Gelcasting Enters the Fast Lane
High manufacturing costs, reliability and reproducibility have always been major barriers to the commercialization of advanced ceramics. The ideal solution would be a near-net-shape process that can be scaled up to high volumes and produce high quality parts consistently. Among the various conventional and novel methods developed to meet these needs, one dubbed gelcasting holds the potential to become the preferred choice for numerous applications.
Although first developed in the mid-1980s by Oak Ridge National Laboratory (ORNL) researchers, gelcasting didn't attract much industry attention until the early 1990s, when AlliedSignal Ceramic Components of Torrance, Calif. (now Honeywell Ceramic Components [HCC]) formed a cooperative research agreement with ORNL. With funding support from the Department of Commerce's Advanced Technology Program, HCC licensed the technology from ORNL in 1994 for manufacturing turbine engine components. A year later, ORNL received an R&D 100 Award from R&D magazine for the gelcasting technology.
The company has made significant progress in scaling up the process with government support. Silicon nitride nozzles, wheels, seals and thin-walled combustors have been produced with comparable mechanical properties to slipcast parts. An example is near-net-shape turbine wheels for air turbine starters used in commercial/military propulsion turbine engines. The wheels have been successfully gelcast, proof-tested to over 100,000 rpm and are currently in engine testing (see Figure 1).
A Versatile ProcessGelcasting is a relatively simple near-net-shape process that is similar to traditional forming methods (see Table 1, p. 28). A slurry of ceramic powder in a monomer solution is poured into a mold and then gelled in situ to the shape of the mold at temperatures between 40 and 80·C. After the part is removed from the mold, it is dried and sintered. Binder burnout is similar to pressed parts (0.5 to 1·C per minute up to 650·C for silicon nitride and alumina). Either a water-based or organic solvent can be used, though an aqueous system is preferred for environmental reasons. Only a relatively small amount of monomer, a gel initiator and in some cases, a crosslinker and catalyst, are required (typically 3 to 4 wt %). Figure 2 (p. 29) shows a schematic of the process.
This process is applicable to a wide range of ceramic powders, including alumina, silicon nitride, silicon carbide, boron carbide, ferrites, aluminum titanate, zirconia, sialon, and sodium zirconium phosphate. Gelcasting can also be used to produce complex-shaped ceramic composites. Since gelcast green parts have high strengths (3 to 4 Mpa or 500 to 600 psi), parts are easily machined, avoiding expensive post-sintering grinding. For turbine engine components under development at HCC, gelation times range from 10 to about 60 minutes, drying times range from six to 60 hours, and sintering cycles are similar to slipcast silicon nitride.
Compared to injection molding, which requires as much as 20% binder, gelcasting only uses up to 3 to 4%. In injection molding, burning out the binder can take up to a week, compared to less than a day for a gelcast ceramic. Since binder burnout is also much more difficult in injection molded parts, defects can be a major problem. Such defects can be avoided in gelcast ceramics if they are properly dried.
According to HCC, gelcasting also has a number of advantages over slipcasting. It is 50 to 80% faster, produces much more uniform powder packing, results in a much higher green part strength, and allows simpler molds (no dewatering of surfaces is required). These benefits also make gelcasting potentially easier to automate.
In addition, recent success with the performance of gelcast structural silicon nitride parts in aircraft auxiliary power unit (APU) turbine engine applications (see Figure 3) has resulted in the demand for larger components. Gelcasting has demonstrated a unique ability to form large size, complex, and thick cross-section components. For example, HCC recently gelcast an approximate 16 in. diameter (green) silicon nitride turbine blisk.
"These capabilities are leading HCC to replace all slipcasting applications with gelcasting," says J.J. Nick, HCC's manufacturing manager.
In some applications, gelcasting could also replace isopressing and extrusion. While both technologies are more rapid than gelcasting for simple shaped parts such as tubes or rods, larger parts using the fine powders required for technical ceramics are much more expensive to manufacture using these methods. "For example, we use isopressing to make silicon nitride tubes under about 6 in. in diameter," says John Pollinger, HCC's technology manager. "But for larger diameter parts, the press size would be very large and expensive, especially when low volume quantities of parts are needed. For these parts we use gelcasting because of the relatively low cost for tooling-aluminum, PVC pipe, etc. The gelcast part strength assures ease of handling and gives the part added robustness during drying."
However, gelcasting has yet to overcome some technical challenges. "The gelcasting process has the limitation of lower achievable powder packing densities when using submicron starting powders due to the need to have a fluid slurry for mold filling," says Nick. "The lower packing densities result in more shrinkage during sintering and can make densification more difficult."
Process ImprovementsSince the solids loading becomes the green density of the cast part (no solvent is lost), a higher solids loading is desired (50 to 60 vol %) as long as the slip viscosity remains below 500 mPa·s. Under these conditions, a fluid slip is produced that is easy to degas. A higher green density also helps to minimize defects. To increase the packing density of the powder, ball milling may be required, and dispersents-such as polyethyleneglycol silane, for instance-may be added at low levels.
The original gelcasting system used an aqueous solution of two monomers-methacrylamide for forming the linear polymer and methylene bisacrylamide for crosslinking, thus forming a water-polymer gel. Though this system has been used extensively to produce silicon nitride turbine components and a variety of other technical ceramics, a new system has been developed that uses only one monofunctional monomer, hydroxymethlacrylamide (HMAM) and a special initiator. HMAM is available commercially, produces lower viscosity slips at comparable solids loading, and has superior mold release characteristics.
Another advantage of HMAM is that the slurry shelf life is over eight hours before initiation, compared to three hours for the conventional system. However, the increased solids loading and the different chemistry require a different binder burnout cycle. Once the binder burnout cycle is optimized, sintered parts >99% of theoretical density have been obtained.
