Production tooling, which traditionally has been fabricated by highly skilled tool and mold makers using computer-numerical-control (CNC) machining and electrical discharge machining, has always posed severe cost problems and slow turnaround times for manufacturers. If tooling is late or does not perform as intended, market introduction will be late--which often is the death knell for a new product.

The worldwide tooling market is estimated in the tens of billions of dollars a year. The market's size and the growing possibility of using RP-based technologies to cut consistently high tooling costs have led many large original equipment manufacturers, their suppliers, and national laboratories to initiate various research programs aimed at rapid-tooling development. Meanwhile, RP system vendors, SFF service bureaus, and various small development companies have been hard at work to accomplish the same task.

Sandia National Laboratories' laser-engineered net shaping directly fabricates rudimentary metal parts from CAD data by injecting metal powder into a pool of molten metal created by a focused laser beam

High on toolmakers' wish list are hard, wear-resistant tools and molds with complex geometries rendered to a high degree of dimensional accuracy--0.001 or 0.002 inch. Extremely good surface quality--about 20 microinches root mean square--is also key. Finish machining and polishing operations must be kept to a minimum, as they add time and cost.

"The fabrication of production-ready tooling is incredibly demanding," said Emanuel Sachs, who leads the Three Dimensional Printing Consortium at the Massachusetts Institute of Technology (MIT) in Cambridge. "Molders require superhigh accuracies; really smooth surface finishes; and very hard, dense materials." Other desirable features include plastic-injection-molding tools with integral conformal cooling lines that can speed processing cycle times and provide better part quality, as well as multicomposition and functionally gradient tools made of two or more different materials.

These efforts to develop practical rapid-tooling methods must bridge the large cultural divide between RP experts and toolmakers, a highly secretive and ultraconservative group of specialists who practice what many call a black art. "Most RP people are oriented toward product-development processes, while toolmakers have to address entirely different issues in trying to figure out how to make those products," said Anthony T. Anderson, technical specialist on rapid tools for the Ford Motor Co. Research Laboratory in Dearborn, Mich. "The problem is these groups don't talk together much, but to do rapid tooling they have to."

"The cultural elements of rapid tooling far exceed the technical issues," said Terry Feeley, president of Laser Fare Inc.'s Advanced Technology Group in Narragansett, R.I., who has been heading a rapid-tooling development project at the laser-material-processing company. "A close dialog is needed between the toolmakers, the RP people, and the end users. It's important, for instance, to get people to have appropriate expectations for a particular process," he said.

Also crucial, Feeley continued, is selecting the rapid-tooling technology appropriate to the task: "For example, you shouldn't build a tool capable of 1 million shots that only has to run off 100,000. And there needs to be the general recognition that a shape is not a tool." For success in this area, he said, "you need a shop-floor-driven development effort based on a partnership between the RP specialists and the toolmakers, because breakthroughs in rapid tooling require a hybridization of RP and tooling techniques."

Feeley spent the last three years leading a joint research project with toy maker Hasbro Inc. in Pawtucket, R.I., to develop a cluster of robust rapid-tooling technologies for plastic-injection-molding tooling.

The rapid-tooling technologies resulting from the research, two of which have come to light, will be commercialized by a new division of Laser Fare called ExpressTool Inc. in Warwick, R.I. "We have taken orders," he said, "but we're marketing this as a beta program so we have the chance to iron out any bugs."

One proprietary technology is fundamentally a powder-metal process that produces production tooling made of chromium carbide, a very hard cermet with hardness levels of 55 to 60 Rockwell C (RC), Feeley explained. The other technology, dubbed Hastool, yields marked improvements in tooling build times and molding cycle times, he said. Feeley confirmed that Hastool uses conformal cooling channels, and revealed future plans to institute thermal pulsing--preheating and postcooling the mold with hot and cold fluids to speed cycle times. He did not specify dimensional tolerances for the tooling.

