Development of the powder reaction moulding process

Research Article Received: 14 May 2008 Revised: 14 August 2008 Accepted: 22 August 2008 Published online in Wiley Interscience: 9 October 2008 (www.interscience.wiley.com) DOI 10.1002/jctb.2063 Development of the powder reaction moulding process Lei Zhao, Maurice N. Collins∗ and Colin Birkinshaw Abstract BACKGROUND: The powder reaction moulding process uses a reactive monomer as carrier and binder for the moulding of metal or ceramic powders. De-binding is achieved using thermal depolymerisation which is followed by sintering to give the finished component. Binder can be recovered for re-use. RESULTS: Moulding compounds, with various powder volume fractions, have been prepared using stainless steel, silicon nitride and alumina with n-butyl cyanoacrylate as binder, and the stability of the compounds established. Rheological properties of the compounds have been measured using both pressure flow and drag flow methods. Compounds are strongly pseudoplastic. Comparison of experimental results with theoretical models, describing suspension flow behaviour shows that experimental maximum volume fractions are close to the theoretical volume fraction of 0.42 for silicon nitride, 0.68 for alumina and 0.7 for stainless steel. Differential scanning calorimetry and thermogravimetry have been used to simulate de-binding and show a rapid loss of binder through depolymerisation. Post-sintering porosity of the ceramic materials is high but this is thought to arise from the low pressure moulding techniques used. Porosity of the stainless steel mouldings is much lower. CONCLUSIONS: The results validate the powder reaction moulding idea and demonstrate applicability to three widely different powder materials. c 2008 Society of Chemical Industry Keywords: powder forming; sintering; cyanoacrylate; moulding; compression forming INTRODUCTION 454 The powder reaction moulding process was proposed and described, about 10 years ago, as an advance in the technology of the powder processing route for the fabrication of components from hard metals and ceramics.1,2 Typically this method of manufacturing components relies upon compression moulding or injection moulding of a powder-binder composition to give a green compact which is then subjected to a thermal debinding and sintering process to form the component. Binders are usually waxes or low molecular weight polymers. In the powder reaction moulding process a reactive monomer is used as carrier for the powder and the green component is formed by moulding the powder–carrier compound accompanied by polymerisation of the monomer to give a tough compact. The monomer is then recovered by thermal decomposition of the binder prior to sintering of the component. Figure 1 illustrates the process cycle. Anticipated advantages of the process, relative to more conventional powder processing methods, are more rapid solidification of the binder to give a strong and tough green component, which can be machined if appropriate; very rapid de-binding as the decomposition product is a small monomer molecule which readily diffuses through the powder; and no net loss of binder material. In existing processes the de-binding stage can take several hours and requires thermal decomposition of the binder whereas the powder reaction moulding process is built around an ‘unzipping’ polymer–monomer system characterised by high reaction rates in both directions. J Chem Technol Biotechnol 2009; 84: 454–460 The monomer system suggested was the cyanoacrylate family of materials, well known as super-glue and used for adhesive bonding of a wide range of substrates. Polymerisation of the cyanoacrylates can be initiated by weak bases such as amines and even by water, and subsequent reaction is very rapid.3,4 Polymerisation can be retarded or inhibited by addition of weak or strong acids.5,6 Thermal decomposition of the polymer occurs at temperatures above about 180 ◦ C via a chain end initiated un-zipping reaction with quantitative conversion to monomer.7 Monomer Thermal Basic −− −→ −− −→ Monomer Polymer depolymerisation initiator The cyanoacrylates in common industrial use include the methyl, ethyl and n-butyl materials with slight differences in polymerisation and decomposition rate. The n-butyl monomer has an advantage in that the corresponding polymer is soluble in tetrahydrofuran and this facilitates molecular weight measurement. Successful operation of the process requires preparation of monomer–powder compounds which are stable for a sufficient period of time to allow moulding or extrusion, but then can be ∗ Correspondence to: Maurice N. Collins, Department of Materials Science and