Center for Science, Technology, and Economic Development (CSTED) > Selected Reports

The Role of NSF's Support of Engineering in Enabling Technological Innovation

II. REACTION INJECTION MOLDING

CHOOSING THE FOCUS OF THE CASE

As we began to examine the topic "advanced polymer matrix composites" more closely, we found it increasingly difficult to bound the technology satisfactorily in a way that would meet NSF's criteria for selecting innovations as well as prove feasible for study, given limited research resources. One problem was that polymer matrix composites were developed before World War II, so the basic inventions and discoveries long predate the formation of NSF. Another was that "advanced" has no clear operational meaning. Currently, "advanced" composites appear as automobile bumpers, dashboards, tennis rackets, golf club shafts, medical prostheses, and F-18 tail fins. What makes each one advanced is different. In some cases, "advanced" refers to a processing technology that made composites slightly cheaper than alternatives (e.g., reaction injection molding). In others, it refers to the particular combination of materials that yields a desired performance feature (e.g., in aerospace). Finally, polymer composites have relatively slow market penetration rates (usually upwards of 20 years). The slow penetration appears to be caused as much by market factors (unit cost, demand, investment required) as by technological roadblocks. Thus, we were unable to identify a single product that embodied the features of advanced composites yet satisfied NSF's selection criteria.

This led us to consider another approach entirely: to focus the case on the knowledge and technologies that permit application of predictive theory in the field of advanced matrix composites. We were prompted to consider this direction by a special issue of Scientific American (October 1986), which, together with summary articles appearing in that magazine in January 1962 and July 1973, offered a 25-year history of progress in composite materials. The articles appeared to reflect a shift over this period from observation and systematic experimentation to theory-driven experiments and applications. In the case study, we would examine how knowledge advances and new technologies applied to specific product areas in one or more of the following areas: aerospace, automobiles, and sports equipment. However, this approach also proved unpromising. As we explored this possibility informally with researchers in the field, we found that there was disagreement among them about the extent to which predictive theory was actually contributing to current applications of polymer matrix composites. No one doubted that the field was becoming increasingly theory based, but we found it difficult to argue that current, significant products (by NSF's criteria for cases in this project) were produced by using predictive theory.

Returning to our initial interviews with materials scientists and engineers at SRI, we realized that yet another choice was available: a process technology that was relatively recent and, through its use, polymer matrix composites had significant impact in particular markets. Reaction injection molding, including its more recent offshoots, reinforced reaction injection molding and structural reaction injection molding, was thus selected as an alternative to a product technology (or technologies) using advanced polymer matrix composites. By this tangled path, we ended up studying a process technology, which provides an opportunity for unexpected but useful comparisons with the two product technologies studied in the first year.

DEFINING REACTION INJECTION MOLDING

In 1979, Walter Becker of Mobay Chemical Corporation published the first major book on reaction injection molding (RIM). He observed that, even in the context of the rapidly changing plastics industry, the pace of RIM development had been so rapid as to be considered "remarkable" (Becker, 1979: v). Becker was reluctant to label RIM as revolutionary, but at the same time he was hard pressed to identify a technology from which RIM could be said to have evolved. For Becker, the most likely candidate was cast polyurethane elastomers.[7] This technology, used extensively during the 1960s, involved mixing the ingredients of polyurethane (typically a diisocyanate and a polyol[8]) at about 130C, adding a chain extender[9], pouring the mixture into a waxed metal mold heated to about 110C, letting the polymer cure for about 30 minutes, demolding, and then post-curing for 10 to 15 hours at 100C. During the 1960s, the ingredients (prepolymers) were made commercially available; process machinery was developed to meter, mix, and dispense the urethane ingredients semiautomatically; and demolding time was reduced to about 10 minutes (Becker, 1979: 5). By the early 1970s, cast polyurethane technology had matured; if the products (urethane elastomers) were to experience any significant increase in their applications, particularly in industries requiring high-speed production, the cycle time would have to be drastically reduced. RIM, first introduced in the mid-1960s by Mobay Chemical Corporation and its German parent, Bayer AG, was a response to this problem.

