A Prime Approach for the Moduling of Conduit Ceramic Parts

Journal of Materials Processing Technology 133 (2003) 181–188 A PRIME approach for the moulding of conduit ceramic parts J.S. Ridgway*, J.B. Hull, C.R. Gentle Department of Mechanical and Manufacturing Engineering, The Nottingham Trent University, Burton St., NG1 4BU Nottingham, UK Abstract The processing of advanced materials such as refractory ceramics and metallic alloy powders has been investigated intensely over the past two decades. Ceramic injection moulding has therefore become a prime method for manufacturing complicated parts from a robust material. Typically, powder is dispersed within a thermoplastic carrier (or binder) before it is moulded at high temperatures and pressures. Further removal of the binder by thermal or solvent degradation methods yields a component that is suitable for sintering. Within industry, components have been manufactured with densities greater than 95% of theoretical. However, this processing route has its drawbacks. Debinding can take days due to slow heating rates and changes in viscosity of the polymeric carrier that can delay production and increase costs. A solution has been found by using a reactive binder that can polymerise within seconds and degrade back to a monomer within a fraction of conventional de-binding times. This technology, known as powder reaction injection moulding engineering (PRIME), has been developed using a cyanoacrylate binder that is commonly used as an adhesive, thus introducing difficulties when moulding. This paper describes the processing limitations of this binder and the method for moulding a conduit ceramic part. # 2002 Published by Elsevier Science B.V. Keywords: Powder moulding; Cyanoacrylate; Binder; Ceramic processing; Alumina 1. Introduction Modern applications of technical ceramics such as automotive components and bio-prosthetics [1–6] have forced development and intensive research into alternate forming methods. Products within modern society can include complex shapes for mechanical and aesthetic reasons, but more importantly, there is a drive for wear resistance, longevity, economy and tolerance to harsh, volatile environments. Such components can be developed using a highly successful approach known as powder injection or compression moulding [7–12], combining traditional polymer injection moulding techniques with powder processing technology. Typically, a ceramic or metallic powder is dispersed within a molten polymer binder (or carrier) before it is formed under high pressure into a pre-heated mould. Thermal degradation or solvent extraction methods are then employed to remove the polymeric binder prior to firing, whereas the ‘brown body’ is sintered to near net density. Whilst processing polymer melts, variables within the moulding cycle such as temperature and pressure have to be constantly measured, leading to a complex system with accurate control. A new technology developed by Hull * Corresponding author. Tel.: þ44-115-941-8418; fax: þ44-115-948-6166. E-mail address: j.ridgway@domme.ntu.ac.uk (J.S. Ridgway). 0924-0136/02/$ – see front matter # 2002 Published by Elsevier Science B.V. PII: S 0 9 2 4 - 0 1 3 6 ( 0 2 ) 0 0 2 2 9 - 7 et al. [13] simplifies the PIM process by using a reactive binder. Powder reaction injection moulding engineering (PRIME) is a method that relies on using a reactive monomer such as a cyanoacrylate to bind ceramic or metallic powders within the mould. It is well documented [14–17] that when polymerised, cyanoacrylate can de-bind by reverting to a monomer within minutes, thus significantly reducing processing time in comparison to viscous polymer melts used in PIM. Cyanoacrylate polymerisation reactions occur at room temperature thus simplifying the requirements for a moulding platform. However, problems can exist with this highly reactive binder when using conventional moulding apparatus. This paper demonstrates the use of PRIME to develop a conduit part, highlighting processing limitations and solutions. 2. Feedstock Feedstock is a name given to the combination of a powder and binder matrix for injection moulding. This matrix is created by carefully mixing the powder into a binder under a controlled atmosphere, typically a vacuum. A ceramic powder that can be mixed satisfactorily to cyanoacrylate is alumina (aluminium oxide). Alumina has inert properties and excellent wear resistance, allowing it to be used in many applications including automotive components and medical 182 J.S. Ridgway et al. / Journal of Materials Processing Technology 133 (2003) 181–188 Fig. 1. Working window for acidity levels of inhibited monomer. prosthetics. However, when alumina is mixed with cyanoacrylate a reaction occurs, instantly curing the binder. Therefore, this polymerisation reaction has to be delayed until the feedstock has been moulded. A strong acid used in quantities