Dielectric film etching in semiconductor device manufacturing: Development of SiO 2 etching and the next generation plasma reactor

Abstract

The progress of the dry etching technology for dielectric materials and the ASET project were reviewed. Reactive ion etching (RIE) replaced a wet chemical etching in 1970s and has been forwarding the size shrinkage of electron devices and the improvement of device capacity and speed greatly. The technology is continuously evolving this last two decades to meet the requirements from the device performance and the productivity. Developments of SiO 2 etching technology during that period were introduced with the concept to use fluorocarbon (CF) gases. Magnetically enhanced plasma reactor was also reviewed especially about the improvement for a charging damage. The RIE system has been improved with a new magnetic field arrangement and a new control methodology of chemical species, especially for a self-aligned contact (SAC) hole etching. The capacitive-coupled (parallel plate) plasma sources are the standard for the SiO 2 etching nowadays. Conventional developments were conducted in a very empirical way, such as a trial and error, with many speculation using qualitative data. This approach requires more and more resources and time for the development of future devices with a design rule below 100 nm in the SOC (system on a chip) era. It is necessary to establish a systematic methodology for process development and qualification. ASET Plasma Laboratory had been found to research a basis for the systematic development of the plasma etching technology. CF plasma for the etching of high-aspect-ratio contact holes in SiO 2 was investigated intensively in the 5-year program that had finished in March 2001. They introduced five plasma sources that can etch 0.1 μm contact hole on a 200 mm wafer in production, and state-of-the-art diagnostics tools for the plasma and etched surface. The SiO 2 etch mechanism was revealed from the etch species generation to the reaction in a deep hole. The number of electron collisions to CF gas molecule is proposed as an important parameter to control the gas dissociation and etch species flux to the surface. An etch reaction model was also proposed using the estimated surface reaction probability that is a function of ion energy and CF polymer thickness that reduces the net ion energy to the reaction layer. The CF polymer thickness was determined by a balance equation of generation term (radical fluxes) and loss terms (etching by ions, radicals and outflux oxygen from SiO 2). A program was developed and successfully predict the etch rates of Si-containing materials including organic dielectrics. Requirements for the next generation plasma etch tools were also discussed.