One of the more-promising rapid-tooling technologies now available is the Keltool powder-metal-sintering process originally developed by 3M Corp. and recently purchased from Keltool Inc. by 3D Systems Inc. in Valencia, Calif. (see "Rapid Prototyping Is Coming Of Age," July 1995). The process is particularly useful for small tools, according to Paul Jacobs, 3D Systems' director of research and development.

Although much of the Keltool process remains proprietary, the procedure generally runs as follows. First, an RP master pattern is used to make a precision silicone room-temperature-vulcanizing (RTV) rubber mold. This temporary mold is then filled with finely powdered A6 tool steel (or stainless steel) and even finer particles of tungsten carbide. The powder sizes are specified to ensure tight packing. An epoxy binder is then added to the specially designed "bimodal" powder mix, which after curing forms a green part. Following demolding, the green part is fired in a hydrogen reduction furnace that removes the binder and sinters the powders together. The fused part, which has about 30-percent void space, is infiltrated with copper. Finally, the composite tool is rough machined and shipped.

The Keltool process can attain accuracies of ±0.0016 inch for a 10-inch part if all steps are performed correctly, Jacobs said. "While this is adequate for prototype tooling, we feel that true production-grade tooling will take two to three more years to perfect." Parts larger than 6 inches suffer from warpage problems, he noted, an issue that is also being addressed. In the meantime, 3D Systems is organizing a group of prominent tool and die makers to assist in the continuing development effort and to try out the results as they become available. "We're not good enough yet for production tooling," Jacobs said, "so we decided that we need to get the toolmakers on board early if we're going to make Keltool technology really work."

An RP-based process to produce sturdy prototype and short-run production tooling for plastic injection molding is being developed by CEMCOM Research Associates Inc., a small company in Baltimore. The procedure, still a year or so away from commercialization, has the potential to fabricate fully functional matched die sets more inexpensively and in one-half to one-third the time needed to machine conventional metal tooling. Turnaround times of six weeks or less are envisioned, along with lower costs.

The NCC Tooling System uses plastic RP models as master patterns for the fast fabrication of nickel-ceramic composite (NCC) tooling for intermediate-volume plastic-injection-molding runs--tens of thousands of shots. The secondary tooling method is based on plating nickel over plastic stereolithography patterns, then reinforcing the thin, hard nickel face with a stiff ceramic material.

The near-net-shape process is particularly suited to larger components (greater than 10 by 10 inches). This capability would fill a niche in the RP toolmaking field, which tends to be size-constrained because of limited RP build envelopes, said Sean Wise, CEMCOM's director of research and development. Dimensional accuracies for the molds are on a par with those of the original stereolithography patterns (±0.005 inch).

With development partners including 3D Systems, Pitney Bowes Inc., Eastman Kodak Co., and Ford, CEMCOM engineers have shown the NCC Tooling System to be a viable method to make high-pressure forming molds. They are now refining it to improve dimensional accuracy, ease of use, and process speed.

Several projects are now under way. To date, the process has been shown to produce a minimum of 5,000 injection-molded plastic parts. They include parts composed of unfilled resins and reinforced resins formed via injection molding, gas-assisted injection molding, and compression molding. A benchmarking mold has been demonstrated for Kodak, and an internal part for a mailing machine measuring 15 by 6.5 by 5 inches was built for Pitney Bowes.

This plastic-injection-molding core and cavity set indicates the fine feature detail capability of 3D Systems' Keltool rapid-toolmaking technology

"The CEMCOM process has the potential to be very accurate," said Douglas Van Putte, technical associate at Kodak in Rochester, N.Y. "So far, however, exact duplication of the stereolithography model has not been demonstrated."

The toolmaking process begins with a CAD representation of the desired mold, which is altered to include a 0.5-inch-thick separation insert at the mold parting plane, Wise explained. "Imagine a closed matched die tool," he said; "separate the mold halves by half an inch and build a model that is exactly like the void space between the mold halves." This modified mold design is then used to create a high-quality stereolithography model using an RP machine supplied by 3D Systems. The model is then coated with a conductive silver-based material and placed in a electroforming bath of nickel sulfamate where a thin nickel layer is plated over it.

