2001 Progress Report: Alternative Wafer Cleaning Using HF-H2O ProcessingEPA Grant Number: R826737
Title: Alternative Wafer Cleaning Using HF-H2O Processing
Investigators: Sawin, Herbert H.
Institution: Massachusetts Institute of Technology
EPA Project Officer: Richards, April
Project Period: January 1, 1999 through December 31, 2001 (Extended to December 31, 2002)
Project Period Covered by this Report: January 1, 2000 through December 31, 2001
Project Amount: $340,000
RFA: Technology for a Sustainable Environment (1999) RFA Text | Recipients Lists
Research Category: Sustainability , Pollution Prevention/Sustainable Development
A novel hydrofluoric acid/water vapor (HF/H2O) process for in situ dry cleaning of microelectronic wafers was studied to: (1) understand its surface kinetics, and (2) develop it as a replacement for current wet cleaning processes. Successful development of such a dry cleaning process could greatly reduce the number of wet cleaning processes needed in microelectronics manufacturing, thereby reducing the deionized water needed for rinsing; reducing the amount of acids, bases, and solvents consumed and disposed; and improving worker safety.
The use of HF and other acids, bases, and solvents in the cleaning of wafers between process steps is problematic for the microelectronics industry in terms of environmental impact, material costs, worker safety, productivity, and disposal costs. The avoidance of aqueous baths, which risk worker exposure, would increase worker safety and reduce facility floor area needed for cleaning processes.
The objective of this project was to: (1) find the reaction mechanisms of oxide etching both HF/H2O and HF/alcohol processes; and (2) develop a vapor phase HF cleaning process to remove the metallic contamination and natural oxide on the silicon surface. Even though the HF vapor process has been studied intensively for several decades, the commercial application is not very successful due to the unknown nature of the process. This study, performed at the Massachusetts Institute of Technology (MIT), focused on possible applications in the semiconductor industry where a replacement to the aqueous phase cleaning process is desirable. The ultimate purpose of this project is to provide feasibility for the HF vapor process to be vacuum compatible and clustered with the cleaning process.
The oxide etching in the HF/H2O vapor process occurs in both a gas phase regime (sub-monolayer, monolayer, and multilayer adsorbed regime) and a condensed phase regime, depending on the partial pressures of HF and H2O in the gas phase and temperature of the substrate. The condensation of HF and H2O occurred at lower partial pressures of reactant gases than was previously predicted by vapor liquid equilibrium data. The ternary mixture of HF, H2O, and SiF4 from the oxide etching reaction caused this depression of the condensing point.
In the condensed regime, the etching rate is less sensitive to the temperature and the partial pressure of the reactants at a high pressure of HF. The etching rate in this regime generally is one to two orders of magnitude higher than that of the gas phase regime. The etching rate in this regime also is affected greatly by the mass transfer rate in the gas phase. The etching rate is proportional to a scaling factor, (QD/Lp)1/2 for the mass transport.
In the multilayer adsorption regime, the etching rate that is linearly dependent on the partial pressures of reactants is relatively low. The etching rate of oxide at a high reactant pressure can be affected by the product concentration on the surface when mass transfer resistance is present in the gas phase. The etching rate in this regime is greatly affected by the temperature of the substrate. The mass transfer rate limits the etch rate of oxide in the multilayer adsorption regime.
In the monolayer adsorption regime, the etching rate is expressed by Langmuir-Hinshelwood kinetics. The etching rate is governed by surface kinetics in this regime. Advantages of the monolayer etching regime are: (1) smoother etched surface, (2) low selectivity to tetraethyl orthosilicate, (3) haze free etched surface, (4) no metal attack, and (5) vacuum compatible process. Although the monolayer etching regime showed promising results, the etching rate in this regime is greatly affected by the surface state of the oxide layer, which often caused irreproducible etching results. The electrostatic charge on the surface and its polarities are responsible for the irregular etching results.
The positive charge enhanced the etching reaction in the sub-mono and monolayer etching regime, while negative charge mainly enhanced the etching in the multilayer etching regime. Direct ionization of HF on the oxide surface is responsible for enhancement in the monolayer regime. The multilayer is believed to form a thicker adsorbed layer by negative charge on the oxide, resulting in a higher etching rate. The adsorption of reactant also is enhanced by the vapor pressure reduction of HF and water from fluorosilicate formation, instead of SiF4 under basic condition induced by the negative charge.
The use of a clustered plasma etch, ash, plasma oxide growth, HF/water dry clean, and metal deposition sequence has been successfully demonstrated in collaboration with researchers in this lab supported by the Semiconductor Research Corporation.
HF vapor cleaning machines have been introduced commercially as a dry cleaning method since the late 1980s. None of them have been successfully applied in the manufacturing process, moreover in a vacuum cluster tool. Based on our study, we have reported a true gas phase and vacuum compatible HF vapor process operated in the monolayer adsorption regime at an elevated temperature. This process sequence is ideal for a vacuum cluster configuration, in which a single wafer is processed at a time and is not exposed in the ambient.
A "real" cluster tool type machine cannot be achieved without a vacuum compatible cleaning module because every single process in semiconductor manufacturing involves an aqueous phase or gas phase cleaning process before or after the main process steps such as chemical vapor deposition (CVD), oxidation, and photolithography. The HF vapor process is not able to remove particulate contamination on the wafer surface, therefore it might be required to have a dry particle cleaning process in order to maximize the cleaning performance of a cluster tool. This process can be accomplished by the "dry ice jet" cleaning method. We strongly believe that the combined cleaning procedure of HF vapor and ice jet cleaning operated in a vacuum cluster tool can replace more than 50 percent of liquid phase cleaning steps.
Future research will focus on refinement of these HF/water vapor wafer cleaning processes, which could reduce chemical consumption by a factor of five. Although the application of an in-line type vacuum cluster machine is now at the initial stage in industry, there is no doubt that this process is a future manufacturing method. At the present time, higher productivity, environmental concern, and cost reduction are the main driving forces behind the development of cluster tools.