 2 and Cl 2 /O 2 plasmas. [40][41][42][43][44][45] However, the ASCeM-2D was found to be insufficient to reproduce the nanoscale surface roughness formed during plasma etching, probably because the formation and evolution of roughness is 3D as well as stochastic. 24,28 Therefore, we have recently developed a 3D ASCeM model (ASCeM-3D) to simulate the evolution of 3D feature profiles, including nanoscale surface features such as roughness, during Si etching in Cl-based plasmas.…”

Kinetic Monte Carlo simulations for transient thermal fields: Computational methodology and application to the submicrosecond laser processes in implanted silicon

“…Its applications cover different topics: e.g., erosion [1], crystal growth [2], surface modification by plasmas [3,4], cell and macromolecule interactions [5], and so on. The KMC method simulates the kinetics by a stochastic sequence of elementary events (e.g., displacement, chemical reaction, bond formation, dissolution, etc.)…”