Reaction injection molding involves the high-speed mixing of two or more reactive chemicals, or prepolymers, just as they are being injected into a mold. The mixture flows into the mold at relatively low temperature, pressure, and viscosity. Curing occurs in the mold, again at relatively low temperature and pressure. The entire process, from mixing to demolding, typically takes less than a minute. Not only is it much faster than casting, it also requires much less energy. Different types of RIM are characterized by different ways in which reinforcement can be introduced into the final product. In reinforced reaction injection molding (RRIM), reinforcing materials such as chopped glass or glass flakes are introduced with one of the prepolymers; the reinforcing material and the prepolymers are then mixed and injected into the mold. Structural RIM (SRIM) refers to a process in which a fiber reinforcing mat is placed in the mold and the reacting polymers are injected through and around it. There are several processes related to RIM that differ in important ways (Slocum, 1990: 105-107):

• Injection molding of thermoplastics is superficially similar to RIM but differs in important ways. First, most thermoplastics are made in large-scale, continuous processes that are completely separate from molding operations. The bulk polymer must be heated to melt it before it is injected into a mold, where it must be cooled until it is sufficiently rigid to be removed. Both pressure and temperature requirements are much higher than in RIM, so equipment and energy costs are higher.
• In thermoset compression molding, the polymer is partially formed and sometimes mixed with a reinforcement such as chopped glass before the molding operation. There is no impingement mixing; the material is placed in an open mold and squeezed throughout the mold cavity as the mold is closed. Polymerization is activated by the heated mold. Energy and equipment costs are higher than for RIM.
• Resin transfer molding (RTM), like thermoset compression molding, is conceptually similar to SRIM. A monomeric material is injected into the mold, where it reacts to form the part. There is often a preformed structural reinforcing material such as a woven mat in the mold. RTM uses different chemical constituents than SRIM, allowing injection rates to be slower and mold pressure to be kept very low. This keeps tooling costs low, but the process is slow and appropriate only for low-volume production.

Although the focus of our case is a process technology, neither the technology itself nor the contribution that research has made to its evolution can be understood fully without including RIM reactants in the analysis. Stated differently, improvements in RIM were not independent of improvements in understanding the material that it processes. Our case therefore includes both the technological elements that comprise RIM and the related polymer chemistry.

Because it is a process, the elements of RIM technology can best be detailed by describing the separate processes involved[10]:

• Raw material storage
• Compounding intermediates
• Metering (one of two key elements in RIM)
• Mixing, including transition to laminar flow (the second key element)
• Molding
• Demolding
• Mold recovery
• Finishing
• Storage and shipment.

Compounding intermediates. In this step, the raw materials as delivered by suppliers are converted into intermediates ready to be used in the RIM process. This can be done in-house by the processing company or by the supplier before delivery.

Metering. The high-pressure metering unit is a key element of RIM. It must deliver the highly reactive starting ingredients to the mixhead with precise synchronization of two liquid streams, and within a few seconds inject them into the mold. Rapid opening and closing of the valves in the mixhead create backpressure transients that must be controlled, and the liquids must be mixed at very high velocities and pressures. Axial piston or radial piston pumps are the heart of the metering system. The pumps must operate in two modes: recirculation of the materials at low pressure, and delivery at high pressure for short bursts. Computers monitor and regulate the metering process.

Mixing. The other key element is the mixhead, which must deliver each stream at high velocity into the mixing chamber, accomplish this under precisely synchronized conditions, develop turbulence in the mix chamber so that the two streams are fully and rapidly mixed, and clean the chamber so there is no buildup. The mixing process also includes an aftermixer, which is placed between the mixhead and the mold. Its function is to convert turbulent flow to laminar flow to ensure even, bubble-free filling of the mold.

Molding. The mixed components, or intermediates, flow into the mold, gel within 5 to 10 seconds, and polymerize in place within a minute or less. Heat generated by the chemical reaction helps cure the material, but the mold must be heated so that the edges do not cool more rapidly than the core. The mold surfaces must be free of flaws, and the mold halves must fit tightly. Yet provisions must be made to allow air to escape as the mold fills. The mold is held in a mold carrier, which can be either moving or fixed.

Demolding. After a minute or so of curing, the product is firm enough to be removed or ejected from the mold.

