Feature profile evolution during the high density plasma etching of polysilicon

Abstract

This thesis work deals with one of the critical steps in the manufacturing of modem integrated circuits - the plasma etching of thin polysilicon films used to form the metaloxide- semiconductor transistor gate. The etching of very small features (-0.25 jim) in the -2500 A thick films, performed at low operating pressures (-10 mTorr), must be accompanied with minimal etching artifacts that can degrade device performance. This thesis aims to understand the causes for the etching artifacts observed during the etching of polysilicon line-and-space features in C12 and HBr plasmas so that better etching recipes can be developed. The second goal of this work was to develop a generalized simulator capable of predicting the feature profile evolution for the above plasma etching system as well as systems involving other materials and chemistries. The 2V2-dimensional simulator developed used Monte Carlo techniques to compute the transport and surface kinetics combined with a cellular representation of the feature. Using the Monte Carlo algorithm permitted the incorporation of all dominant physical and chemical mechanisms of the etching process such as angle-dependent ionenhanced etching, physical sputtering, ion scattering, surface recombination, plasma deposition, and line-of-sight re-deposition without encountering numerical difficulties. The technique allowed the calculation of surface kinetics rates based on the surface composition; simultaneous composition-dependent etching and deposition could be handled easily. A modification of the cellular representation of the feature was developed to determine neutral species interactions with the surface correctly. A surface normal calculation algorithm involving least-squares fitting of the surface was developed to handle specular ion scattering. Designed sets of plasma etching experiments were performed on photoresist masked and silicon oxide-masked polysilicon samples using C12 and HBr chemistries varying the inductive power (controls the ion density, radical concentrations), the rf biasing power (controls the ion energy) and the gas flowrate (controls the reactant and product concentrations). The interesting features exhibited in the experimental profiles included: 1) the increased sidewall deposition associated with photoresist-masking and isolated features, 2) the greater curvature of the sidewalls associated with the combination of photoresist and Cl2 plasmas, 3) the more vertical sidewalls achieved with HBr, 4) the double faceting of the feature sidewalls under etching conditions accompanied by significant deposition, 5) the delay in the onset of microtrenching at the feature bottom while etching photoresist-masked samples with C12, 6) the greater microtrenching exhibited with silicon oxide-masking and C12 plasmas, and 7) the lack of microtrenching for the HBr etching. The experimental results suggested strong dependencies of microtrenching, tapered sidewall profiles and photoresist-mask faceting on the feature aspect ratio, product formation rate and product residence time in the etching chamber. The etching artifacts were explained using the profile evolution simulator. The microtrenching was associated with two mechanisms - ion scattering from tapered sidewalls and the focussing of directional ions by bowed sidewalls onto the feature bottom. The former mechanism led to trenching initially while the latter mechanism gained importance midway during the etching. The absence of tapered sidewalls initially and the relatively straight sidewall profiles developed during the etching explained the non-occurrence of microtrenches when using HBr. Under processing conditions accompanied by significant deposition, facets at two distinct angles were predicted. The top facet depended on the composition of the material on the photoresist-mask line and its etching angular dependence. The lower facet angle and the polysilicon sidewall profile were governed by the feature aspect ratio, the sticking probabilities and fluxes of the depositing material, and the depositing material etching angular dependence. The phenomenon of feature charging was incorporated in the Monte Carlo simulator to understand its role in the profile evolution. Two electrical approximations were made for the feature - the perfectly insulating and a novel resistive approximation. With an insulating feature, the potential profiles were obtained by determining the space charge on the feature surface and solving Poisson's equation over the entire simulation domain. Calculation of the potential profiles with the resistive feature representation involved treating the feature as a large resistive network, determining the steady-state currents to the feature surface and solving the conductivity equation and Laplace's equation in the solid and gas, respectively. The role surface and bulk conductivities played on the potential profiles were studied. The potential profile in a completely etched polysilicon (conducting) feature with a silicon oxide (insulating) feature bottom was generated. Higher ion currents were calculated at the lower part of the polysilicon sidewall. These currents can etch the passivating material deposited at lower portion of the sidewall enabling spontaneous etching of the sidewall, and cause notching of the sidewall.