Powder injection molding breaks new ground.

New developments from on-going research in powder injection molding promise to overcome some of the many challenges in producing these parts. Powder injection molding, or PIM, is a multi-step, high-volume manufacturing process capable of producing parts with complex geometries from highly filled material (see PT, March '91, p. 65). The process uses fine metal or ceramic powders that are combined with as much as 40% by volume polymeric binder to form a feedstock for injection molding. After molding, most of the binder is removed from the "green" parts via pyrolysis, and the parts are then sintered into finished parts with densities 95%-98% that of solid metal.

Although the powder injection molding business is still quite small, with roughly 40 U.S. molders generating an estimated $140 million in sales in 1993, its future appears bright. Randall German, Brush Chair Professor in Materials at Penn State University, University Park, Pa., the site of an industry consortium on PIM, predicts market growth of 40% a year in the U.S. and Japan. A number of current research projects in PIM processing may help foster that growth.

BINDERS & DEBINDING SYSTEMS

One area of research activity is alternative debinding systems for multi-component binders. The Penn State consortium is investigating new immersion debinding systems for widely used wax/PP binders. The currently prevalent solvents used for this system--trichloroethane and heptane--each have practical problems. Heptane provides rapid debinding rates, but also attacks and softens the PP, which can lead to stresses that may cause cracks and ruptures in the part. Trichloroethane provides rapid debinding rates with negligible softening of PP. But it is an ozone depleter and is being phased out of use.

Newer citrus-based solvents that are less likely to pose environmental threats are being investigated. Another alternative is to use different binder systems. One candidate is food-grade PE/glycol, in which water is used as the solvent, according to Dr. Karl Hens, program director of the PIM consortium. PE/glycol has been shown to be acceptable for use with larger powders, typically above 10 microns; although he adds that it is less effective for very fine powders.

INSTRUMENTATION & CONTROLS

In cooperation with Engel Canada, Guelph, Ont., and Kistler Instrument Corp., Amherst, N.Y., the Penn State consortium is modifying an Engel press and controller in its lab for faster response to correct for feedstock inconsistencies. This involves optimizing control schemes for in-situ corrections to the molding conditions. Modifications involve calculating the cavity-pressure switchover point and packing-pressure profile for each shot based on viscosity measurements during the filling cycle of the same part. Feedstock viscosity and bulk modulus measurements are made during the filling stage by analyzing hydraulic, melt, and cavity pressure profiles before the entire part is filled.

To make in-situ corrections requires a powerful controller. To that end, the Penn State consortium is in line to receive a RISC-based controller card being developed by Engel, which Hens expects to help push tolerances to below 0.05%. Part of the drive toward more sophisticated controllers is due to the high thermal conductivity of PIM feedstocks, which promotes rapid cooling. This, in turn, necessitates fast injection and short fill times--often below 0.4 sec, says Hens. Fast injection, particularly on small parts, can push controllers to their limits. The use of hot runners for PIM, such as one system currently being developed by Mold-Masters Ltd., Georgetown, Ont., will put even more demands on controllers by further reducing the shot size, Hens adds.

Ultrasonic mold sensors are now being developed by the Penn State consortium to analyze material inconsistencies within the mold and the melt-front progression during the molding cycle. The ability to "see" the melt-front position and velocity at any given time permits changeover to holding pressure just before the mold is filled, avoiding flash. The sensor is also able to detect voids, whether internal or external, and short shots. Each of these capabilities has already been demonstrated by Penn State in its lab; another leg of the project that is still under development is the ability to flag defective parts for the robot.

A separate project is the development of a neural-network artificial-intelligence system that can "learn" the characteristics of the process and optimize molding conditions. One feature of neural networks in general is the ability to recognize patterns based on limited pieces of information--in the same way that a person can recognize a face from a few features. Hens predicts a need for such a system because the hundreds of parameters that are monitored by sophisticated controllers make it difficult for a human operator to optimize the process. Hens says commercial availability is still two to four years away. Initially, the neural-network system is expected to make suggestions on molding conditions; eventually it would be fully integrated into the machine controls to be able to optimize the process without human interference.

HOT RUNNERS FOR PIM

A hot-runner system that has been modified for PIM is under development by Mold-Masters and the Penn State consortium. A prototype of the system was shown molding a complex, thin-wall heat sink for an electronics application at the Engel exhibit during NPE in June. One problem encountered in PIM during injection, says Hens, is separation of the fluid from the solid portion of the feedstock, which he compares to a stream of water and stones. This powder segregation can result in non-uniform shrinkage if it occurs in the part, or clogging if it occurs in the runner system. Hot runners offer the possibility of eliminating these separations, Hens says.

Mold-Masters cites three basic benefits of its modified hot runner: First, by eliminating the sprue and runner, injection speed can be optimized for cavity filling. Second, the lower viscosity of the molten feedstock reduces tool wear. And third, hot runners eliminate regrind, which can adversely affect the properties of the final sintered product.

The hot-runner system used was a modified Dura Centi-Shot valve gate. The valve gate reportedly prevents freezing at the gate area; a copper-alloy insert prevents heat loss and a titanium exterior insulates the mold from the hot-runner system.

FLOW ANALYSIS FOR PIM

A 3-D computer mold-filling flow-analysis program for PIM has been developed by Concurrent Technologies Corp. (CTC), a nonprofit institution located in Johnstown, Pa. The software is applicable to both metal and ceramic powder molding, and takes into account several critical parameters: mold filling, heat transfer, freezing of the binder, and air pressure in the cavity during filling. It reportedly can determine the likelihood of weld lines, jetting, short shots, and entrapment of air in the mold. The program displays flow-front locations, velocities, temperature and pressure, shear rates, and air pressure in the mold cavity.

In addition, an expert system is being developed to assist designers with mold cavity design. Part of that program is the prediction of shrinkages that occur during debinding and sintering.

According to CTC's PIM project manager, PIM feedstocks have more complex rheologies than standard thermoplastics and require different software algorithms to predict flow in the mold. (On the other hand, the software developed for PIM feedstocks is also applicable to the simpler rheology of thermoplastics.) One possible hurdle to PIM mold filling analysis is the wide variety of proprietary feedstock formulations; over two-thirds of molders involved in PIM mix their own feedstocks, and there are no industry standards as yet. Testing to determine thermal and rheological properties of specific feedstock formulations is required to accurately predict the flow behavior.

As a research organization, CTC is not in the business of selling software, although CTC believes that the PIM program has commercial potential. CTC is exploring options for distributing the mold-filling simulation code, including potentially making it available to users via electronic networking.