The Facility for Applied Manufacturing Enterprise

San Diego State University was contracted to fabricate and evaluate the processing of a group of low-loss Cyanate-Ester and Epoxy resin screening plaques. These materials it was hoped, once mixed with a ferroelectric ceramic filler, would produce samples with a dielectric constant of around 15-20 and a low tan. Measurement of loss tangent and dielectric constant by TRW will determine if fabrication of additional specimens is warranted. A "process lessons learned" record was requested by the sponsor to facilitate transition of technology to TRW.

An epoxy resin was included in this study as a baseline matrix material because previously measured dielectric data exists within TRW and because epoxies have the potential for process flexibility and good compatibility with film bonding tapes. Cyanate-Ester resin was selected as a second candidate by TRW through experiences with this and other systems in previous development efforts. Cyanate-Ester was found previously to exhibit lower dielectric loss (due ostensibly to lower moisture absorption) when compared with Epoxies and ease of processing when compared to another potential candidate: flouropolymers. Figure 1 depicts several model predictions for the bulk dielectric constant of epoxy plaques filled with the CG ferroelectric ceramic powder. A volume fraction of around 63% filler is needed to reach a dielectric constant of 20 in the more conservative Maxwell model.

The original screening test matrix (see Table 1) included one epoxy and two cyanate-ester systems at ceramic particle concentrations of 5, 15, and 25% by volume. The epoxy used was Masterbond product EP121CL 2-part, low-moisture formulation with =3.3 and =0.013@1MHz. The two Cyanate-Ester systems were Arocy XU 366 with =2.64 and =0.0004@1MHz and Arocy M-20; both were from Ciba. Arocy XU 366 and M-20 are single-part systems designed for use with a catalyst. The CG Ferroelectric ceramic powder is graded at 1 micron and has a reported =3.3 and =0.013@up to several GHz. Additional physical data for the materials is presented in Table 2.

Matrix	Plaques	Filler	Catalyst
EP121CL epoxy	1	None	None
EP121CL epoxy	1	Ceramic 1@5%	None
EP121CL epoxy	1	Ceramic 1@15%	None
EP121CL epoxy	1	Ceramic 1@25%	None
XU-Cyanate	1	None	None
XU-Cyanate	1	None	Recom. 2W%
XU-Cyanate	1	None	5W%
M-Cyanate	1	None	Recommended
XU-Cyanate	1	Ceramic 1@5%	TBD
XU-Cyanate	1	Ceramic 1@15%	TBD
XU-Cyanate	1	Ceramic 1@25%	TBD
M-Cyanate	1	Ceramic 1@5%	TBD
M-Cyanate	1	Ceramic 1@15%	TBD
M-Cyanate	1	Ceramic 1@25%	TBD

Symbol	Value	Units	Information (source)
_c	8	g/cc	density of ceramic (supplier)
_e	1.30	g/cc	density of epoxy (supplier)
_cy1	1.14	g/cc	density of first cyanate ester (supplier)
_cy2	1.13	g/cc	density of second cyanate ester (supplier)
v_t	81.9	cc	nominal volume of plaque (meas.)
t	0.305	cm	nominal plaque thickness (meas.)

SDSU's original plan was to produce 12"x12" plaques for later cutting into 6" squares. A flat-platen tool previously constructed from scrap material for another program was layered with titanium foil to provide a smooth finish for part molding. Mr. Rob Riddle at TRW suggested that plaques in the range 0.075" to 0.125" thickness offer acceptable mechanical strength. The 12"x12" plaque size proved a waste of material during the learning phases of the work and was abandoned in favor of the 6" x 6" plaque size produced from the depicted picture-frame insert.

All work would be made much easier with access to complete cure information on each resin product, but the suppliers are seldom responsive in this area. All processing was carried out in a well ventilated room. All personnel wore latex gloves. A custom processing apparatus was fabricated from spare lab equipment to permit control of the temperature and degree of mixing of reacting polymers while under vacuum. Vacuum is required to remove trapped air introduced with the fillers or later during relatively high-speed, convective mixing of the suspension. It consists of an evacuated glass bell jar, a fixtured aluminum base-plate for the jar with integral vacuum and electrical pass-throughs, a DC stirring motor, a shaft mounted mixing impeller, a speed controller, a regulated hot plate, a thermocouple reader and a vacuum pump.