Mold recovery. Once the mold has been opened, the product must release from the mold and the mold must be cleaned as necessary to remove flash and buildup of release agents. Release agents aid in getting the product free of the mold surface. They can be water based or solvent based, and can be applied to the surface of the mold between operations (usually by spraying) or incorporated into the material itself, migrating to the surface during the curing process. Subsequent finishing of the product, such as painting, generally determines the choice of release agents.

Finishing. These operations include trimming the flash, cleaning off release agents, post-curing (required only for some RIM urethanes), priming, and painting. [12]

Storage and shipment. Polyurethanes present no significant storage or shipment problems. They are nontoxic and not highly flammable.

RIM CHEMISTRY[13]

As we said earlier, in addition to these process elements, there is RIM chemistry, primarily the chemistry of polyurethanes. Indeed, RIM was developed to process urethanes, and most use of RIM has involved these materials. There are several types of urethanes for different applications:

• Polyurethane rigid structural foam is light and rigid, with a solid skin and a lower-density cellular core. It is used for business equipment housings, furniture, and building materials.
• Low-modulus elastomers are used for automobile fascias, bumper covers, and trim. Most are not pure polyurethanes but polyurethane/polyurea hybrids that have better processing characteristics than pure urethanes.
• High-modulus elastomers are used for large industrial and consumer parts requiring toughness and rigidity. They are made by using modifications to the chemistry of low-modulus elastomers.
• Integral skin foams have high-density skins and are used for steering wheels, arm rests, and protective covers that combine a tough surface and soft feel.
• Polyurea materials, composed entirely of ureas, have better high-temperature stability than urethanes, react extremely rapidly, and allow short cycle times in production.

Because urethane and urea react very quickly, give no by-products, and the reaction goes to a high degree of completion, they are well suited for RIM. Polymerization occurs in a stepwise fashion to form a linear chain. The metering of RIM machines can handle the balancing of components necessary to achieve complete urethane formation in seconds. But for the part to be ejected from the mold, it must also solidify quickly, and in the case of linear polymerization alone that would require that the mold be heated to complete the reaction, then cooled to solidify the polymer. This process is too slow and impractical for most RIM applications.

One solution to the solidification problem is to introduce crosslinking simultaneously with polymerization. If diisocyanate reacts with a triol, for example, branches will form, eventually creating a complete network. When the reaction is about 70% complete, some of the molecules are so large that they are limited only by the size of the mold; an insoluble gel begins to form.

The most common method of creating rigidity in RIM products is to create a block copolymer during polymerization. Block copolymers can be made from urethanes by adding a third reactant, the chain extender, a low-molecular-weight diol. The resultant block copolymer controls the rate at which viscosity increases during polymerization, and also affects the elastic modulus of the final polymer. Thus, most urethane-based RIM systems contain four components: an isocyanate (usually a derivative of MDI), an oligomer (usually based on polypropylene oxide), a chain extender (a wide variety are used, typically a glycol), and catalysts and other additives (e.g., fillers, blowing agents[14], surfactants, internal mold release agents, and pigments). The basic properties of RIM materials, flexibility and impact resistance combined with substantial rigidity, are the result of "microphase separation" of the combined materials into "hard" phases (the reaction of the glycol chain extender and the diisocyanate produces a crystalline material) and "soft" phases (the long, flexible chain of the polyol). These two are incompatible, so they separate into very small, two-phase regions. The attraction among the hard domains, between which the soft domains are interspersed, gives the polymer its integrity. [15]

A RIM CHRONOLOGY

Before we can identify which of these technological elements are intrinsic to RIM and which are supporting, we must present a brief chronology of events in RIM's development. Development is still occurring, as is evident from a view of the technology 10 years after Becker's book appeared:

The chemistry that was standard 10 years ago is almost completely replaced. The list of applications of the late 1970s has expanded severalfold with the advances in both chemical and processing technology. The chemistry that is standard today promises to be replaced in the coming years, perhaps more than once. (Slocum, 1990: 105)

The RIM process was developed in Germany by Bayer AG. The first application appeared publicly about 1966; RIM attracted considerable attention at the 1967 Duesseldorf Plastics Fair in Germany, where Bayer displayed an all-plastic car. [16] Becker regards this as the catalyst that started the "RIM reaction" (Becker, 1979: 1). Bayer had assembled the key elements of RIM: introduction of reactants directly into the mold without turbulence, even distribution throughout the mold cavity, and a smooth transition from high-pressure impingement mixing to laminar flow.