The typical nickel plating thickness over the tool face varies from 0.04 to 0.20 inch. "Electroforming can replicate the geometry of the model very accurately," he said. The high-resolution nickel shell reproduces fine surface detail of the mandrel and provides mechanical integrity for the most highly stressed areas of the mold.

After electroforming, the stereolithography model continues to serve as a fixture, holding the nickel shell in place. Alignment and accurate dimensions are maintained during the chemically bonded ceramic (CBC) casting process by stabilizing the shell to the stereolithography model. Prior to casting, cooling lines are custom-fitted and fixed in place. Because the ceramic has almost zero shrinkage, casting can take place in the mold base pocket.

The resulting precision nickel-shell and stereolithography-model assembly is then attached to a standard pocketed steel mold frame using a high-strength CBC called COMTEK 66. This stiff backing material is a water-hardened, metal-filled, cement-based composition with low-shrinkage properties. The ceramic is then vacuum-cast through a small opening in the back of the frame. After the CBC cures for about a day, the opposing side is cast. Later, the two halves are separated, the model is removed, and the CBC is postcured. Once cured, the ejector pins are drilled and installed.

The ceramic transfers loads from the nickel shell to the steel frame. The difficulty with a very stiff backing material is that dimensional changes result in shear stress at the interface with other stiff materials, so the material must not shrink. "We push the chemistry to completion so the ceramic is stable," Wise said.

The resulting NCC mold has a high-tensile-strength, abrasion-resistant surface, and the high-compressive-strength backing provides support and mechanical coupling to the steel mold frame, which provides containment and alignment. The similar thermal-expansion characteristics of the nickel mold face, the stiff ceramic backing, the steel frame, and the net-shape forming characteristics of the nickel and the ceramic all help maintain an effective bond and precise location of tooling components.

"We feel that the CEMCOM technology is very promising for the fast production of tooling," said Vadan Nargasheth, business-unit fellow at Pitney Bowes Inc. in Shelton, Conn. Despite the promising results, Nargasheth believes it will take another year of development before the process will be ready for commercialization. Key development areas, he said, include improving the quality, accuracy, and finish of stereolithography models; figuring out how to integrate the different operations to get rid of bottlenecks in the process; and devising new ways to cool formed parts, given that the ceramic is a good insulator. "We're looking to embed metal heat sinks in the ceramic," Nargasheth said.

A recently announced technology grabbing significant attention is an RP-based powder-metal (PM) forging process for making steel inserts for molds and dies. Developed by Rapid Dynamics LLC, a tiny start-up company in Fresno, Calif., the PM forging method offers highly accurate finished tooling in days rather than weeks.

The process was invented and developed over the past seven years by Paul Vawter, an expert in RP-related casting processes. He provided a general description of the rather straightforward press-to-shape procedure.

First, an RP model of an object (or the object itself) is used as the pattern for the room-temperature casting of a negative (mirror) image in a special ceramic formulation with very low-shrinkage properties (less than 0.5 percent). After curing, the ceramic pattern and a quantity of very fine H13 or P20 steel powder is placed in an enclosure (die), where it is pressurized with a hydraulic ram while heated just to forging temperature. The applied heat and pressure consolidates the metal powder into a solid steel block (more than 97-percent theoretical maximum density) that is shaped like the RP pattern. The forging process has minimal distortion and shrinkage problems, he said.

The result is a tool steel insert with accuracies within 0.001 to 0.002 inch per inch, Vawter said. "What you see is pretty much what you get. The process is similar to powder forging or hot pressing," he explained. "The key is to use the lowest possible forging temperature so no molten metal is created (no phase change occurs), which typically produces distortion."