above 0.1% by volume can be dissolved into the monomer solution thus delaying a reaction. Para-toluene sulphonic acid is capable of preventing reactions and if used in greater quantities inhibition can be for hours [14], provided that a strong base initiator is not introduced to the monomer. Moreover, polymerisation can be initiated by using surface catalysts such as caffeine, pyridine, t-butylamine and free surface moisture. The inhibition level is very important when the feedstock is moulded. Premature polymerisation could ruin a moulding machine and take hours to clean and, therefore, a balance has to be made. This leads to the development of a workable window for mixing feedstock as shown in Fig. 1. The use of the window is also of value when another parameter is introduced to a feedstock, such as a new powder. The alkalinity of individual powder batches is not discrete, thus changing the inhibition characteristics of a feedstock. Therefore, if parameters such as powder type are adjusted, the acidity levels need to be empirically determined. A practical solution is to use a polymeric disposable barrel insert. Polypropylene tubing is relatively inexpensive and suited to the application, and hence a technique for moulding was developed as shown schematically in Fig. 2. Part of the powder reaction moulding process is to react the feedstock within the mould. It is thought that this can be achieved by one of the three methods. Firstly, surface initiation, secondly, a method of entering the catalyst into the mould, or finally, air within the mould may be sufficient to cause a reaction. Each of these methods will be investigated to find the most efficient approach. It is difficult to design a mould, because the rheological flow characteristics of the feedstock are unknown. Therefore, an empirical approach has to be employed. The basic requirements necessary in the mould design are as follows: (i) sufficient mechanical strength to withstand the moulding pressures; (ii) room temperature moulding; (iii) fixture of the moulding barrel to the mould inlet; (iv) a high quality surface finish; (v) releasing agent; (vi) material compatible with cyanoacrylate; (vii) inlet; (viii) vent(s); and (ix) release pins. 3. The moulding process Historically, cyanoacrylate was created as an adhesive, bonding a wide range of materials from metals to rubbers in seconds. When used as a binder it presents a problem associated with processing. Conventional PIM machines cannot be used for moulding because the binder will adhere to the barrel or screw surface. Therefore, it is feasible to develop a simple system for manufacturing low volume parts. A barrel for such a machine should be made from a material that cyanoacrylate will not adhere to such as polypropylene or polyethylene. However, a machine made from these materials would have little mechanical strength. Fig. 2. Moulding system. J.S. Ridgway et al. / Journal of Materials Processing Technology 133 (2003) 181–188 183 Fig. 3. PTFE mould for solid conduit valve (dimensions: mm). 4. Mould design Based on the above requirements, a polytetraflouroethylene (PTFE or Teflon) mould representing a solid model of a conduit was designed and manufactured. This design was used to test the mouldability of feedstock and provide useful information about curing. Fig. 3 is a diagram of the design (no release pins were implemented at this preliminary stage of design). The Teflon mould was pre-prepared for use by treating the inner surfaces with a silicon oil surfactant to aid release of the formed part. A feedstock consisting of 0.45 volume fraction powder and 7% acid (by volume of binder) was injected into the mould cavity using the arrangement shown in Fig. 2. The mould was left at room temperature for 7 days whilst the feedstock cured. On opening, the formed surface in contact with the mould displayed a close replication and exhibited satisfactory curing. However, inspection revealed that the cyanoacrylate had not fully polymerised at the centre of the part. Further investigation of the mould highlighted a problem with machining. The Teflon had warped whilst the mould was being manufactured, probably caused by high temperatures produced when cutting. Therefore, the inner surfaces did not seal correctly during moulding, producing a part with excess flash. This can be eradicated by either changing the material to Nylon-66 or bracing the mould halves before and after the machining process. Three methods of eliminating the problems encountered with moulding a ceramic part are: (a) provide a passage for a catalyst to enter the mould; (b) control the levels of acid inhibition using the working window method; (c) limit the parts to a small thickness. The development of a conduit part automatically meets two of the above recommendations. Firstly, a conduit could have a small thickness in its cross-section and secondly, a sacrificial