It is significant that impingement mixing, the heart of the RIM system, had already been invented by Breer, Weinbrenner, and Hoppe (presumably at Bayer), who applied for a patent in 1955 (Wirtz, 1979: 182). The use of impingement mixing to produce urethane foams was first reported by Harris (1969) in the German publication, Kunststoffe. Pahl and Schlüter described the first RIM equipment with a self-cleaning mixhead in 1971 in the same journal. First applications of RIM occurred in Germany for automobiles and furniture (Macosko, 1989: 2). A key patent for a RIM mixhead is held by the German firm Krauss-Maffei, patented in the United States in 1972 (Kuerleber and Pahl, 1972).

RIM development in the United States was stimulated substantially by the 1972 federal standards requiring that all cars sold in the United States be able to sustain a
5-mile-per-hour impact without damage. The adoption in 1974 of mandated "corporate average fuel economy" (or "CAFE") standards added weight reduction as an important design goal for automobile bumpers. The preferred solution was to install lightweight, flexible polyurethane fascias on the front and rear of cars, supported by a steel support structure mounted on shock absorbers. RIM turned out to be the most economical way to produce fascias. U.S. production using RIM began in 1974 at 2,000 tons and grew to 17,000 tons by 1978 and to 70,000 tons by 1987. Of the latter production, auto fascias accounted for more than 50,000 tons, and other auto parts amounted to an additional 10,000 tons. More than 95% of production used polyurethane or urea-urethane mixes (Macosko, 1989: 3-4). The market significance of RIM processing can be appreciated by placing these data in the larger context of 1988 U.S. shipments of all thermoplastic and thermoset composites. In that year, the transportation industry received 311,000 tons, representing 25% of the entire market for these materials (Newman, 1990: 10). However, in the 1990s automakers began to shift from polyurethane RIM (PU-RIM) to injection molded polyolefinic elastomers for bumpers. Global production of PU-RIM fell from 130,000 tons in 1989 to 110,000 tons in 1994. Machine sales in the mid-90s focused on RRIM and SRIM for door panels and use of PU-RIM for interior auto applications (Mapleston, 1995: 54).

Rapid cycling times for RIM were constrained by the problem of mold release, which initially required that molds be sprayed with a release agent, typically wax or industrial soap, after each "shot." The basis for the problem is that the finished polymer adheres to the mold surfaces. Internal mold releases have proven to be a major step forward because they reduce the mold spraying requirement to once every 20 to 100 shots (Macosko, 1989: 28). Internal mold release agents, such as zinc stearate in polyurethane and acidic siloxanes in polyureas, are introduced into the initial materials and migrate to the surface while in the mold. One basic process was patented in 1983 by researchers at Texaco (Dominguez, Rice, and Lloyd, 1983), although the first extended production demonstration was accomplished by Mobay and Dow Chemical in 1984 (Slocum, 1990: 110).

Reinforced reaction injection molding (RRIM) machines were introduced in the mid-1970s. All were based on conventional RIM machines, modified to handle fillers such as chopped or flaked glass. Among the technical challenges introduced by RRIM are settling of the fillers during storage and delivery over long distances to the mixhead, abrasion of the mixhead by fillers, and obtaining effective orientation of the fibers in the mold (Coates and Johnson, 1981). Increasingly, RRIM is shifting to polyureas because, for a given stiffness requirement under heating, reinforced polyureas require less filler than do polyurethanes[17] (Slocum, 1990: 138).

Structural RIM (SRIM), another approach to enhancing the structural characteristics of RIM-produced materials, began to be commercially applied in the late 1980s. Glass has been the primary fiber used in SRIM. Because of the long, entangled fibers of the mats used in SRIM, it is difficult to obtain glossy surfaces. As a result, SRIM is used for non-appearance parts. Since the mixed polymers must flow easily through the fibrous mats in SRIM, the viscosity must be low at the beginning of mold fill and remain so until the mold is full, then rise rapidly. Ashland Chemical has developed one system, patented in 1983, that accomplishes this by using ethylenically unsaturated esterols. [18] Dow has developed another system based on more traditional isocyanate chemistry (Slocum, 1990: 116), and Mobay patented a third approach using urethane chemistry. According to Macosko, a major contributor to research on RIM, the RIM bumper market is shrinking, but knowledge gained from research and experience with RIM is yielding spinoffs that contribute to composites technology. In particular, SRIM is emerging (SRI interview with Macosko, June 3, 1996).