The minimum viscosity of the suspension at process temperatures ultimately determined the suitability of any potential processing methods. Filled plaques required compression-assisted molding to complete the formation of flat plaques prior to gelation of the matrix. This is because particulate fillers greatly increase the apparent viscosity of a fluid. Casting, rather than compression-assisted molding was used for all the unfilled plaques. Because direct measurement of viscosity is seldom possible during processing, several observational trials were made to visually gage the decrease in viscosity associated with heating the polymers. Figure 2 depicts a visual "degree of gelation" graph for neat epoxy.

The degree of gelation observed was corroborated by motor controller current data provided by the motor controller readout. For a constant voltage, motor current is a reliable indirect indicator of viscosity. This data will not be reported because of its dependence on the exact apparatus used for processing. Notice the small meniscus. When the normally much deeper meniscus starts to decrease in depth for a constant motor voltage, gelation is immanent and the suspension should be removed at once and spooned into the plaque mold.

Processing of the XU366 resin was approached in much the same way as that for EP121CL. The most important difference between the two in processing is the much more rapid increase in viscosity observed for the cyanate systems near gelation. This presents a problem in determining the time to remove the suspension from the process apparatus. Several runs ended with solid polymer in the apparatus. Once a late stage increase in viscosity is noted for these systems, they must be removed immediately and spooned into the mold. Any delay is catastrophic.

The EP121CL epoxy plaques (denoted "E") that SDSU delivered to TRW are marked with E5, E10, E25, etc. Physical data for these plaques is found in Table 3. The number after "E" is the weight percentage of filler; the volume percentages are only 0.8%, 1.5% and 5.1%. Additional specimens at 35% and 50% by weight were fabricated and delivered to TRW. One specimen with 75% by weight, which is roughly 40% by volume, was discarded due to its lack of integrity. Flexural modulii were measured for many of the plaques.

Code	Matrix	Plaque Weight	Density	Filler Weight fraction	Filler Volume fraction
E	Epoxy 1	72	1.3	0.0%	0.0%
E5	Epoxy 1 @ 5%	107	1.35	4.8%	0.8%
E10	Epoxy 1 @ 10%	112	1.42	10.0%	1.8%
E25	Epoxy 1 @ 25%	139	1.64	25.0%	5.1%
E35	Epoxy 1 @ 35%	166.0	1.63	35.0%	18.6%
C	Cyanate 1 w/o catalyst	47	1.14	0.0%	0.0%
CC	Cyanate 1 w/ catalyst	95	1.14	0.0%	0.0%
C5C	Cyanate 1 w/ 5% cat.	96	1.19	5.0%	0.7%

Dielectric property measurement performed at TRW and relayed to SDSU by Fax in January of 1997 is presented in Figure 3. The unfilled plaque verifies the product data. The plaques with 5 and 10% catalyst appear lossy and their dielectric constant values do not correlate well with their filler concentrations.

A number of side issues such as filler-settling-induced concentration gradients, catalyst concentration v. dielectric properties, etc. were of notable addressed. Concentration gradients through the thickness may be a desirable way to reduce dielectric mismatch and associated plane reflection effects. Such a "functional gradient" material was produced accidentally. At low filler concentrations the effective suspension viscosity was too low (for either polymer) to retain the particles in suspension for the entire pre-gel phase of processing. An early plaque, made by casting, showed pronounced settling and stratification of the filler as evidenced by gross curling due to differential shrinkage of different strata. This method of producing a "functional gradient" produces too many side effects. Multiple layer casting of progressively denser layers may prove more feasible.

Catalyst concentration was varied to explore its influence over dielectric behavior. Since the recommended catalyst for the Arocy systems is a zinc additive, the loss tangent was expected to increase with increasing catalyst concentration. This is confirmed by the TRW generated data in Figure 4. The dielectric constant is seen to decrease for increasing catalyst concentration.

This work successfully demonstrated laboratory-scale methods for compression molding polymer suspensions of interest in signature reduction programs. Further refinements to process procedures are certainly attainable. Scale-up will be limited by equipment fabrication, polymer temperature control and batch size effects, and timing of the gel point. The ferroelectric filler of interest appears to fail to produce the expected bulk composite properties for which it was chosen. Catalyst concentration demonstrated a significant influence over the loss tangent of the XU366 system.