core similar to that used in mould manufacture in the ‘lost wax’ process will create a passage for a catalysing agent. The ‘lost wax’ process involves thermally removing a wax insert after the feedstock has been formed around it. The wax Fig. 4. Aluminium mould half for wax inserts. fits inside the mould and its profile is transferred to the internal surface of the final part. Thus, it is essential that the wax be moulded to a high tolerance with an excellent surface finish. Removal of the insert was performed by suspending the sealed mould within a water bath, at a temperature above the melting point of the wax. After leaving the mould for a period, the wax melted out and evaporated exposing the inner surface of the moulded part. This then allowed water to enter the mould and subsequently polymerise any remaining uncured feedstock. To test the process, an aluminium mould was designed to create wax inserts with a diameter of 8 mm. This insert was fitted inside a Nylon-66 mould with an internal diameter of 19 mm, hence a hollow cylinder with a thickness of 5.5 mm was created. Fig. 4 is a CAD model of the aluminium mould for wax inserts and Fig. 5 details the arrangement of the wax insert inside the Nylon-66 mould. Fig. 5. Nylon-66 mould half with wax insert included. 184 J.S. Ridgway et al. / Journal of Materials Processing Technology 133 (2003) 181–188 Inserts were created by casting molten paraffin wax in a pre-heated mould, to eradicate distortion caused by rapid cooling. These inserts and the surface of the Nylon-66 mould were then treated with a silicon oil surfactant as a releasing agent. After the Nylon mould was clamped, a 0.45 volume fraction feedstock inhibited with 7% acid was injected via the standard method and left to cure for 7 days. Before opening the Nylon mould, it was suspended in a water bath at 90 8C for 2 h, subsequently the wax melted out. On opening, the part had adhered to the inner surface, but no flash was noticed at the joining seam. Slight tapping on a steel block released the part for further inspection. Pitting was noticeable on the part surface due to the difficulty releasing the part from the mould. However, the feedstock seemed to have completely polymerised on the internal surface. The external surface still showed signs of fluidity caused by lack of polymerisation even though the mould was left undisturbed for 7 days. The sample was dimensionally accurate to the mould showing no signs of shrinkage through polymerisation. Fig. 6 shows a photograph of the finished part. 5. Improved mould design Observations made with the Teflon/Nylon/wax technology highlighted problems that were overcome by changing the moulding process. A novel metallic alloy mould and core was designed to accommodate the form of a complex conduit part, offering many advantages over the basic mould design, as detailed below. (1) Any mould material can be used with a melting temperature lower than the de-binding point such as a low melting alloy. (2) Removal of the alloy mould and core by melting could allow the part to polymerise very quickly on the inner and outer surface, reducing processing time from 7 days to within hours. (3) The part can be easily removed after processing, as there is no surface contact with the mould. Fig. 6. Photograph of moulded cyanoacrylate/alumina. (4) An alloy material will not be damaged or collapse with the injection pressures associated with the process. (5) The use of a mould material such as an alloy eliminates the warping problem encountered with machining Nylon. (6) A new material would be welcomed, as Nylon is expensive and prone to wear. (7) No release pins are required. (8) No mould release agent is required. (9) The mould and core can be re-cycled. Fig. 7 is a schematic diagram of the mould arrangement, with the sacrificial core in situ. The inner core was typically melted out at ffi100 8C and the outer shell at ffi140 8C to maintain integrity of the moulding prior to polymerisation. If the temperature raised above the latter value then the cyanoacrylate would start to thermally de-bind within the mould cavity. Initially, wax was investigated as a material for the mould and insert because of its lower melting temperature. Five different waxes were compared for casting the shape of the outer mould. Table 1 details casting data and observations. The types of wax are detailed below. Candelilla carnauba paraffin investment casting hard natural wax, very brittle and cracks easily, hard natural wax not as brittle as candelilla, soft wax, mixture of polystyrene and paraffin wax. The wax inserts could not withstand the pressures of moulding and they cracked whilst feedstock was being injected. The observation highlights the fact that wax is an incorrect choice of material for the moulds even though the injection pressures are relatively low. Therefore, an alternative material such as a metallic