DARPA, industry mull MCM foundry efforts

Electronic News, Feb 8, 1993 by Jack Robertson

AUSTIN, TEX.--Development of next-generation multichip module production equipment, including a pilot foundry consortium, emerged as key goals last week in a meeting of industry groups and the Defense Advanced Research Projects Agency (DARPA) here.

Also a hot topic at the kickoff workshop to draft a plan for an MCM consortium was creating synergy with the manufacturing needs of flat panel display makers.

The workshop developed preliminary technical inputs in eight MCM process areas to be used in submitting an eventual consortium foundry proposal to DARPA. Bill Barker, DARPA program manager for the project, said he hopes the Pentagon agency can seek proposals early this year to fund the MCM consortium.

The pilot foundry goal is to make 15-inch square substrates with defects as low of 0.1 percent per layer. The substrates would be populated with chips, and then cut into 6-inch square MCMs for packaging and testing.

The MCM consortium also hopes to standardize equipment needs with the evolving U.S. flat panel display industry--especially since both products will be using semiconductor-type processes on 14-inch to 15-inch substrates. "If the same production equipment can be used by both industries, we will create economies of scale for suppliers that should drive down costs and increase performance levels," said Len Schaper, director of the University of Arkansas High Density Electronics center and workshop chairman.

In addition to participation by the Electronic Industries Association (EIA) and Semiconductor Equipment and Materials International (SEMI), more than 175 workshop attendees were told that companies backing the effort would include IBM, Motorola, Texas Instruments, nChip, MCC (Micro-electronics and Computer Technology Corp.), MicroModule Systems, and Polycon.

Although the lure of new funding drew a big attendance, one possible restraint is DARPA's requirement to have industry put up equal matching funds for any grant for the pilot foundry and equipment development. The predominantly small business equipment producers don't always have the financial resources to come up with matching funds of their own. Sources said some of this burden will be assumed by the much larger MCM builders, who are willing to absorb some of the cost because of the essential need for next generation equipment.

The MCM consortium plan to be submitted to DARPA is aimed at joint development of all the critical process equipment types needed in depositing substrate patterns, forming insulating layers and vias for connections, populating chips on the substrate, dicing and packaging and testing the completed modules.

The technical inputs for the projected consortium roadmap could set off vigorous debates among various equipment vendors--all touting their particular technology to be selected. The equipment areas included: lithography, laser processing, CVD/sputtering, plating/wet etch, substrate materials, ceramic processing, assembly and test.

"We need new generations of equipment--significantly lower in cost, with higher yields over much larger substrate areas," said Mr. Schaper. "Everyone thinks that since MCMs may need only (line geometries of) 10 microns width that you can just use old semiconductor production equipment in a class 10,000 clean room. Nothing is further from the truth. We need extremely low defects -- as low as 0.1 per layer over a 15-inch square substrate."

DARPA's Mr. Barker said the first consortium funding will be aimed at joint development of equipment. Each category of gear will be designed to work together in any MCM foundry. After the critical equipment is developed, DARPA is then planning to fund a pilot foundry.

Industry sources here, however, said individual MCM companies are likely to buy the next generation MCM production equipment as quickly as possibly to install in their own plants. "The market for the new MCM equipment won't have to wait on any pilot foundry. I would expect much of the equipment will be up and running at all the MCM manufacturers long before any pilot foundry is set up," said a MCM pioneer executive, who didn't want to disparage the pilot foundry plan.

Mr. Schaper said it isn't clear which equipment type should be used to form lines on substrates--whether wet etch, dry etch, or laser-assisted deposition. Similarly insulating layer deposition can be done by anodizing metal, thermal oxidation, spray coating, curtain coating, screening, laminating, and CVD. Connecting vias can be formed through layers either by E-beam, RIE, wet etch, punch holes, and lasers. He felt that the common form of drilling used in today's MCM laminate processes won't produce the sharp 7 micron holes needed for new highly dense MCMs.

Sources expected an entrenched battle between existing MCM technologies in how to interconnect chips to the substrate--by flip chip, wirebond or tab. Texas Instruments is also pressing its High Density Interconnect process, a joint project with General Electric, that inserts chips in predrilled holes in the substrate and deposits interconnect circuity as the final step on top of the chips.