Figure 1 shows the number of U.S. patents granted by year for both RIM and RRIM. It is one indication of the timing of each innovation's introduction and the rate of technological change from its introduction to the present. The peaking of interest in RIM around 1990 and the subsequent emphasis on its offshoot, RRIM, are apparent.

Another perspective on the evolution of RIM and its offshoots over time can be gained by observing papers presented at annual Composite Conferences of the Society of the Plastics Industry for the period 1970 to 1992. Using the keywords RIM, RRIM, and SRIM to search titles and keyword lists, the profile shown in Figure 2 emerged.

Finally, the profile of papers contained in the ENGINEERING database is a further indication of the level of interest and scholarly activity in RIM, RRIM, and SRIM from 1985 (the earliest entries to the database) until the present. The more recent emergence of interest in SRIM is apparent in Figure 3, as is its supplanting of RIM and RRIM as foci for engineering research. Note that RRIM and SRIM are included in the set of 665 articles identified using the keywords "reaction injection molding" together with "reinforced" and "structural."

TECHNOLOGICAL DECOMPOSITION

The chronology of RIM indicates that the innovation, as introduced by Bayer in the mid-1960s, combined existing polyurethane chemistry, elements of injection molding, and mixheads developed for making polyurethane foam to produce a revolutionary process innovation (to use Becker's characterization). This case is therefore not a matter of tracking the origins of "intrinsic" technologies, but rather of identifying the major improvements in the key elements of RIM technology, and who made those improvements, and tracking the origins of both the technologies and the contributors. Given that the contribution of basic RIM to its primary market began to decline about 1990, we will focus on the period from the mid-1970s to that time. SRIM, on the other hand, is still showing promise of increasing use as improvements in both chemistry and processing continue.

RIM Chemistry[19]

As we have seen, the basic chemical components of RIM materials when RIM was introduced consisted of a polyol, a diisocyanate, and a chain extender-the same components used for cast urethane systems. Early RIM systems based on this chemistry suffered from several problems: the hard segment of these RIM materials softened considerably between -30C and +65C, creating variations in properties with temperature; glycols were incompatible with the long-chain polymers, leading to mix quality problems; reaction times were so slow that demold times were 1 minute, and longer for large parts. The solution to these problems was amine-based RIM systems, introduced in the mid-1970s by Mobay and Bayer. [20] In these systems, the glycol chain extender was replaced by diamine. The product was still called a polyurethane, but it was actually a polyurethane-urea hybrid (Slocum, 1990: 108). Macosko (1989: 29) points out that the major directions in RIM are based on the basic chemical formulation: polyurethane, mixed urethane-urea, and all urea. The formulations actually used in production are proprietary.

Polyurea-based RIM systems were being developed during the 1980s by Texaco, Dow, and Mobay. In such systems, the bonds developed are more thermally stable than those of urethanes and can be formed without a catalyst. Unfortunately, they react so quickly that molds cannot be filled before the reactants' viscosity becomes too great. Thus most of the developmental efforts have been focused on slowing down the reaction (Slocum, 1990: 111). These improvements reduce demold and cycle times, improve surface appearance, and yield higher thermal stability, but increase material costs. As noted above, urea-based RIM is growing in significance relative to urethane RIM. To Macosko's knowledge, "no one in industry is doing research on urethane RIM now" (SRI interview with Macosko, June 3, 1996).

Since over half of all RIM materials are filled, the interaction between fillers such as glass fibers and various polymer constituents is an important aspect of RIM chemistry. (Recall that RRIM offers major advantages over RIM in reducing thermal expansion and improved mechanical properties such as modulus and impact strength.) Primary fillers for RRIM are hammer milled glass, glass flakes, mica, chopped glass, and wollastonite; these are usually added to the polyol (Macosko, 1989: 213-214). According to Slocum (1990: 113), much is still not understood about the interaction of polymer and reinforcement.