alloy was investigated. Two common types of alloy have melting temperatures below the de-binding threshold of approximately 150 8C. These metals are alloys of tin/bismuth and tin/bismuth/lead, the latter having the lower melting temperature; indicating a Fig. 7. Moulding arrangement. J.S. Ridgway et al. / Journal of Materials Processing Technology 133 (2003) 181–188 185 Table 1 Wax casting information No. Wax type Melting temperature (8C) Casting temperature (8C) Mould temperature (8C) Observations 1 2 3 4 5 Carnauba Candelilla Paraffin 1 Paraffin 2 Investment casting 85 85 52–65 75–82 63–70 100 100 80 90 90 110 110 90 100 100 Little shrinkage, but cracks when cooling Very brittle and cracks when cooling Casts with little cracking but high shrinkage and very soft Casts with little cracking but high shrinkage, soft Casts easier than natural waxes but cracks easy, extreme care required on cooling rates, very tough wax and malleable, noticeable shrinkage Table 2 Metallic alloys casting information Alloy Component Melting temperature (8C) Casting temperature (8C) Mould temperature (8C) Observations Tin/lead/bismuth Tin/lead Insert Outer mould 104 138 120 150 85–90 150 Brittle but hard and dense Hard dense material suitable candidate for the insert. Table 2 lists casting temperatures and observations. 6. Results A feedstock consisting of 0.48 volume fraction alumina and 8% acid was used for moulding experiments. This level of inhibition is in the ‘safe zone’ section of the working window, reducing the risk of premature polymerisation. Wax and alloy inserts were used for experimentation to find the one that produces the highest surface quality of the conduit part. Table 3 details observations made from forming the feedstock around wax and alloy inserts. The moulding process was successful; each part was effectively removed from the mould without any external damage. However, the feedstock collapsed in all cases as the insert was melted out. Further investigation of the damaged parts indicated that the collapse is probably caused by the lack of support for the feedstock when the insert is removed. Hence, a method of Fig. 8. Moulded conduit (sample 1). accelerating the polymerisation reaction is required. Samples 2 and 3 were moulded with a surface initiator applied to the insert, however, it had no noticeable effect and other methods are required to polymerise the feedstock. Fig. 8 shows sample 1 as moulded. A novel design has been implemented that uses a wax skeleton to surround and support the feedstock whilst in the mould. Removal of this skeleton allows the passage of a Fig. 9. The new mould (a) showing the skeleton formation and (b) with an insert in position. 186 No. Volume fraction Acid (%) Insert type Mould type Insert melting temperature (8C) Mould melting temperature (8C) Observations Cracks; 1: bad, 10: good Surface qualitya; 1: bad, 10: good General 1 2 3 4 5 0.48 0.48 0.48 0.48 0.47 7 6 6 6.2 8 Paraffin wax Investment wax Sn/Bi/Pb Sn/Bi/Pb Sn/Bi/Pb Sn/Bi Sn/Bi Sn/Bi Sn/Bi Sn/Bi 95 80 110 110 110 155 155 150 155 150 8 9 3 8 N/a 6 9 6 5 2 6 0.48 7.5 Sn/Bi Sn/Bi/Pb N/a 110 N/a N/a 7 0.48 7.5 Sn/Bi Investment wax N/a N/a N/a N/a Surface pitting related to internal collapse Internal collapse near injection point Major internal collapse Internal collapse, exotherm noted whilst moulding Inversed mould whilst melted out core which caused serious collapse Tried to melt outer mould first; feedstock had not polymerised causing a termination Wax cracked whilst moulding even though supported by mould clamp a Surface quality represents pitting and smoothness. J.S. Ridgway et al. / Journal of Materials Processing Technology 133 (2003) 181–188 Table 3 Moulding observations No. Insert type Mould type Skeleton type Skeleton melting temperature (8C) Insert melting temperature (8C) Mould melting temperature (8C) Observations Cracks; 1: bad, 10: good Surface qualitya; 1: bad, 10: good General 8 Sn/Bi/Pb Sn/Bi Paraffin 85 120 155 10 8 9 Sn/Bi/Pb Sn/Bi Investment wax 80 120 155 8 9 10 Sn/Bi/Pb Sn/Bi Investment wax 80 120 155 10 9 Slight internal imperfections, paraffin wax compressed with moulding pressures indicated by raised lumps on outer surface Slight internal imperfections, investment casting wax caused little external surface imperfections Virtually no internal imperfections, although external pitting; noticed that a large amount of feedstock appeared through wax injection points whilst moulding a Surface quality represents pitting and smoothness. J.S. Ridgway et al. / Journal of Materials Processing Technology 133 (2003) 181–188 Table 4 Moulding observations with wax skeleton included 187 188 J.S. Ridgway et al. / Journal of Materials Processing Technology 133 (2003) 181–188 catalyst direct to the surface of the binder, causing polymerisation soon after contact. The feedstock then has extra support from the polymerised shell enabling the