AVS2011 Session PS-FrM: Plasma Modeling

We present results of particle-in-cell/Monte Carlo collision simulations of kinetic effects in low pressure capacitively coupled plasma discharge.[1] In particular, we examine discharges of various gases (including Ar , Xe, and others) in the pressure range of 10s of mT and the frequency range of 10s of MHz. We track the formation of high energy electrons (e.g., at the ionization threshold or greater) as a marker for enhanced ionization, and look at the effects of elastic and inelastic collisions on the formation of these high energy electron bunches. [2,3] We show results for 2D and 3D simulations where we include density gradient effects, and results for plasma chemistry effects on the bulk electron energy distribution function and the ion energy distribution function at a plasma surface interface. We discuss the role of the bunches on electron heating in the plasma bulk and on their presence on how electron heating is treated in fluid simulations of plasma sources.

The development of technologies for the plasma treatment of living tissue is in large part based on controlling plasma sources to deliver the desired fluxes of radicals and ions to surfaces. This process is complicated by scientific and technical issues. From a scientific standpoint, although it is generally accepted that reactive oxygen species (ROS) and reactive nitrogen species (RNS) are important in, for example, wound healing, sterilization and cancer treatment, it is not clear which species and in what proportions are optimum for each type of treatment. The situation becomes more complex when considering the UV photons, energetic ions and electric fields produced by the discharge which also interact with the tissue. From a technical view point, the interaction between the tissue (and wounds in particular), the surrounding materials and the plasma can significantly affect the plasma. For example, the shape of the wound and the permittivity of the fluid in a wet wound can warp local electrical fields which then feed back to the plasma. Given this complexity and interdependencies, computer modeling of plasma-tissue interactions might provide insights to these interactions. In this talk, results from computer modeling of plasma-tissue interactions will be discussed. The modeling platform solves for charged particles, neutral and photon fluxes while also solving Poisson’s equation, and resolving spatial scales on reactor-to-cellular levels. Plasma transport through gases and liquids are included. Two types of plasma sources will be considered - dielectric barrier discharges (DBDs) where the plasma is in direct contact with the tissue and remote plasma jets, where dominantly neutral species and photons reach the tissue. We will discuss the treatment of wounds through a liquid layer covering exposed cells wherein the blood serum contains blood platelets. The characteristics of the plasma sources, and the interaction of plasma generated species and electric fields with the wound, fluid and underlying cells will be discussed.

* Work supported by the Department of Energy Office of Fusion Energy Sciences.

A key aspect of the plasma processes here considered is that some type of work is done at the plasma / surface boundary layer, and realistic simulations must therefore incorporate the surface material and the etch product chemistry. This increases drastically the complexity of the problem but is the only way to represent all of the appropriate physics. Radical species from the surface entering the gas phase will take part in the phase and surface reactions that are associated with the parent gas, including negative ion formation and electron dissociation among others.

Here we present 2D simulations of an inductively driven SF₆ silicon etch process in the GEC Reference Cell [1], building upon previous calculations of SF6 plasma chemistries using Quantemol-P [2]. Etch rate, pressure and power trends along with chamber wide contour plots of gas-phase species concentrations and fundamental plasma properties are considered. We have found a good agreement with experimental results [3], which validates the underlying model and points to the important role of simulation-assisted plasma process development and optimization.

REFERENCES

[1] P. J. Hargis et al, Rev. Sci. Instrum. 65, 140 (1994).

[2] J. J. Munro and J. Tennyson, J. Vac. Sci. Technol. A 26, 865 (2008).

[3] G. A. Hebner, I. G. Abraham, J. R. Woodworth, “Characterization of SF6/Argon Plasmas for Microelectronics Applications”, Sandia Report, Sand2002-0340 Unlimited Release, March 2002.

InP-based optoelectronic devices need reliable dry etching processes characterized by high etch rate, profile control and low damages. High density plasma etching, using inductively coupled plasma ICP reactors, has been found to be very important for the transfer of patterns from the mask to InP substrate and InP-based layers. In order to investigate the role of N₂ in the InP etching process under Cl₂/Ar/N₂ plasma discharge, we have developed an InP etching simulator permitting to determine the InP etch profile evolution through the mask as a function of the operating conditions and the initial mask geometry.

Three-dimensional measurement and prediction of atomic-scale surface roughness on etched features become increasingly important for the fabrication of next-generation devices; however, the feature profiles are too small or too complex to measure the surface roughness on bottom surfaces and sidewalls of the etched features. To predict the surface roughness on atomic or nanometer-scale, we have developed our own three-dimensional atomic-scale cellular model (ASCeM-3D) [1] and feature profile simulation. Emphasis is placed on a better understanding of the formation mechanisms of atomic-scale surface roughness during Si etching in chlorine-based plasmas and the relationship between the ion incident energy and angle and etched feature profiles.

