HDI Substrates Enable Future Electronics

When will the American Organic Substrate (Printed Circuit Board) Industry climb aboard the Microvia/Build-up Train?

By Harvey S. Miller, Contributing Editor

Conventional PC boards do not permit enough wiring density to route next-generation equipment servers, workstations or base stations, such as with their increased population of integrated circuits in chip-scale packages and ball grid arrays. The wiring density of conventional PC boards is limited by the following factors:

1. Subtractive etching below 100 mic-rons (4 mils) leads to extreme yield reductions.
2. Mechanically drilled through-holes become increasingly expensive below 250 microns (10 mils).
3. Through-holes of any size reduce the wiring-available area on all layers, and the dogleg needed to distance them from the component mounting pads (to prevent solder wicking) also reduces top-layer wiring area.

Substrate Combinations

1. Additive technology over a deposited resin surface. High precision permanent resists enable 50 micron and below lines and spaces.
2. Blind microvias (in-pad), 50 microns and less, are the key wiring-density enhancement which differentiates HDI substrates from conventional PC boards.
3. Other HDI technologies are outlined in Table 2 and described from the Japanese experience, below.

Japanese Dominate

But the Japanese companies have been putting it all together since 1994, transforming "PC boards" into "organic HDI substrates." How they are doing it is best illustrated with a case-by-case analysis of the major Japanese company approaches. A discussion of differences with the U.S. position will address the "why" of the lag and provide the basis for a perspective. Finally, we will offer suggestions to the North American companies.

Five major Japanese company papers delivered at the IMAPS '98 conference in San Diego, November 3-5, illuminated their HDI substrate constructions directed both to IC packages and to the boards that mount and interconnect them. Table 2 describes the elements of each.

• Hole formation laser "wins" 4 to 1, Pulsed and CW. (Fujitsu, not included in our five, is moving from photo to laser via in its FLLD construction.)
• Build-up layer is coated or deposited in two examples, discrete laminate or film in three. (Fujitsu will even the score.)
• Circuit formation is additive in three, enabling 50 (and lower)-micron lines and spaces, subtractive in two.
• Via metalization uses conductive paste in three, plating in two.

Figure 1. This board, manufactured for a portable application, represents an interim design. Note the blind microvias (in-pad) and alongside the dogbone vias. (Courtesy HADCO)

1. In line with the pattern of Japanese industry, four of the five companies in our example are vertically integrated from manufacturing (including PC board substrate fabrication) to product marketing. This pattern is rare in North America, where the industry is almost entirely merchant. But this factor is not as critical as the next.
2. In Japan, corporate partnerships are honored, and there is no need to make Wall Street's next quarterly report expectations. That makes long-term investment easier to implement.
3. Finally, the small, digital-consumer products of the Japanese companies required HDI by 1995. American products, with larger boards in non-portable applications, are just beginning to require HDI for routing.

Figure 2 Closeup of a laser-drilled microvia: A CO2 laser was employed to sublimate epoxy and glass with a bounce pad. (Courtesy HADCO)

That was when the Japanese companies were beginning to use additive in an entirely different, much-higher- potential context, as an element of HDI. The embittered veterans of the "Additive Wars" should be sought out for their value to serious American HDI fabricators.

Laser buried vias were used by Don Seraphim to increase connectivity in IBM's 3081 System in 1981. The program lasted two years.

As 1999 begins, many U.S.-based PC board fabricators are acquiring laser drills. Using materials technology similar to Hitavia 3 (Table 2), but with resin-coated copper (RCC), they are beginning to fabricate HDI substrates. Hadco uses a YAG IR laser to ablate the copper, then a CO2 UV laser to remove dielectric down to the next copper surface— a common technique.

Ambitious Plans

1. Johnson Matthey ACI intends to be the largest North American supplier. By the end of 1999 it will have 10 laser drills„four YAG to ablate copper (for RCC processing) and six CO2 units to remove dielectric. It has a history of successful photovia prototyping that will help, and a cross-license with Kyocera for build-up technology.
2. AlliedSignal's ASTI has three captive materials available for HDI building blocks: a) resin-coated copper; b) Ultrastable, a premium thin copper clad, thin rigid material, both from the laminate sister division; and c) two-metal-layer polyimide film.

1. Seek strategic partnerships with equipment and materials companies wherever possible. Hard bargaining over the last 5% of price can be counter productive if it means losing a valuable supplier-partner.
2. Seek support from OEM customers who need HDI. Some of these are the companies that once cast off fabrication as a non-core competency. They must be wondering now whether HDI substrate fabrication is truly a non-core competency.
3. Consider licenses such as those offered by Matsushita for ALIVH or Dyconex for plasma.