Characterization of Non-Woven-Fiber Printed Wiring Boards

Project Number : C95-09

Point of Contact:

Dr. Michael Pecht CALCE EPSC Email: pecht@calce.umd.edu Phone: (301)-405-5323 Fax: (301)-314-9269

Objective

Background

Work Accomplished

Objective

To characterizes the mechanical and thermo-mechanical properties of nonwoven, randomly dispersed, short fiber laminates, and identify potential failure mechanisms which must be addressed in the design and utilization of printed circuit boards using nonwoven technology.

Background

A basic building block of many electronic systems is the printed wiring board (PWB), which is a composite of organic and inorganic materials (resins and fibers) forming a base material, or laminate, onto which conductive materials are bonded to the surface for interconnecting electrical components. PWBs can be single or double sided, or a multilayer construction of either rigid or flexible composite laminate materials [Pecht, 1991]. A typical organic PWB is composed of laminates comprised of woven or non-woven fabric embedded in a resin. In a non-woven randomly dispersed short fiber fabric, fibers are generally cut short and lie in a random pattern. In general, non-woven laminates have compositions which are more homogeneous, can be made smoother, and have more isotropic properties than woven laminates. These properties are all important for fine pitch surface mount applications, where thermal mismatch and coplanarity are key to the ease of manufacture and component attachment (solder joints, direct attach or flip-chip) reliability. In the non-woven laminate investigated in this study, the matrix material is polyimide and the fabric is composed of non-woven, short aramid fibers.

Work Accomplished

Fiber distribution examination indicated that the fibers are much longer than the distances between conductors common in today’s electronic equipment. Therefore, use of non-woven-fiber laminates may reduce, but certainly not eliminate certain board failures, such as electrical shorts caused by conductive filament formation (CFF), whereby copper ions migrate along the interfaces between reinforcing fibers. However, randomized fiber orientation decreases the likelihood of a fiber intersection between two PTHs, thereby decreasing or eliminating the number of pathways for copper ion migration

The laminates were measured to have a z-direction CTE (80 ppm/oC), which is higher than those typical of FR-4 (60 ppm/oC), BT (50 ppm/oC), and CE (40 ppm/oC). Although the in-plane CTE (14-15 ppm/oC) is lower than FR-4, it was found to be higher than that claimed by the laminate vendor (8-12 ppm/oC).

The five different experimental conditions disclosed that both temperature cycling and humidity cycling could cause internal damage. Damage from temperature cycling was observed as the high temperature of the cycle approached the glass transition temperature (Tg) . This was evident in the many large cracks caused during the high-temperature cycling (25 to 200oC), in which the highest temperature was only 60oC below the Tg of the board (260oC). These cracks were visible on the surface as well as in the cross-sectioned area. When low temperature cycling (-50 to 125oC) was used (135oC below Tg), there were cracks after cross-sectioning, but only minor surface degradation. In experiment #1 (25 to 85oC), there was no significant degradation since 85oC, the highest temperature of this experiment, was 145oC below the Tg of the board and the change in temperature was small (60oC).

The effects of holding temperature constant while varying RH are notably different. In experiment #4 (T = 85oC, RH = 25 to 85%), the surface shows some delamination due to hygroscopic stress, as well as some cracks in the laminates. Cross-sectioning revealed that these defects were mainly superficial and did not penetrate deeply below the surface. Though none of these boards failed electrically, cracks and delamination can introduce contamination, accelerate corrosion of metallization, and thus degrade the laminates’ electrical performance.