ALBERT

Notes: This article reports a simulation of argon inductively coupled plasma. Experimental measurements of the electron energy distribution function (EEDF) are fit to a power-law model and used to calculate electron impact rate coefficients in the simulation. Simulation results are compared to experimental measurements of electron density and temperature with good agreement, especially at the lower pressures investigated. At higher pressures, the disagreement between experiment and model is analyzed in terms of the nonlocality of the EEDF. Diffusive transport, neutral heating, gas phase electron impact reactions, and surface quenching all contribute to the predicted metastable profiles. Predicted metastable densities and neutral gas temperatures are compared to experimental results from the literature with reasonable agreement. © 2002 American Institute of Physics.

Notes: Surface recombination coefficients of O and N radicals in pure O2 and N2 plasmas, respectively, have been estimated on the stainless steel walls of a low-pressure inductively coupled plasma reactor. The recombination coefficients are estimated using a steady state plasma model describing the balance between the volume generation of the radicals from electron-impact dissociation of the parent molecules, and the loss of the radicals due to surface recombination. The model uses radical and parent molecule number densities and the electron energy distribution function (EEDF) as input parameters. We have measured the radical number density using appearance potential mass spectrometry. The parent neutral number density is measured using mass spectrometry. The EEDF is measured using a Langmuir probe. The recombination coefficient of O radicals on stainless steel walls at approximately 330 K is estimated to be 0.17±0.02, and agrees well with previous measurements. The recombination coefficient of N radicals is estimated to be 0.07±0.02 on stainless steel at 330 K. © 2000 American Institute of Physics.

Notes: A tuned, cylindrical Langmuir probe has been used to measure the electron energy distribution function (EEDF) in atomic and molecular gases in an inductively coupled plasma. We have discussed the precautions necessary for making Langmuir probe measurements in fluorocarbon plasmas. The ionic and neutral composition of the plasma is measured using mass spectrometry. While the EEDFs in argon are non-Maxwellian, the EEDFs in molecular gases are found to be approximately Maxwellian at low pressures (〈20 mTorr) in the gases studied (N2, O2, CF4). The EEDFs in argon–molecular gas mixtures change from Maxwellian to two-temperature distributions, as the fraction of argon is increased in the plasma. At higher pressures, the molecular gases exhibit EEDFs reflecting the electron collision cross sections of these gases. In particular, N2 plasmas show a "hole" in the EEDF near 3 eV due to the resonant vibrational collisions. O2 plasmas show a three-temperature structure, with a low-energy high-temperature electron group, a low-temperature intermediate-energy electron group, and a high-temperature high-energy tail. The fractional degree of dissociation in the N2 and O2 plasmas is below 0.1, with the parent molecules and molecular ions being the dominant species. The spatial variation of the EEDF in an oxygen plasma at low pressures (10–20 mTorr) is found to be consistent with the nonlocal theory. © 2000 American Institute of Physics.

Notes: Sputter yields Y and sticking coefficients S are essential inputs for feature profile evolution studies. Molecular dynamics simulations are used to compute sputter yields and sticking coefficients for Cu+ ions impinging on a Cu surface at various incident energies 15〈Ei〈175 eV, and incident angles 0〈θi〈85°. Threshold energies for sputtering Eth are also predicted and shown to vary with θi. We show that for energies below what is experimentally considered as threshold for physical sputtering (Eth(expt)∼60 eV) a yield between 0.01 and 0.1 Cu/ion is observed for some off-normal angles of incidence [C. Steinbrüchel, Appl. Phys. Lett. 55, 1960 (1989)]. We show that Y∝Ei−Eth below Eth(expt) when Y is a maximum with respect to θi (at θi=45°). We find that Y∝Ei1/2−Eth1/2 at other angles of incidence. We show that S is sensitive to Ei and θi in this regime. In particular, when θi=85°, we see that ln S∼1/Ei for Ei≥20 eV. We discuss some assumptions commonly used in profile simulation studies which may now be relaxed, with an eye toward improving the predictive power of those simulations. © 1999 American Institute of Physics.

Notes: The development of a Tersoff-type empirical interatomic potential energy function (PEF) for the Si–C–F system is reported. As a first application of this potential, etching of a:Si by CF3+ using molecular dynamics (MD) simulations is demonstrated. Aspects of CF3+ ion bombardment through a fluence of 4×1016 cm−2 are discussed, including overlayer composition and thickness, Si etch yields, and etch product distributions. The formation of a 1-nm-thick steady-state SixCyFz overlayer occurs in the simulation, and this layer is an active participant in the etching of the underlying Si. At an ion energy of 100 eV, a steady state the etch yield of Si is predicted to be 0.06±0.01 Si/ion. A comparison of the simulation findings and experimental results from the literature leads to the conclusion that the new PEF performs well in qualitatively modeling the atomic-scale processes involved in CF3+ ion beam etching of Si. Simulations of this kind yield insight into fluorocarbon etch mechanisms, and ultimately will result in phenomenological models of etching by fluorocarbon plasmas. © 1999 American Institute of Physics.