Technology, University of Limerick, Limerick, Ireland. E-mail: maurice.collins@ul.ie Department of Materials Science and Technology, University of Limerick, Limerick, Ireland. www.soci.org c 2008 Society of Chemical Industry Development of the powder reaction moulding process Powder www.soci.org The objective of the work reported here was to establish additional fundamental information on powder–binder relationships that would be useful in moving the process further towards industrial realisation. The shaping technology used is essentially compression forming rather than injection moulding, but the overall objective is to provide information that will allow development of an injection or compression moulding process. Three powders were chosen for the work, stainless steel, alumina and silicon nitride, and the binder used was n-butyl cyanoacrylate. These powders were chosen as representative materials with very different particle characteristics and surface chemistry. Parameters of interest are maximum powder volume fraction achievable, stability of the prepared powder-monomer compound, rheology of the moulding compounds under various shear conditions, polymerisation or curing rate and sintering and de-binding behaviour. Monomer Mix Mould Monomer recovery Cure EXPERIMENTAL Debind and sinter Figure 1. Powder reaction moulding process flow chart. polymerised rapidly once the desired shape has been achieved. Although commercial monomer does contain small amounts of polymerisation inhibitor, contact with a high surface area powder is liable to exceed this inhibitor capability and so additional inhibition is needed. Inhibition of polymerisation can be achieved using p-toluenesulfonic acid5 and initiation of polymerisation using caffeine. For example an inhibited monomerpowder compound could be introduced into a mould which has been pre-treated with caffeine and polymerisation will be rapidly initiated on the surface with slower progression into the bulk. Since the original process proposal a number of investigations have been conducted. Ridgway et al. reported8,9 on the manufacture of a conduit heart valve using alumina and cyanoacrylate and Ng et al10 carried out green state machining on compacts made from the same materials, this work demonstrating a further advantage of the process in that the green compacts are very tough, can be handled easily and shapes further modified by conventional machining technology. Details of the powders used are given in Table 1. n-butyl cyanoacrylate monomer, (Henkel, Ireland) was used with ptoluenesulfonic acid as inhibitor, which was dissolved in the monomer prior to blending with the powder. Powder monomer mixing was carried out by hand using a beaker and spatula and the maximum powder volume fraction that would give a homogenous compound, without free monomer or powder, was visually estimated. The cure time was measured as the time that the compound remained fluid, and this was tested by extrusion using a polypropylene syringe. The flow properties of particulate filled reactive fluids are not easy to measure in a meaningful way. Two approaches to rheological characterisation were adopted here; measurement of a flow behaviour index using an extrusion technique, and cone and plate rheometry. To measure the flow behaviour index an extrusion cylinder was installed in a universal test machine, set up in such a way as to allow measurement of the volume of material extruded under constant load. Prepared moulding compound was put into the cylinder and extruded under a load of 270 N, and the time t to travel 1 mm was noted. As the volume V of the syringe is known the flow index can be described as: Flow Index = V/t Although this system could in theory be analysed using pressure flow equations applied to the exit capillary, it is impossible to know how much work is being done within the body of the syringe and so a simple comparative result was accepted. A more formal measure of viscosity was obtained by using a HAAKE RotoVisco1 (Staffordshire, UK) fitted with a C35/2 cone Table 1. Characteristics of the powders used Surface area (m2 g−1 ) Particle size (µm) Powder name α-silicon nitride (HCST) Alumina AC2-325 (Aluchem) Stainless steel 316L (Sandvik-Osprey) Experimental Manufacturer Experimental (BET) – 5 (d50 ) from sedigraph see Table 2.3 22 (d90 ) 10.5 (d50 ) 4.0 (d10 ) 0.9 (d50 ) 13 (d50 ) – 0.9 m2 g−1 22 12 – – 26 c 2008 Society of Chemical Industry 455 J Chem Technol Biotechnol 2009; 84: 454–460 Manufacturer www.interscience.wiley.com/jctb www.soci.org Table 2. Maximum powder volume fractions achievable at various inhibitor concentrations Silicon nitride Acid level (v%) 0.1 0.2 1.0 2.0 4.0 Practical