insert to be melted from the mould without dragging the binder and powder with it. Fig. 9 demonstrates the new design with the cavity for a wax skeleton. This method works very well and the problem of massive internal collapse has been resolved as noted in Table 4. However, there was still a slight imperfection near the injection and venting point of the mould, common to all the moulded conduits. This phenomenon is either connected to the venting position of the mould or an internal pressure build-up within the feedstock. The latter of these suggestions indicates that when the insert was removed a pressure relief caused slight collapse. From these observations, the wax skeleton provides either: (1) a passage for the catalyst to contact the surface of the part, polymerising the feedstock therefore supplying a structural support for the feedstock whilst the core is melted out; or (2) two extra points of venting which may relieve the internal pressure built-up in the feedstock whilst moulding. 7. Conclusions (1) Powder reaction moulding is a novel process that requires a re-design of conventional moulding apparatus. High pressures and temperature control are not required for a low viscous cyanoacrylate feedstock. Therefore, a simple press is used to apply low forces within the injection cycle. (2) The volatile nature of cyanoacrylate has forced the requirement of a polypropylene insert within the moulding barrel. This novel approach to moulding has enabled a simplistic method of cleaning the apparatus after moulding, by disposal of the insert. (3) Nylon and Teflon are suitable mould materials because cyanoacrylate will not adhere to the surfaces, however, these materials are expensive and difficult to machine. Therefore, a system has been developed utilising a metallic alloy mould that is melted away after processing, thus eradicating any problems associated with adhesion of the binder. Removal of the mould and insert allows the ingress of a catalysing agent such as water. (4) The utilisation of an alloy moulding platform has enabled the creation of complex conduit parts which can be debound within minutes. This novel system is relatively simple and inexpensive in comparison to conventional moulding equipment, which has high costs and complexity. References [1] C.R. Gentle, The use of ceramics in prosthetic heart valves, in: Proceedings of ImechE Conference on Heart Valve Engineering, December 4–5, 1986, pp. 31–33. [2] C.R. Gentle, G.D. Tansley, Development of a ceramic conduit valve prosthesis for corrective cardiovascular surgery, Biomaterials 16 (3) (1995) 245–249. [3] J.S. Ridgway, J.B. Hull, C.R. Gentle, Development of a novel binder system for manufacture of ceramic heart valve prostheses, in: Proceedings of the Conference AMME, 1997, Gliwice, Poland. [4] CHAD Research Laboratories pty, Sydney, Australia. [5] H. Juden, C.R. Gentle, D. Dowson, SEM examination of a porous alumina prosthetic heart valve, Biomaterials 4 (1983) 139–141. [6] A. Yamagami, et al., Porous alumina for free-standing implants. Part I. Implant design and in vivo animal studies, J. Prosth. Dent. 59 (6) (1988) 689–695. [7] R.M. German, Powder Injection Molding, Metal Powder Industries Federation, 1990. [8] M.Y. Anwar, Injection moulding of 316L stainless steel powder using novel binder system, Powd. Metall. 38 (2) (1995) 113–118. [9] K.F. Hens, R.M. German, Advanced processing of advanced materials via powder injection molding, Powder Injection Molding Consortium, Particulate Materials Centre, Penn State University, University Park, PA, 1993. [10] B.C. Mutsuddy, R.G. Ford, Ceramic Injection Moulding, Chapman & Hall, 1995. [11] P.J. Vervoort, R. Vetter, J. Duszczyk, Overview of powder injection molding, Adv. Perform. Mater. 3 (1996) 121–151. [12] P.J. Vervoort, et al., Overview of powder injection molding, Adv. Perform. Mater. (3) (1996) 121–151. [13] J.B. Hull, C. Birkenshaw, M. Buggy, Novel Binder/Carrier System: Powder Reaction Injection Moulding Engineering (PRIME), Patent No. 9615698.9 (1996). [14] C. Birkinshaw, M. Buggy, A. O’Neill, Reaction moulding of metal and ceramic powders, J. Chem. Tech. Biotechnol. 66 (1996) 19–24. [15] A. O’Neill, To evaluate the suitability of ethyl-cyanoacrylate as a binder for powder reaction injection moulding, B.Sc. Thesis, University of Limerick, Ireland, 1993. [16] J.S. Ridgway, J.B. Hull, C.R. Gentle, Development of a novel binder system for manufacture of ceramic heart valve prostheses, in: Proceedings of the Sixth International Scientific Conference on Achievements in Mechanical and Materials Engineering, AMME’97, Gliwice, Poland, November 28–30, 1997, pp. 165–168. [17] J.S. Ridgway, J.B. Hull, C.R. Gentle, Design and selection of an alumina-cyanoacrylate feedstock for porous conduit heart valve prosthesis, in: Proceedings of the Mechanics in Design ’98, The Nottingham Trent University, UK, July 6–9, 1998, pp. 816–826.