Vawter said he pressed his first insert just three months ago. Potential customers, including automakers and first-tier suppliers, are now evaluating the results. Rapid Dynamics' current prototype machine can produce inserts with a maximum dimension of 4 inches. A large prototype system capable of making 1-cubic-foot inserts is now in development.

An eight-jet printhead, developed by researchers at MIT's Three Dimensional Printing Consortium, sprays binders onto metal powders in a layerwise fashion to build green parts ready for infiltration with molten metals

Vawter noted that his PM forging process can make carbide face tools and conformal cooling channels. In addition, "you can weld and grind on it, and you can do in situ heat treatment." He added that he is working on capability to handle insert designs with undercuts.

Although the technology is still at an early stage, some of Vawter's results are striking. For example, the new process faithfully replicates the face of a dime as well as the tiny machining lines and hand-etched identification numbers on a production tool.

Three-dimensional printing, an RP technology invented by MIT mechanical engineering professor Emanuel Sachs and materials scientist Michael Cima, uses electrostatic ink jets to spray polymer binder onto powders, selectively hardening slices of a CAD-defined object layer by layer. Successive layers of powder are spread on top, and the process is repeated until an object is completed. Unbound powder temporarily supports unconnected portions of the component, permitting overhangs and undercuts to be fashioned. "It's basically a process for printing with materials. Instead of red and blue ink, you have, say, aluminum and nickel Ôink,' " Sachs explained. The print device, which is as big as a Coca-Cola machine, makes 6-inch-long parts in about 10 hours.

Process development has been supported by an industrial consortium of 11 companies including Boeing, Hasbro, Johnson & Johnson, the Pratt & Whitney Division of United Technologies, and 3M, with financial support from the National Science Foundation and the Defense Advanced Research Projects Agency.

The MIT engineering team's recent research focus has been on the rapid fabrication of production tooling for plastic injection molding. In a typical run, a stainless-steel powder is printed with a colloidal acrylic binder. After the resulting porous preform is sintered in a furnace, it is infiltrated with copper/tin to form a solid tool. Accuracies can be held to 0.002 inch per inch. "So far, we've used the core and cavity sets to mold prototype parts from demanding polymer systems such as polycarbonate, glass-filled nylon, and polystyrene," Sachs said.

"We can produce extremely complex parts with internal features that might otherwise not be possible," Sachs said. One of the 3-D printing technology's best tricks is its ability to incorporate conformal cooling channels into molds. Cooling lines are typically drilled into the mold at a later stage. "We're having good luck with conformal cooling lines, which allow you to run coolant below the surface of the mold. You can hold a more uniform mold temperature, which produces much less thermal distortion and less internal stress," he explained. "This is reflected in faster production cycle times (maximum 15-percent improvement), better part quality, or a little of both." He noted that the results have been corroborated by his industrial partners.

"We're also looking at rapid thermal-cycling techniques, where you heat the tool up with hot oil, shoot the plastic part, then run cold oil through the lines for a fast cool down," Sachs said. "It looks promising," he stated, adding that it would require tools that can also withstand high-cycle thermal fatigue and changes to injection-molding machines.

The MIT research group is also investigating the possibility of printing tools designed with stiff, lightweight truss structures, often having a tetrahedral shape. By incorporating voids in the tool design, its thermal inertia can be greatly reduced, which will cut cycle times.

Considering future development, Sachs listed several planned improvements to 3-D printing technology. "We're exploring process changes to attain better dimensional control--an order-of-magnitude improvement. Though the shrinkage is predictable, it must be further reduced." Tool hardness, he said, must also be increased. Current 420 stainless-steel/copper/tin tools have a hardness of 30RC. "We want to get to 50RC." He concluded that "the next horizon is direct printing of metal parts. With multiple printhead nozzles, we feel that the process should be scalable to make actual toolless production feasible."