maximum volume fraction (%) 22 29 43 45 49 Alumina (AC2-325) Acid level (v%) Practical maximum volume fraction (%) 1.0 2.0 4.0 5.0 54 62 65 68 Stainless steel (316L) Acid level (v%) Practical maximum volume fraction (%) 0.1 0.2 0.5 1.0 2.0 4.0 33 38 60 65 70 75 and plate sensor system with 2◦ angle cone. At low powder volume fractions the cone and plate instrument was relatively successful in measuring viscosity but as the volume fraction was increased the scatter in the results increased considerably and abrupt changes of apparent flow behaviour were observed, associated with breakdown of laminar flow between the cone and plate of the measuring system. Where possible the Haake software was used to fit the flow behaviour to a power law equation to give the flow behaviour and consistency indices. All experiments were carried out at 22 ◦ C. Comparing the two methods of rheological assessment, the first approach uses pressure flow and resembles anticipated processing methods, but only gives a relative result, whereas the second approach, using drag flow, will give some indication of actual viscosity. It is appreciated that the shear rates used in these methods are low relative to plastics injection processing but they are relevant to compression moulding. Compacts were compression moulded at room temperature from the prepared compounds using simple disc moulds machined from PTFE, and were allowed to ‘self-cure’ without the use of initiator. De-binding was simulated by thermogravimetric analysis of material taken from the green compacts using a TA Instruments (West Sussex, UK) system and a Perkin Elmer (Watford, UK) differential scanning calorimeter (DSC). Sample size was approximately 15 mg and a heating rate of 5 ◦ C min−1 was used between ambient and 300 ◦ C under nitrogen. Moulded specimens were heated to 220 ◦ C to remove the binder. After de-binding, the samples were placed in a desiccator which was evacuated for a period of 1 h. Distilled water was then dropped into the evacuated chamber until the samples were immersed, and after 2 days they were weighed in water (m1 ). Excess water was then removed from the samples using tissue and they were again weighed (m2 ) in the wet condition after which they were placed in an oven to dry and weighed (m3 ). The bulk density was then calculated from equation shown below: (wet weight − immersion weight) = m3 ρL /(m2 − m1 ) The bulk volume Vb is : ρa = m3 ρL /(m3 − m1 ) Vb = (m2 − m1 )/ρL The open pore volume Vop is : 456 The apparent porosity (%Pa) is : www.interscience.wiley.com/jctb Table 3. Inhibition times for various concentrations of inhibitor Powder Silicon nitride Alumina Stainless steel Inhibitor level (%) Volume fraction (%) 0.1% 0.2% 1% 1% 1% 1% 2% 2% 4% 23 (Max.) 28 (Max.) 30 35 40 43 (Max.) 40 45 (Max.) 40 20 11 5300 5040 861 150 4968 3900 4944 1% 1% 1% 1% 2% 2% 2% 2% 4% 4% 4% 40 45 50 53 (Max.) 50 55 60 62 (Max.) 55 65 68 (Max.) 454 75 69 0.2 409 266 81 0.5 1653 743 333 0.1% 0.1% 0.2% 0.2% 0.5% 0.5% 0.5% 1.0% 1.0% 1.0% 1.0% 2.0% 2.0% 2.0% 2.0% 2.0% 4.0% 4.0% 4.0% 30 33 (Max.) 30 38 (Max.) 50 55 60 (Max.) 50 55 60 65 (Max.) 50 55 60 65. 70 (Max.) 65 70 75 (Max.) Working time (h) 0.5 0.25 5 0.17 5 2 1.5 8 7.5 5 2.5 37 34 11.5 10 2 23 4.5 2.5 To sinter silicon nitride, mouldings containing 6 wt% Y2 O3 and 2 wt% Al2 O3 as sintering additives were heated to 1800 ◦ C under a nitrogen atmosphere for 2 h. For alumina powders, 0.3 wt% MgO was used as additive and samples were sintered at 1700 ◦ C for 2 or 3 h. Stainless steel moulding was sintered under nitrogen atmosphere at 1400 ◦ C. RESULTS ρb = dry weight × liquid density/ The apparent density is : L Zhao, MN Collins, C Birkinshaw Vop = (m2 − m3 )/ρL %Pa = (Vop /Vb ) × 100% Table 2 shows the maximum powder volume fractions that can be safely combined with monomer containing various concentrations of polymerisation inhibitor, and Table 3 gives details of the curing times of various formulations. Figure 2 compares the flow indices for the various compounds. Figures 3 and 4 show, as examples, relative viscosities of the compounds at two specified shear rates. In this context relative c 2008 Society of Chemical Industry J Chem Technol Biotechnol 2009; 84: 454–460 Development of the powder reaction moulding process www.soci.org Flow Index Table 4. Derived power law equations Flow Index (mm3/s) 100 Silicon Nitride 10 Alumina 1 29% 34% 39% 44% 49% 54% 59% Volume Fraction Figure 2. Flow indices as a function of powder volume fraction. 