Abstract

Some recent results obtained using three quite different experimental approaches to fast in-situ surface metallization are presented. One of the goals is to make high density interconnects for multichip modules with metal contact width, height and pitch respectively of the order of 10, 5 and 25 μm. Direct writing speeds should be in excess of 1 cm/s. The first approach tried is to push classical pyrolytic laser chemical vapor deposition of copper from its bis-hexafluoroacetylacetonate Cu(hfa)₂ to its limits. This is done by increasing the Cu(hfa)₂ vapor pressure and by seeding the transparent surface of our substrate with a thin layer of a strongly light-absorbing substance. Speeds in the order of a few mm/s have already been obtained with good electrical properties and adhesion. The second approach tested implies metallization by thermal in-situ decomposition of a metal complex layer on a surface with a laser beam. The precursors used in this work are metal cluster coordination complexes, in particular au₅₅(pɸ₃)₁₂c1₆ (ɸ stands for phenyl group -C₆H₅). Such molecules have a high metal content and advantageous thermal decomposition properties which enable writing speeds in excess of 1 cm/s while maintaining acceptable electrical properties of the metal lines. Finally, in a third approach, the fast in-situ step is limited to deposition of a thin metal line on the surface. These lines of only a few Å height are then "developed" in a second step to much high (of the order of one μm) and better conducting metal lines. This second step may be relatively slow, as many devices which have been patterned on their surface with the thin metal "prenucleation" lines can subsequently be developed simultaneously in a parallel processing step. An example of this approach is surface seeding using metals like Pt, W or Ir, followed by development by either selective low pressure CVD of copper from Cu(hfa)₂, selective decomposition of a copper formate solid surface film, or selective electroless deposition from a copper bath. Possibilities of processing with much higher resolution using focussed charged particle beams are also briefly discussed.

© 1991 Optical Society of America