Notes: Point-of-use plasma abatement (PPA) has been proposed as one way to eliminate perfluorinated compound (PFC) emission from various tools used in integrated circuit manufacturing. PPA employs a high density plasma between the process tool turbomolecular pump and the backing pump. Oxygen is added to the process tool effluent upstream of the PPA tool. The mixture of oxygen and PFC-containing tool effluent enters the PPA tool and the PFCs are converted to products that can be scrubbed downstream of the backing pump. In this article, we present a model for the PPA tool operation, illustrating the principles with a mixture of C2F6/O2. A plasma model is coupled to a neutral transport and reaction model, including electron-impact molecular dissociation and subsequent gas phase chemistry. © 1999 American Institute of Physics.

Notes: A cylindrical Langmuir probe has been used to measure the electron energy distribution function (EEDF) in atomic and molecular gases in a shielded inductively coupled plasma. We report the EEDFs in these gases as a function of pressure. While the electron properties in a discharge depend on the product of the neutral number density (N0) and the effective discharge dimension (deff) for a given gas, this dependence is different for different gases. We find that pressure is a convenient parameter for comparison of the EEDFs in these gases. The EEDFs in inert (Ar, Kr, Xe) and molecular gases (H2,N2,O2,H2O,CO2,CF4) in the low pressure limit (below 1 mTorr) show a "three-temperature" structure. Since this wide range of gases display similar EEDF shape, we propose this structure to be common to all gas discharges in this limit. The EEDF in all of the gases shows a two-temperature structure with apparent tail depletion at 3 mTorr. The similarity of the EEDFs in all of the above gases is probably due to nonlocality of the electrons at these low pressures. The molecular gases exhibit a nearly Maxwellian EEDF between about 10 and 30 mTorr, while the EEDF in argon is non-Maxwellian in this range. At pressures above 30 mTorr, the EEDFs in molecular gases show deviations from a Maxwellian distribution, reflecting the electron-neutral collision cross sections of each gas. The EEDFs in molecular gases at 100 mTorr show significant deviations from a Maxwellian distribution. We find that the EEDF in molecular gases can be approximated by a Maxwellian distribution over a fairly large pressure range of 3–50 mTorr for the purposes of modeling these discharges. © 2000 American Institute of Physics.

Notes: Comparative analyses of molecular dynamics (MD) simulation studies of reactive ion etching of Si are presented. A recently developed empirical potential is used to model the Si–F system, and applied to the simulation of Si etching with energetic F+ at 10, 25 and 50 eV. These results are compared to those of a similar study using the Stillinger-Weber Si–F potential. This analysis leads to the expected result that different potentials lead to quantitatively different results with regard to Si etch yield, surface structure and composition, etching mechanisms, and product distributions. More importantly, however, it attests to the robustness of the qualitative nature of these results. The degree of qualitative agreement between systems studied with the two potentials is high enough for us to conclude that MD simulations have revealed valuable qualitative insights into the complicated system of reactive ion etching of Si. © 2000 American Institute of Physics.

Notes: Surface reactions of atomic halogen atoms play important roles in various plasma etching processes, commonly used in microlectronics manufacturing. However, relatively little is known about the surface chemistry of these key reactive intermediates. Previous measurements of the recombination coefficients of Cl, Br, and F on various surfaces in a molecular beam apparatus indicated that the recombination reaction is pseudofirst order [G. P. Kota, J. W. Coburn, and D. B. Graves, J. Vac. Sci. Technol. A 16, 270 (1998); 16, 2215 (1998)]. One mechanism that would result in pseudofirst order kinetics is a two-step process in which the first halogen atom adsorbs into a relatively strongly bound chemisorbed state, and the second atom reacts with it either through a direct reaction, or after being physisorbed onto the halogenated surface. In this article, we report experiments in which surfaces are first exposed to a molecular beam of one type of halogen atom, then the surface is exposed to a second type of halogen. During the second exposure, the heteronuclear reaction product is monitored with a mass spectrometer. Finally, the surface is sputtered and the mass spectrometer is used to detect any remaining presence of the original halogen atom. Analogous experiments were also performed with isotopically enriched mixtures of chlorine. These experiments unambiguously demonstrate that halogen atom surface recombination involves a two step adsorption-abstraction mechanism. Under all conditions studied, the surface recombination reactions proceeded at rates on the order of surface collision frequencies. The relative magnitudes of the heteronuclear rates (as a function of surface composition and halogen atom type) scaled in the same way as the homonuclear recombination probabilities measured previously. In every case examined, after the second halogen exposure, the surface retained a significant coverage of the halogen that had been originally exposed to the surface. This leads to the conclusion that only a fraction of the strongly bound surface sites are available for abstraction by free radical attack. Absolute calibration of the incident and evolved species fluxes allowed an estimate to be made of the reactive site densities for several surfaces. These ranged from 1012 to 1015 cm−2 depending on the surface. © 1999 American Institute of Physics.