Several companies are producing composites created by placing a glass fiber mat in the mold before injecting the polymers. Macosko (1989: 220-221) has compared the properties and processing conditions of SRIM composites offered by Dow (isocyanurate), Mobay (urethane), Ashland (acrylamate), and Shell (epoxy). One key issue in SRIM is the bonding of epoxies and resins to the glass fibers; another is achieving a smooth finish. Thus wetting, permeability, and heat transfer have been the subject of recent research (Macosko, 1989: 225-232). The Center for Low-Cost, High-Speed Polymer Composites Processing at Michigan State University, an NSF State Industry/University Cooperative Research Center, is doing research on the flow of resin into complex fiber preforms. The Center's industrial partners are mostly materials supplier firms (SRI interview with Bogdan, July 11, 1996).

Macosko, a chemical engineering professor at the University of Minnesota, has been a major contributor to research on many aspects of RIM. Macosko recounts that it was Frank Critchfield of Union Carbide who first got him interested in the area. In the early 1970s, Macosko was working on how the rheology (flow properties) of reacting polymer systems depend on the state of gelation of the forming polymer. He chose polyurethane as a model system for such studies after reviewing the literature on condensation polymers. At the time, he was not very aware of industrial applications of polyurethanes. Macosko gave a presentation at a Gordon conference on polyurethanes and, as a result, was invited to give a seminar at Union Carbide. One outcome of this connection was that Union Carbide funded a program at the University of Minnesota to develop a better understanding of the fundamentals of the RIM process (SRI interviews with Macosko, June 3, 1996, and Critchfield, July 15, 1996).

Metering

Metering units (basically, pumps) must deliver the reactants at precisely controlled flow rates and pressures, as well as recirculate them when mixing is not occurring in between shots. Typically, for nonfilled RIM and SRIM, piston pumps are used because they are the lowest-cost alternative. For RRIM, lance pistons (essentially large syringes) are used (Slocum, 1990: 125). Most current machines are designed for RRIM and thus use lance pistons driven by hydraulic pumps.

Mixing

Macosko (1989) observed that "despite its central role to RIM and considerable research, impingement mixing is still not well understood" (p. 87). He lists the features common to all mixheads used in RIM:

• Recirculation of reactants near the mixing chamber to maintain temperature and dispersion and to permit rapid startup of mixing.
• Rapidly opening valves to control stoichiometry from startup of mixing.
• Inlet nozzles that accelerate reactant jets to high velocity.
• Chamber where streams impinge and initial mixing occurs.
• Some means of ensuring that the chamber is filled rapidly and proper impingement occurs from the beginning of the shot.
• Clean-out piston to push all reactants out of the mixing chamber (p. 88).

The most common mixhead design, patented by Bayer, enables mixing and circulation/cleaning to occur sequentially via the movement of a single piston.[21] Other designs have been introduced, but Macosko says that all of them were developed by trial and error (1989: 93). Macosko reports the results of several research projects that provide performance data for various mixhead configurations as well as alternative material formulations (pp. 95-98). Research indicates that the mixing mechanisms that must be controlled occur at the micro-level and consist of fluid mechanics at the point where the two reactant streams impinge, micromixing processes, and molecular diffusion to bring the components into chemical contact (Macosko, 1989: 106-107). According to Critchfield, formerly at Union Carbide,

Bayer made important contributions. It pioneered in developing the high-pressure, high-speed impingement mixing technology used in making rigid polyurethane foam molded products. CFC was the blowing agent. An example of the products made this way is doors for buildings. (SRI interview with Critchfield, July 15, 1996)

Critchfield continued by noting that important contributors to the development of the mixing technology and to the development of RIM machines were Krauss-Maffei (Munich), Cannon (Italy), Cincinnati Milacron, and Hennecke Machinery Company, a Bayer subsidiary.

Slocum (1990: 130-131) succinctly summarizes the parameters industrial RIM users employ when setting up a system:

• Flow rate of polyol is a simple function of the weight of the part, the weight ratio of polyol to isocyanate, and the injection time.
• Nozzle diameter to achieve a Reynolds number greater than 200 (to achieve adequate mixing) is a function of polyol density and viscosity.
• Viscosity required to achieve a given Reynolds number is a function of nozzle diameter, velocity, and density.
• Temperature of the polyol can be determined from the temperature-viscosity relationship of the polyol.

