metal patterns onto thin, prefired ceramic layers, stacking the layers, and firing at high temperatures to densify and strengthen the mass. For alumina, which is the ceramic presently used, the choice of conductor metal is restricted to molybdenum because of the need for excessive firing temperatures. However, the dielectric constant of alumina is too high and molybdenum is too resistive, resulting in relatively poor electric characteristics for the transmission lines. Also, the thermal coefficient of expansion of alumina is not the same as that for silicon, leading to high stress and failure when chips are attached and powered.

New substrate materials must be developed that have lower fabrication temperatures, lower dielectric constants, high strength, and thermal expansion coefficients matched for those of silicon. Compatibility during processing between the substrate matrix, the metal conductors, and the semiconductor chips has been elusive and will require more research into interface interactions, fracture propagation, and thermal relaxation mechanisms. If ceramics are to remain competitive for chip substrates, dielectric constants must be reduced by techniques such as microscopic porosity, and surface roughness must be eliminated to facilitate lithographic processing of metal lines.

There is a significant trend away from ceramics and toward polymers for substrate packaging materials. Polymer layers have intrinsically lower dielectric constants and preparation temperatures, can be applied onto a surface with controllable flatness, and are compatible with lithographic processing of metal transmission lines. Mechanical stability and heat dissipation are areas requiring further research, but hybrid ceramic-polymer layers are now being produced in which the polymer layers are processed on top of or adjacent to chips when metal lines of the highest density and highest performance are required. New polymeric materials will be needed in which thermal expansion coefficients are reduced to match the chip and ceramic, strength is increased to prevent cracking, and dielectric constants are further reduced, perhaps by the use of fluorocarbon-like chemistries. Adding moisture resistance, thermal stability, and perhaps photosensitivity to the list illustrates the complex combination of requirements that apply to new polymer materials.

The printed circuit boards onto which chip-carrier modules are mounted have been made of epoxy-filled glass cloth for some time. The reinforced structure has a thermal expansion coefficient characteristic of the composite, with strength being provided by the glass. Research on this material is seeking to lower the dielectric constant by using aromatic polyamide polymers to replace the glass, and to replace the epoxy with a more compatible material. Structural integrity and compatibility with high wiring densities are further areas for research.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement