A High Performance, Economical RF/Microwave Substrate

A High Performance, Economical RF/Microwave Substrate

Taconic, Advanced Dielectric Division
Petersburgh, NY

During the past decade, the wireless industry has evolved from a predominantly military-driven market to a cost-conscious, consumer-driven commercial market. At the same time, wireless applications are moving up in the frequency spectrum. For example, personal communications service (PCS) systems currently are designed for 1.9 GHz, but the newer design for wireless communications is moving toward 5.8 GHz.

As applications move up in frequency, renewed interest is being expressed in high performance, high frequency, low cost substrates. In many military applications, polytetrafluoroethylene (PTFE) substrates had been considered the materials of choice. With a very low loss tangent (0.0018 at 10 GHz), good resistance to processing chemicals, negligible water absorption and high temperature resistance (PTFE, popularly known as Teflon," has a melting point of 621°F), PTFE substrates were able to meet all the technical requirements of RF/wireless design.

When commercial applications demanded high performance substrates, certain disadvantages were associated with conventional PTFE substrates, such as their cost; typically the price ranged from eight- to 10-times the price of epoxy glass laminates (commonly called FR-4). In addition, due to the inert nature of PTFE, processing of the PTFE substrates requires an extra step: pretreating the hole walls using a plasma or chemical process (such as a sodium napthanate etch) to achieve good adhesion of the copper to the hole walls. Conventional PTFE substrates also are soft, which means that they require special care during processing. Finally, the coefficient of thermal expansion of conventional PTFE substrates is high (typically 180 to 205 ppm/°C, depending on the dielectric constant of the substrates).

To meet the requirements of the commercial arena, several proprietary and generic products have been introduced that offer some advantages over conventional FR-4 substrates. However, these products do not exceed or even match the superior electrical performance of PTFE or offer the cost advantages of FR-4.

The reason why these products fail in commercial applications is most of them are thermoset resin-based systems. (Thermoset resins form an infusible mass after crosslinking.) These resin systems are often processed in an organic solvent medium. Due to the chemical nature of most of these systems, they burn when exposed to an open flame. In order for these resin systems to be viable for PCB use, flame-retardant additives (bromine- or chlorine-based compounds) must be added. Hence, even though the electrical performance of the pure resin might be able to fill the need at high frequencies, the addition of a flame retardant to allow the resin to be used for PCB fabrication renders the PCB unusable at high frequencies. For example, polybutadiene has a loss tangent of 0.0015 at 10 GHz. However, this loss increases to 0.006 when a flame-retardant additive and other fillers are introduced.

Furthermore, some European countries (for example, the Scandinavian countries) are becoming extremely concerned about how their products and processes are impacting the environment, and these products usually are not environmentally friendly alternatives. Most thermoset-based systems have an unsaturated, exposed carbon atom, which leads to high moisture absorption and a greater tendency to absorb processing chemicals during the processing stage. This absorption tendency leads to two unwanted developments. First, an extra baking step must be added after the drilling step to eliminate blistering during the solder reflow process. (Hence, the advantage over PTFE substrates of eliminating the hole treatment process is lost.) Second, high absorption leads to a shift in dielectric constant (hence, a shift in resonant frequency), a higher loss tangent and a phase shift with frequency. This dielectric constant shift is an important factor in applications where PCBs are exposed to high humidity, such as a base station antenna.

Although these products are less expensive than conventional PTFE substrates, they are much more expensive than FR-4. Furthermore, a typical thermoset resin has a low viscosity past its melt temperature. Thus, when the laminates are manufactured under high temperature and pressure, the resin tends to flow. This phenomenon tends to produce a greater variation of dielectric constant and thickness within a sheet that most RF designs cannot handle. For example, a greater variation of thickness within a sheet can adversely affect the linearity of power amplifiers; a loose tolerance on dielectric constant will make it difficult to maintain a tight tolerance on the impedance of a trace width in a microstrip configuration.

As discussed previously, a thermoset resin system forms an infusible mass after crosslinking. Due to this characteristic, the peel strength of some of the latest high performance thermoset products is extremely low, making these products difficult to work with and expensive when rework is warranted (hence, increasing processing cost). Furthermore, due to the low peel strength, some of the newer products are not offered in 0.5 oz copper (17 mm thickness), making it impossible or very difficult to achieve fine lines and features. This situation is not much of a disadvantage for an RF design, but it is a major disadvantage for the newer high speed digital applications. In addition, since the processing of these products is different than traditional FR-4 substrates, a premium is charged for processing. Hence, the cost advantage of processing the high performance FR-4-like products is not realized. All of these issues have brought the high performance, low cost, high frequency substrate industry and its users back to the drawing board questioning whether a high performance, low cost substrate is a reality or still a myth.

A New PTFE-based Material

After an exhaustive engineering analysis, the conclusion was reached that PTFE-based products still offer the best electrical and mechanical properties. Building on all the advantages of PTFE and incorporating several proprietary technologies, a true, low cost RF/microwave substrate has been created.

The type RF-35 ceramic-filled, PTFE substrate is available for less than $8/square foot in quantities greater than 1200 square feet. RF-35 not only satisfies the price requirement, it exceeds every electrical and mechanical property that is sought in a PCB substrate for high frequency applications. The dielectric constant is 3.5 with a tight tolerance (±0.1). The material is offered in thicknesses in increments of 10 mil. (RF-35 currently is offered in thicknesses of 10, 20, 30 and 60 mil.) In addition, class 4 thickness tolerance is offered (that is, the tolerance on a 0.020" substrate is ±0.0015").

What makes RF-35 unique is the fact that the thickness and dielectric constant variations within a sheet are minimal. Work performed at the corporate engineering facility has shown that the standard deviation of the dielectric constant within a sheet is 0.01 and the standard deviation of the thickness within a sheet is 0.00023". The loss tangent Df of the RF-35 material was measured at 500 MHz to 11.2 GHz by an independent test facility. As shown in Figure 1 , virtually no variation exists in dielectric constant Dk across the frequency spectrum, and the material’s loss is 0.0018 at 1.9 GHz and 0.0025 at 10 GHz.

Water absorption is less than 0.02 percent when the material is exposed to water for 24 hours at 23°C. RF-35 is available in 0.5, 1 or 2 oz copper, and peel strength (after the RF-35 substrate is exposed to a solder float at 550°F for 10 s) is greater than eight pounds per inch for 0.5 oz copper.

To address the issue of the PCB substrate’s rigidity, a proprietary technology was incorporated to utilize the structural integrity of the woven-glass reinforcement. As a result, the substrate hardness was increased to 34 (as measured on the Rockwell hardness scale). In addition, RF-35 was engineered to have good dimensional stability. The substrate has a dimensional change of –40 ppm in the x-axis (fill direction) and 100 ppm in the y-axis (warp direction) after etching of the copper. These results translate into less finished substrate warpage and a minimal amount of stress on mounted components. To ensure the mechanical integrity of RF-35 and to dispel the myth that PTFE substrates exhibit creep, RF-35 was tested by an independent lab according to the BSI-125 procedure. The results are shown in Figure 2 . As the graph indicates, RF-35 exhibits negligible creep when exposed to a 4200 psi load up to 650°F (343°C), indicating that the substrate should not exhibit any creep in normal PCB applications.

A proprietary process has been developed that makes RF-35 a truly unique substrate. By modifying the surface chemistry of the ceramic filler, a true interpenetrating polymer network of the ceramic filler and PTFE has been achieved. This network ensures that the ceramic particles are encapsulated with PTFE, which provides RF-35 with the synergy effects of both PTFE and the ceramic filler. Hence, RF-35 has good dimensional stability and mechanical integrity, a tight tolerance on dielectric constant and thickness, and a high peel strength, and is the least-expensive high performance laminate available currently.

A comparison of the processing steps for RF-35 and a typical thermoset resin PCB material, listed in Table 1 , indicates that, contrary to popular belief, processing RF-35 is less complex than processing a thermoset resin-based PCB.

Conclusion

RF-35 offers a combination of good performance and low cost for high frequency applications. The substrate features the electrical advantages of PTFE and the mechanical advantages of the ceramic filler, negates all of the PTFE substrate disadvantages and disproves all the myths surrounding Teflon and its processing.

Taconic,
Advanced Dielectric Division,
Petersburgh, NY
(518) 658-3202.