In the ASCeM-3D model, the simulation domain is divided into a number of small cubic cells of L = r_Si^-1/3 = 2.7 Å, where r_Si = 5.0 x 10²² cm^-3 is the atomic density of Si substrates. Ions and neutrals are injected from the top of the simulation domain, and etch and/or sputter products are taken to be desorbed from etching surfaces into microstructural features, where two-body elastic collision processes between incident ions and substrate atoms are also taken into account to analyze ion reflection on etched feature surfaces and penetration into substrates. The ASCeM-3D takes into account surface chemistries based on the Monte Carlo (MC) algorithm [2-4], including adsorption and reemission of neutrals, chemical etching, ion-enhanced etching, physical sputtering, and redeposition of etch and/or sputter products on feature surfaces.

Numerical results indicated that nanoscale convex features increase in size with increasing etching or plasma exposure time, and surface roughness increases with increasing ion incident energy. The ripple structures of etched surfaces were found to occur depending on incident angle of ions. Ion reflection or scattering on etched surfaces strongly affects the evolution of feature profiles and surface roughness on atomic scale.

[1] H. Tsuda et al., Jpn. J. Appl. Phys. (2011), in press.

[2] Y. Osano and K. Ono, J. Vac. Sci. Technol. B 26 (2008) 1425.

[3] H. Tsuda et al., Thin Solid Films 518 (2010) 3475.

[4] H. Tsuda et al., Jpn. J. Appl. Phys. 49 (2010) 08JE01.

The energy of ions bombarding the substrate is critical in plasma etching and deposition of thin films, especially when precise etching without damage is required. The ion energy distribution (IED) may be controlled by applying “tailored” bias voltages on the substrate, or on nearby electrodes immersed in the plasma. A Particle-in-Cell simulation with Monte Carlo Collisions (PIC-MCC) was conducted of the application of DC voltage steps (and staircases) on an electrode, during the afterglow of a capacitively-coupled pulsed argon discharge, to control the energy of ions incident on the counter-electrode holding the wafer. Staircase voltage waveforms with selected amplitudes and durations resulted in ion energy distributions with distinct narrow peaks, having controlled peak energies and fraction of ions under each peak. A semi-analytical model was also employed to achieve “tailored” IEDs, i.e., distributions with a desired shape and energy spread (for example a nearly-monoenergetic IED with given FWHM). This was again accomplished by applying judicious voltage waveforms on the substrate electrode. Predicted IEDs were compared with experimental data. Strategies to control the energy flux of bombarding ions or to distribute the total ion energy flux to different energies were identified.

Work supported by DoE Plasma Science Center and NSF.

Selective etching of silicon oxide (SiO₂) over silicon nitride (SiN) has been widely used in microelectronics fabrication processes such as contact hole etching in self-aligned processes, formation of a stress liner, and dual/triple hard mask (DHM/THM) etching processes of dual-damascene structures. Opposite selective etching of SiN over SiO₂ with high selectivity would be also desirable for various processes. In general, when a fluorocarbon gas is used for etching processes, a carbon film tends to be accumulated on SiN surface, which is considered to reduce its etching rate. Therefore, there have been various attempts in plasma processing to increase the SiN etching rate by reducing carbon films over SiN with the use of hydrogen reactions with carbon. In such plasma processing, hydrofluorocarbon gases are typically used. In this study, we have performed molecular dynamics (MD) simulations of SiN and SiO₂ etching by CH_xF_y ions and compared their etching rates and surface chemistry, especially focusing on effects of hydrogen on the process. The reactive interatomic potential functions for atomic systems of Si, O, F, C, N, and H were developed in-house for the MD simulations code, based on atomic interaction data of small molecules in ground states obtained from ab-initio calculations. Details of the atomic potential functions used in the simulations will be presented elsewhere. Simulations are typically performed on a small block of a model substrate that consists of several thousand atoms and is subject to bombardment of energetic particles such as CH_xF_y. In the simulations, we evaluate sputtering yields, surface modification during the process, and characteristics of sputtered products. From the simulations, it has been found that hydrogen of CH_xF_y ions tends to reduce F accumulation on SiN surface, forming volatile HF, and sometimes promotes formation of cyanides such as HCN. Detailed simulation results, including sputtering yields and surface chemical compositions, will be given in this presentation.