AVS 45th International Symposium
    Plasma Science and Technology Division Thursday Sessions
       Session PS-ThM

Paper PS-ThM10
Metal Flux Ionization Fraction in Copper Ionized Physical Vapor Deposition@footnote 1@

Thursday, November 5, 1998, 11:20 am, Room 318/319/320

Session: Plasma Applications in Copper Metallization
Presenter: A.E. Wendt, University of Wisconsin, Madison
Authors: T.G. Snodgrass, University of Wisconsin, Madison
J.E. Foster, University of Wisconsin, Madison
S. Lu, University of Wisconsin, Madison
A.E. Wendt, University of Wisconsin, Madison
J.H. Booske, University of Wisconsin, Madison
J.L. Shohet, University of Wisconsin, Madison
Correspondent: Click to Email

A characterization and modeling effort is directed at a more complete understanding of the potential and limitations of copper ionized physical vapor deposition (IPVD) for damascene processes. An rf inductively-coupled IPVD tool operating in argon includes a dc magnetron sputter source mounted in the top of an 18" D chamber. A 14'' D internal single-turn rf induction antenna is positioned between the magnetron and a 12" D substrate holder. Improved filling of high aspect-ratio features depends on the degree to which metal atoms are ionized as they pass through the rf plasma. To identify factors governing the "ionized flux fraction," measurements of metal properties have been made in the gas phase and at the substrate. Optical spectroscopy and Langmuir probes measure gas phase concentrations of neutral and ionized copper, and an improved quartz crystal microbalance@footnote 2@ is used to determine both neutral and ion fluxes at the substrate. The sputter rate from the target as well as the deposition rate radial profile at the substrate location have been characterized in detail. Results with and without the rf plasma show that the rf induction plasma has the primary effect of increasing the ionized metal flux and has only a minor effect on the flux of neutral copper. Selfsputtering of the internal rf antenna has also been examined, and methods to control it will be presented. A model has been constructed that, along with the measurements described, provides a physical explanation of the IPVD operating characteristics. @FootnoteText@ @footnote 1@This work supported by NSF grant #EEC8721545 @footnote 2@T. G. Snodgrass, W. Wang, J. H. Booske, A. E. Wendt, J. L. Shohet, submitted to Rev. Sci. Instr.