A New Approach for Reducing Global Warming Emissions from Etching by Controlling Ion Energy and Neutral FluxesEPA Grant Number: R831459
Title: A New Approach for Reducing Global Warming Emissions from Etching by Controlling Ion Energy and Neutral Fluxes
Investigators: Hershkowitz, Noah
Institution: University of Wisconsin - Madison
EPA Project Officer: Bauer, Diana
Project Period: December 1, 2003 through November 30, 2006
Project Amount: $324,812
RFA: Technology for a Sustainable Environment (2003) RFA Text | Recipients Lists
Research Category: Nanotechnology , Sustainability , Pollution Prevention/Sustainable Development
Fluorocarbon-based gas mixtures are currently used in plasma etching of silicon-based dielectric materials by the semiconductor industry. Non-fluorocarbon feed gases have been used, at a research level, in etching dielectrics such as silicon dioxide. Among these experimental feed gases was a mixture of nitrogen trifluoride (NF3) and a light hydrocarbon, which yielded satisfactory etch rates by industrial standards. According to literature, the use of this gas mixture had been shown to markedly reduce fluorocarbon (CxFy) gas emissions from the exhaust of a plasma etch tool. However, a combination of NF3 and a light hydrocarbon cannot replace fluorocarbons in semiconductor device manufacturing unless etch processes performed with NF3/CxHy mixtures can meet other industrial processes requirements including etch selectivity, which is probably the most important criterion. In this particular case, highly selective etching of SiO2-based dielectrics over silicon nitride (high etch rate of the dielectrics compared to Si3N4 etch rate) is of industry's concern.
Fortunately, gas chemistry in the bulk plasma is not the only factor that controls substrate surface mechanisms, which lead to etching. Ion flux and energy for each plasma species play an equally important role in etch selectivity. In industrial dielectric etching, ion energy distribution at a substrate is very broad (from zero to a maximum energy as allowed by the applied power) because of low radio frequency (0.1 - 10 Mhz) of substrate bias power. By gaining a better control of the ion energy distribution and therefore limiting surface reactions to those facilitated by a desired ion energy, improvements in etch selectivity may be possible. In fluorocarbon-based plasmas, a thin (a few nanometers thick) intermediate layer deposited on the substrate governs etch reactions that are being carried out by bombarding ions and neutral species. If the same type of intermediate thin films can be derived from species in an NF3/hydrocarbon plasmas, the same quality of etch selectivity theoretically can be achieved.
For developing selective etching in NF3/hydrocarbon (possibly C2H2 or C2H4) plasmas, SiO2 will be used as a representative material of the SiO2-based dielectrics class. Etching of SiO2 over Si3N4 will be performed in a high-density plasma source. Improvements of etch selectivity will be attempted by means of controlling the ion energy distribution function (IEDF) at the substrate surface. Since etch or deposition mechanisms are believed to be governed by the energy of ions striking the substrate, it will be easier to study the effects of ion energy on surface mechanisms if the incoming ions are monoenergetic. By using a very high frequency (VHF) compared to the plasma frequencies of the relevant ions in the system, the IEDF can be made narrower than 10 eV wide. The existence of different "threshold" energy values for net etching of silicon oxide and silicon nitride will be taken advantage of, in order to allow etching of SiO2 while prohibiting etch reactions of Si3N4. Measurements of etch rates will be obtained with laser interferometry, and surface conditions of post-processing samples will be characterized with x-ray photoelectron spectroscopy. Plasma density, electron temperature, plasma potential, IEDF, and neutral fluorine density will be obtained and used for determining how ion and neutral fluxes and ion energy affect a substrate surface in this particular process gas chemistry. Eventually, a time-averaged IEDF consisting of a sharp peak at a low energy and another sharp peak at a higher energy will be used to provide, respectively, an etch reaction-regulating thin film and high energy ions for net etching reactions of SiO2 without net etching of Si3N4.
From experimental data to be obtained in the future, values of etching threshold ion energy for both SiO2 and Si3N4 will be used as the basis for improving selectivity. The types of intermediate layers on both materials that will allow good selectivity can possibly be produced when their properties and compositions are strongly affected by changing substrate bias voltage at a sufficiently high radio frequency for the sinusoidal waveform (before modulation is superimposed). Conditions that lead to the SiO2 threshold being significantly lower than the threshold of Si3N4 will be studied in details. Although expected results on selectivity will be exclusive for the plasma etch tool(s) to be used in this study, an understanding of how ion flux and energy affect the formation of an appropriate intermediate layer is essential for adapting this approach to other plasma etch systems, i.e. commercial etch tools.
Estimated Improvement in Risk Management: Our preliminary results from earlier experiments have shown that CF4 emission can be reduced by a factor of 7. A typical dielectrics etch process in a capacitive tool consumes about 100 sccm or more of source gases. Our processes to be developed are based on high-density plasma tools that have an operating pressure between 1 to 20 mTorr. With low-pressure operations, low flow rates will be required instead of high fluorocarbon flow rates necessary for a high-pressure process in a capacitive plasma system. Our processes will use only a maximum of 20 sccm of inlet gases (NF3 and hydrocarbons). If a CF4 emission is markedly reduced per one sccm of the outlet gas flow from the turbo pump, having a smaller total waste gas flow rate will further cut down on emissions as compared to those from capacitive tools. In this case, the emission can be reduced by a factor of at least 35, 7 from the use of NF3/hydrocarbon process times a factor of 5 from the reduction of flow rates to an abatement device.