Last fall, specialty tooling and machine maker Extrude-Hone Corp. in Irwin, Pa., purchased an exclusive license from the consortium to use 3-D printing to produce metal tooling and direct metal parts, according to Mike Rynerson, the company's director of rapid manufacturing. The company is designing and building a 3-D printing machine that makes metal tools. While he would not reveal when Extrude-Hone might enter the market, Rynerson did say that the initial plan is to operate as a rapid-tooling service bureau.

Research efforts at Sandia National Laboratories in Albuquerque, N.M., as well as Stanford University in Palo Alto, Calif., and elsewhere, are starting to make progress toward direct fabrication of metal parts (including tools). Sandia researchers have developed laser-engineered net shaping (LENS), a new technology that can make three-dimensional metallic components directly from CAD solid models.

The new process has the potential to change the way high-value metal parts are made, according to Clint Atwood, RP team leader. These components could include complex prototypes, tools and molds with conformal cooling channels, and small lots of production parts. LENS technology might also be used to fabricate functionally gradient components whose composition varies with location. In addition, it could help repair or retrofit existing parts.

The LENS process is based on injecting metal powder into a pool of molten metal created by a focused laser beam as the substrate below is slowly moved to trace out the geometry of the desired part. This scanning of the beam/powder interaction zone deposits consecutive layers in sequence, thereby building 3-D metal components.

This nickel-ceramic composite tool set, which was produced with CEMCOM rapid-tooling technology, was used to mold mailing-machine components from structural foam thermoplastics

Like many other rapid-prototyping techniques, LENS is an additive fabrication method, but it produces fully dense metal components directly from the final design materials, thus eliminating intermediate (secondary) processing steps. This feature enables LENS to offer dramatic reductions in the time and cost required to realize functional metal parts. As an example, the production of injection-molding dies for Kodak reportedly resulted in a savings of more than 50 percent in time and 70 percent in cost compared with conventional methods.

"We've made parts using a wide range of materials," said Michael J. Cieslak, manager of Sandia's Direct Fabrication Technologies Department. To date, parts have been built from 316 stainless steel, nickel-based superalloys such as Inconel 625, H13 tool steel, refractories such as tungsten, and titanium carbide cermets, according to the materials scientist.

"This process doesn't sinter the powders; it melts them completely which results in a fully dense material," Cieslak said. Microscopy studies show the LENS parts to be fully dense with no compositional degradation. Other tests reveal outstanding as-fabricated mechanical properties as well. For instance, the LENS-fabricated 316 stainless steel exhibited ultimate tensile strengths of 100,000 pounds per square inch and 30-percent elongation--values greater than the properties listed for annealed 316 stainless-steel grades. Samples of H13 tool steel were shown to have hardnesses of 59.3RC. Significantly, the fabricated structures also exhibit grain growth across the deposition layer boundaries.

Current dimensional accuracy in the x-y plane can be held to ±0.002 to ±0.003 inch, and part accuracy depends on the accuracy of the moving stages. As with most additive processes, z-axis accuracy is not as tight at ±0.010 to ±0.015 inch. Accuracy along the z-axis depends on various process parameters, Cieslak said, particularly the volume of the material being melted. In general, the size of melt pool--normally only 0.02 inch in diameter--determines build accuracy.

"The process is really akin to laser-welding or laser-cladding processes," Cieslak explained. The genesis of LENS occurred a couple of years ago with the start of a joint research project with Pratt & Whitney Aircraft Engine Co., which was interested in refining a laser-deposition system, developed in-house, that was designed for the surface repair of turbine blades.

Supported for the first three years with Department of Energy research funding and later with internal Sandia funding, the research team worked to develop "a similar system that could deposit powder in a finer, more controlled manner," Cieslak said. "Eventually, we realized we had a tool that could do more than just repair hardware. We could use it to build original hardware directly from 3-D CAD solid models."

