SMT - Flex Circuit Update

In the March 2008 SMT¹, an article on flex circuits highlighted liquid crystal polymer (LCP) films and the applications where they are preferred. This column offers comments on materials choice and market effects in flexible printed circuitry (FPC), in response to the March report. LCP films have outstanding dielectric properties: low dielectric constant, low losses, and extremely low moisture absorption. These characteristics make LCP the dielectric of choice for high-speed or high-frequency (above 10 GHz) applications. More of the signal gets where you want it to go.

Low moisture absorption is desirable in FPC because entrapped moisture can cause blistering and internal delamination if the circuitry is suddenly exposed to temperatures above 100°C, as happens, for example, during assembly operations. High moisture absorption is one of the major drawbacks of polyimides, the currently favored dielectric. Over 85% of FPC uses polyimides. Good assembly practice for polyimide-based
FPC dictates a bake-out for extended periods at temperatures above 100°C prior to solder assembly, with high temperature exposure shortly after bake-out because moisture uptake is rapid in polyimides. In the case of LCPs, good dielectric properties remain constant across a wide range of environmental conditions because of low moisture absorption. There's no significant change in the circuit performance with elevated humidity.

The March flex circuit article also hinted that LPCs are not widely used because of "circuit design and features." There's a more fundamental mechanism at work here and it's worthwhile to explain and understand how the FPC industry functions.

FPC is a custom manufactured product. There are no catalog items in FPC; every circuit is manufactured to a specific design for use in a specific device using custom tooling. In this sector, the FPC manufacturer has little control — other than design-for-manufacturability (DfM) input — over conductor pattern design, dielectric choice, or inspection criteria. Therefore, it is not a fault of the FPC industry that LCPs are not widely used. If and when circuit designers feel that LCP dielectrics are needed, and that better performance justifies the higher cost, the FPC industry will agreeably build circuitry with LCP films. If it's on the drawing, then it will be in the FPC. The limited penetration of LCP into the FPC market could be one of the industry's chicken-and-egg situations: LPCs are expensive because they aren't used in high volumes, but they won't be used in high volumes until they are less expensive.

Polyimides are near-perfect films for FPC applications; this is not surprising since they were developed for this purpose. Setting aside higher-than-desirable moisture absorption and lossly dielectric properties, they are extremely stiff and strong with decent dimensional stability, available in a wide range of thicknesses, and familiar because in use for more than 30 years in the sector. Additionally, they have excellent high-temperature properties and chemical resistance/stability, better-than-useable electrical properties, and an extremely smooth surface. No process step in any FPC factory threatens the integrity of a polyimide film and few FPC applications ever suffer polyimide failure.

Custom manufacture has other implications. Printed wiring boards (PWBs) are manufactured with the same chemical processes and equipment as FPC. However, shorter process sequences, stable dielectrics, and a cultural bias to find commonality result in PWBs being treated like a commodity. Quotes come to hand in days; delivery times are short and reliable; prices are estimated easily from a "pennies per square inch per layer" formula. FPC can involve three times the number of process steps, each with less than 100% yield, and uses polymer films which are not as dimensionally stable as glass-reinforced thermosets. They are harder to handle, transparent (meaning all those faults which would be hidden in a PWB are right out in the open), and manufactured under an anti-commonality attitude that states that "each design is different." If you manufacture nuts and bolts day in and day out to the same design using the same materials, you know your costs before you sell and can deliver from stock — no delivery problems and no problem maintaining a steady profit. If you build FPC, you never know the yield of a new design until you have run it through the factory. If you lose all of the first release because shrinkage after etch is more than expected and nothing will fit on the tooling, you must battle with all the other designs that are on the floor for make-up time. The result is a late, over-budget production. Custom manufacturing is an entirely different world and reluctance to embrace new materials is not surprising.

A major element of FPC cost is an overemphasis on specifications. FPC was developed for the military market; it provided the reliability and connection density that are critical in military products. This environment is always specifications-rich to assure that nothing changes from lot to lot; that everything is interchangeable; and that nobody gets killed because a manufacturer tried to increase his profit. If this raises product cost, so be it. One reason why Asian FPC manufacturers can deliver at lower cost is the elimination of superfluous documentation and inspection. In the U.S. market the question is, "Does it meet spec; do you have the paperwork?" In the Asian market the question is, "Does it work?" When I taught FPC workshops I was flabbergasted by a questioner who wanted to know why his FPC circuits varied in color from one production lot to the next. Turns out that the design did not specify the material thickness — one vendor used 0.001" polyimide and the next used 0.002" stuff and both were acceptable to the terms of his purchase contract. In my military-product experience this was unthinkable. My student had never considered materials specification as part of his design but it surely is. FPC is profoundly materials-sensitive.

Conclusion
The extreme importance of tiny details is nowhere more vividly realized than in the impact of FPC on product quality and cost. FPC forms the production backbone of every product where it is used — components are mounted directly onto it and nothing can proceed through the assembly process without the FPC. However, FPC cost is a tiny part of product cost, and grudgingly at that. It has taken almost 50 years for FPC to become the standard technique for automating interconnections in electronic manufacture. Rate of growth has increased in the past 10 or 15 years as the March SMT flex article says; this is a proof of the maturity of the technology and the arrival of at least a degree of common sense.

Thomas H. Stearns, president, Brander International Consultants, may be contacted at thomasstearns@comcast.net. Readers are encouraged to submit questions and thoughts on flexible circuits to Stearns for future columns. He previously designed and developed connectors, flexible printed circuits and rigid-flex production technology for InterFlex Corporation, Lockheed-Sanders Interconnect Products Division, and the Raytheon Company. He is author of Flexible Printed Circuitry, McGraw-Hill 1996, and teaches workshops in interconnections technology for the IPC, Uwisconsin Extension, and PCD&M.

Add RSS Feed