Although these can be calculated, according to Slocum they are usually set by trial and error for basic types of systems such as glass-reinforced polyurethane-urea.

Molding

The two main methods for measuring how material flows into and fills a mold are to monitor the heat generated by the reactants and to trace the pressure over time at various points in the mold. Temperature is a direct measure of the rate of reaction, whereas pressure changes across the mold are related to the material's viscosity, which in turn is a complex function of phase separation, the reaction rate, and gelation (Slocum, 1990: 132). Slocum, of Mobay Corporation, cites Macosko's research on temperature changes in insulated vessels. [22] Pressure measurements have been used to determine the viscosity of the material as a function of time after mixing, so that the effects of velocity, mold temperature, and formulation of the polymer on viscosity change can be studied. According to Slocum, "Such measurements have already proved to be of practical use in developing and optimizing systems, and in advancing our basic understanding of polymerization kinetics" (Slocum, 1990: 133).

Slocum reports that programs have been written that predict flow patterns and pressures in molds, and that these have been used successfully in RIM and RRIM despite their having been developed originally for injection molding (p. 137). They can also be used in SRIM if the permeability of the glass mats is taken into account. According to Slocum, the predictions can be very useful in avoiding air entrapment through proper mold design. Macosko (1989) devotes an entire chapter in his book to modeling of flows in molds, focusing separately on flow fronts under conditions of laminar flow, unstable flow, and reaction rates as measured by temperature changes over time. Flow modeling yields results indicating what the proper range of Reynolds numbers is to achieve laminar flow during mold filling (p. 117). If the reaction rate is too rapid and gelation occurs too soon, of course, the mold will not fill completely-thus the value of modeling reaction rates. The modeling problem is eased because the reactions occur so rapidly that little or no heat transfer through the mold walls occurs, so that adiabatic (perfectly insulated) conditions can be assumed. Working with Castro and Critchfield at Union Carbide, Macosko reported results of comparing predictions against experimental data using an industrial RIM formulation (Castro et al., 1980). These kinds of experiments have resulted in "moldability diagrams," which plot material temperature against filling time for conditions of poor mixing, air entrapment, thermal degradation, incomplete mold filling, and the parameters of the RIM machine. The result is a "window of moldability" for particular molds, formulations, and RIM machines (Macosko, 1989: 127-129).

An important byproduct of Macosko's work (and that of his students) has been software models that simulate the mold filling of reactive materials in SRIM applications. One example is the creation of AC Technology in Ithaca, NY, a firm that grew out of NSF's program at Cornell in injection molding. The software, however, was developed from the Minnesota RIM studies; AC Technology brought the software to Macosko's lab for testing (Macosko, personal communication, October 1996).

Demolding

Once the mold is filled, the part cures rapidly, becoming rigid enough to endure demolding without distortion. "During curing, modulus and strength must build quickly for rapid demolding but also, perhaps aided by a post-cure step, build optimum final use properties. It is this delicate balance between speed and performance which means the success or failure of a RIM product" (Macosko, 1989: 139). Because premature opening of the mold can allow the product to expand beyond the dimensions of the mold, prediction of curing times is important. As in the case of mold filling, modeling the process and checking against experimental data yield moldability diagrams that determine whether and when a part can be demolded (Macosko, 1989: 160-166).

Development of internal mold release agents for polyurethane systems was a major step in reducing cycle time. Most polyurethane and urea RIM systems use internal mold release agents; the most commonly used is zinc stearate together with a fatty acid in 1% to 2% concentration. Numerous formulations for mold release agents for use in RIM processes have been patented; the earliest was Bayer's, granted in the United States in 1973 (Boden, von Bonin, Kleimann, and Mergard, 1973). The problem of mold release plagued early RIM production, and component suppliers tried for some time without success, until Bayer succeeded. Losing that race to Bayer had significant consequences: Oldsmobile and GM Guide shifted from Union Carbide to Bayer as a materials supplier, contributing to Union Carbide's decision to end its involvement in materials for RIM in 1982-83 (SRI interview with Critchfield, August 31, 1996).