Sandia engineers used low-power, solid-state continuous-wave lasers for precision control of the metal deposition. Two laser systems have been used to develop LENS at the national laboratory. The research platform uses a 1.8-kilowatt neodymium-yttrium-aluminum-garnet (Nd:YAG) laser, while the applications platform has a 500-watt Nd:YAG laser. "You could use higher-power lasers if you needed to lay down material at a higher rates," he said. "The process seems to be scalable."

The lasers are incorporated into a dry box system (a glove box) using a controlled atmosphere of inert argon gas with a low oxygen content. The enclosure also provides the opportunity to contain metal powder.

The beam is brought into the box through a window mounted on top and is directed to the deposition region using a 6-inch-focal-length plano-convex lens. The powder-delivery nozzle is designed to inject the powder stream directly into the focused laser beam. Sandia engineers developed an improved powder-delivery system in which powder is entrained in a stream of argon gas. The key in this part of the process is to deliver the powder accurately and continuously.

The patent-pending powder-delivery and nozzle system avoids the low-frequency fluctuations (or pulsing) of commercially available systems, according to Dave Keicher, former senior member of the Sandia technical staff. "Unsmooth flow tends to lead to unstable build up. The idea is to keep the melt pool small and stable." The lens and powder nozzle move as an integral unit.

The lab's applications-development team, led by Atwood, uses a three-axis CNC positioning system that moves the substrate forward, back, right, left, up, and down. The accuracy of the stage motion is also key to controlling deposition accuracy.

The laser beam is focused onto a point on the solid substrate used as a base for building an object. The substrate is generally the same material as the deposition powder. Powder particles are injected into the molten puddle to build up each layer. The substrate is moved beneath the laser beam, which causes the deposition of a thin metal line that creates the geometry for each layer. After each layer is deposited, the nozzle and lens assembly is moved incrementally away from the substrate a distance equal to the layer thickness to maintain a constant focal position. The powder is delivered to the deposition region via a carrier gas, and the powder volume is regulated by the powder-feed unit.

Computers control system parameters including laser power, x-y-z positioning stages, powder-feed rate, and argon flow. The system is run in a preset manner (based on accumulated operating experience), Cieslak said. There is no sensor feedback control as of yet.

Regarding software, the process currently functions similarly to other RP techniques. A faceted solid CAD model is generated, then sliced into a sequence of layers. The sliced file is input into another interpreter program (a preprocessor) that converts it into a series of tool-path patterns required to build a layer. The component is fabricated by first generating an outline of the key component features, then filling in using a rastering technique. The software file is used to drive the laser system to produce the desired component one layer at a time.

Now that the Sandia team has demonstrated the feasibility of the process, a precompetitive research consortium of 10 to 20 companies (including prospective machine vendors and end users) is forming a consortium to expedite the development and commercialization of process. A core group of companies, including Kodak, 3M, and Pratt & Whitney, will provide direction. Apportionment of final R&D results will be decided by consortium members at a later date.

One consortium member will be Optomec Design Co., a laser- and optical-systems integration firm in Albuquerque that applied for a license on the LENS technology. "We want to take the technology, commercialize it, and provide systems to industrial users," said Keicher, who is now a partner in the company. "It's time to get the process out into industry so we can get some feedback."

Cieslak listed needed improvements to the LENS process: The system has to be re-engineered to be fully self-contained; sensors need to be developed to allow real-time feedback; precision and accuracy need to be improved; and higher deposition rates must be demonstrated.

LENS technology still can produce only "extruded" part geometries in which the cross section does not change much along the vertical axis. Unless this capability is extended, the process could have limited applications. Cieslak envisions LENS technology serving a niche market for injection-molding tooling, as well as high-resolution, fine-featured components fabricated out of expensive materials. "It's a natural for retrofitting and repairing existing hardware, and for building monolithic parts from different materials." Presumably these multicomposition and functionally gradient parts could be fabricated using multiple hoppers of powder under computerized feed control.

home | features | weekly news | marketplace | departments | about ME | back issues | ASME | site search

© 1997 by The American Society of Mechanical Engineers