Other work at HCC and ORNL has developed a simple one-dimensional physical model of the drying process. The model predicts the instantaneous moisture content of a gelcast part with an accuracy of better than 10%, when the dryer humidity, dryer temperature, and sample thickness are specified over a broad range of these parameters. The model has been used to develop optimized drying methods that have reduced drying cycles for turbine wheels by 60 to 80%. Further research is needed to develop more complex models.
An alternative to in situ polymerization of monomeric systems is a gelcasting system under development at the University of Illinois (Urbana-Champaign), which is based on an aqueous alumina-polyvinyl alcohol suspension crosslinked by an organotitanate coupling agent. The gel strength of components made from this system can be tailored by modifying the properties of the polymeric gel or its filler content.
Automating the ProcessHCC has already demonstrated that gelcasting is a production-viable process capable of producing the required quantities. However, scale-up and optimization of slip preparation and gel development is critical. A gel initiator system has been developed that improves the gelcasting process. The slurry must have low viscosity, high specific gravity and produce strong, rigid gel bodies. Under these conditions, prototypes of a stator, turbojet wheel and starter wheel have successfully passed proof or engine tests.
But can gelcasting be used to reduce production costs? While the tooling can be less expensive in some applications, the effect of gelcasting on the total cost per piece is generally small compared to slipcasting. Materials preparation costs usually contribute the least to total production costs. The major cost drivers in any facility are direct labor, primary materials and equipment. However, technical cost modeling of gelcast silicon nitride stators and rotors shows that automation always leads to a reduction in costs. Therefore, HCC is now focusing on developing automated gelcasting with support from DARPA/
The automated process is being developed for a turbine wheel representative of a component for potential use in small turbojets. The example part is shown in Figure 4. In 1999, HCC designed and built an automated gelcasting system (Figure 5) and demonstrated the capability of forming cycle times of approximately 15 minutes per wheel, meeting project goals of a production volume capability of 10,000 wheels per year for a single forming machine.
All steps except for removing the part from the forming machine and transferring it into the drying chamber are automated. The system incorporates a modular design with a conveyor system to transport the molds between individual specialized stations. The modular design allows stations to be easily added for increased capacity. Another advantage is that each station can be programmed to function as an individual unit independent of the other stations, so that if one station experiences difficulties, production levels can be maintained at the other stations.
The closed loop automated system is composed of eight process stations and an upper and lower conveyor system connected by two lift stations. The mold is attached to a pallet, and is transferred via the conveyor to each process station. There are eight stations: mold release, mold assembly, mold fill, mold heating, mold cooling, mold disassembly, part removal and mold washing/drying. Each station is equipped with integrated programmable logic controllers. Machine functions are performed by either pneumatic or electrical devices. To heat the mold (which reduces gelation time), an infrared heater was chosen since it provides excellent temperature control, can be scaled to each mold size, has long life and provides high output.
The remaining technical issues being tackled, according to Nick, include eliminating surface bubbles by improving the mold filling process, and improving sealing of the complex molds to prevent gel slurry leakage during filling and gelation. This leakage results in flash and difficulty in mold opening and cleanup. Sealing improvement approaches include soft seals between mold interfaces, modifying mold shut off features, and increasing silicon nitride slurry solids contents while maintaining low viscosity.
Despite the progress in automating gelcasting processes, commercial applications requiring large volumes of silicon nitride turbine wheels are still immature, according to Pollinger. However, he is optimistic that commercialization will occur in the near future. Although interest in automotive ceramic turbocharger wheels and hybrid vehicle turbogenerators with ceramic hot-sections has dwindled, interest is surging in advanced microturbines for distributed power generation.
"The Department of Energy initiated a program in late 1999 to develop advanced microturbines for distributed power generation, including ceramic hot section components, in order to meet emissions and efficiency goals," says Pollinger. "Manufacturing viable processes for ceramic components is a key focus of this program." In addition, the Department of Defense continues to explore small advanced jet engines for advanced missile and UCAV applications, with ceramics being considered to improve thrust and range.
Expanding ApplicationsORNL is currently involved in a number of collaborations and cooperative research and development agreements (CRADAs) with various industrial partners. Most of the work involves using gelcasting to develop powders or parts for automotive applications. For instance, LoTEC of Salt Lake City, Utah, has a license to use gelcasting to make automotive parts from sodium zirconium phosphate.
Gelcasting is also being investigated for electronic materials. Researchers at ORNL have shown that gelcast lead zirconate titanate (PZT) components (see Table 2) made using water-based slurries have comparable properties to those made by die pressing and are superior to tapecast parts. Binder burnout of PZT gelcast parts also occurs at the same rate as die pressed parts. Chinese researchers have used gelcasting to make porous perovskite ceramics and complex alumina parts.
Other non-automotive applications are also being developed. The ORNL researchers are working on a CRADA with Ceramic Magnetics, Inc., of Fairfield, N.J. The goal is to use gelcasting to fabricate circular magnets-more than 50 cm, or 20 in., in diameter from ferrite-for a high-energy physics accelerator.
Under another CRADA, Norton Inc., Northboro, Mass., is collaborating with the ORNL group to produce complex-shaped silicon carbide parts for semiconductor wafer processing. Yet another application under investigation is the use of gelcasting to manufacture artificial bone. ORNL is also involved in two CRADAs with Advanced Ceramics Research, Inc. of Tucson, Ariz., to investigate rapid prototyping using gelcasting.
The potential applications are virtually endless. Over the past decade, gelcasting has emerged as a viable production technology for high-performance ceramics. Future improvements are certain to provide the momentum required to bring gelcasting into the forefront of forming technology.