100 1/s Shear Rate Relative Viscosity Silicon Nitride Alumina Stainless Steel 1000 Silicon nitride 25% Silicon nitride 30% Silicon nitride 37.5% Silicon nitride 40% η = 8.5 × 105 · S(0.6267 – 1) η = 2.2 × 106 · S(0.7295 – 1) η = 1.8 × 107 · S(0.152 – 1) η = 7.1 × 108 · S(0.0724 – 1) η η η η Stainless steel 45% Stainless steel 50% Stainless steel 55% Stainless Steel 10000 Power law equation Alumina 50% Alumina 55% Alumina 60% Alumina 62.5% 64% 0.1 0.01 Powder and volume fraction η = 3.8 × 105 · S(0.5778 – 1) η = 3.8 × 106 · S(0.4264 – 1) η = 5.7 × 107 · S(0.1279 – 1) = 1.5 × 106 · S(0.5737 – 1) = 1.5 × 107 · S(0.4083 – 1) = 2.0 × 107 · S(0.4184 – 1) = 1.1 × 108 · S(0.0069 – 1) 100 10 1 20.0% 30.0% 40.0% 50.0% Volume Fraction 60.0% 70.0% Figure 3. Relative viscosity values as a function of volume fraction measured at 100 s−1 shear rate. Relative Viscosity 182 1/s Shear Rate 450 400 350 300 250 200 150 100 50 0 20.0% Silicon Nitride Alumina Stainless Steel 30.0% 40.0% 50.0% Volume Fraction 60.0% Figure 5. TGA simulated de-binding of silicon nitride mouldings. 70.0% Figure 4. Relative viscosity values as a function of volume fraction measured at 182 s−1 shear rate. viscosity is defined as the viscosity of the compound compared with the viscosity of pure monomer measured under the same circumstances. Table 4 shows the derived power law equations for the moulding compounds. Figures 5 and 6 show, as typical results, the de-binding behaviour of silicon nitride mouldings assessed by thermogravimetric and DSC analysis. Table 5 compares experimental weight losses with theoretical weight losses for the cured compounds prepared with the three powders. Table 6 gives the density and porosity values for the de-bound mouldings before and after sintering. DISCUSSION J Chem Technol Biotechnol 2009; 84: 454–460 that the fine particle silicon nitride can only be compounded at low volume fractions relative to the other two powders and presumably both physical and chemical factors are responsible for this. The observation that with all three powders increasing inhibitor concentrations are required to give stable higher volume fraction compounds confirms that all powder surfaces are able c 2008 Society of Chemical Industry www.interscience.wiley.com/jctb 457 Achievement of the maximum powder volume fraction is clearly desirable as this will reduce overall shrinkage on de-binding and sintering. Considering the data in Tables 1 and 2 it is clear Figure 6. DSC simulated de-binding of silicon nitride mouldings. www.soci.org L Zhao, MN Collins, C Birkinshaw Table 5. The experimental weight loss and theoretical weight loss on de-binding TGA weight loss (%) Sample name Theoretical weight loss (%) Silicon nitride 30% Silicon nitride 35% Silicon nitride 37.5% Silicon nitride 40% 35 34 31 28 42.73 37.26 34.77 32.42 Alumina 55% Alumina 60% Alumina 62.5% Alumina 65% Alumina 68% 18 12.5 11 12 11.5 18.75 15.83 14.47 13.18 11.72 8.5 5 3 8.40 6.89 5.56 Stainless steel 60% Stainless steel 65% Stainless steel 70% Figure 7. Working windows for the three types of compound. Table 6. Density and porosity of de-bound mouldings before (B) and after (A) sintering Bulk density (g mL−1 ) Apparent density (g mL−1 ) Bulk volume (mL) Open pore volume (mL) Apparent porosity (%) SN 30% B SN 30% A SN 35% B SN 35% A SN 37.5%B SN 37.5%A 1.2909 1.7218 1.4655 1.7863 1.4905 2.1559 3.0915 3.2122 2.9852 3.1543 3.0029 3.1671 1.0683 0.6319 1.2486 0.7861 1.4542 0.4605 0.6222 0.2932 0.6356 0.3409 0.7324 0.1470 58.24 46.40 50.91 43.37 50.37 31.93 AL 55% B AL 55% A AL 60% B AL 60% A AL 65% B AL 65% A AL 68% B AL 68% A 2.0127 2.2987 2.3538 2.5303 2.2936 2.6240 2.2337 2.4543 3.8530 3.9226 3.8104 3.9407 3.8454 3.8601 3.6553 3.9308 2.2491 1.8378 2.3749 2.1115 3.0367 1.8091 3.5726 2.8395 1.0742 0.7608 0.9078 0.7557 1.2255 0.5793 1.3894 1.0666 47.76 41.40 38.23 35.79 40.36 32.02 38.89 37.56 SS 60% B SS 60% A SS 65% B SS 65% A SS 70% B SS 70% A SS 74% B SS 74% A 4.7526 6.5988 4.8059 6.7627 4.6615 6.7312 4.4988 6.4627 5.5296 7.1265 5.4919 7.1125 5.3889 7.0493 5.2934 7.2014 1.9643 1.4249 2.6823 1.8946 3.7941 2.6195 5.0118 3.5092 0.2760 0.1055 0.3350 0.0932 0.5121 0.1182 0.7523 0.3599 14.05 7.40 12.49 4.92 13.50 4.51 15.01 10.26 Sample name 458 to initiate polymerisation, and for each powder, the inhibitor concentrations required reflects the chemical activity of the total initiating powder surface. The chemical nature of the actual initiating species remains unknown. Stainless steel is presumed to have a surface rich in Cr2 O3 and may be slightly basic, while the alumina and silicon nitride should be neutral. The cyanoacrylate monomer possesses two strongly electron withdrawing substituents and so initiation is readily achieved by weak nucleophiles such as water. It is possible that initiation occurs here through adsorbed water, but the very www.interscience.wiley.com/jctb different inhibitor requirements and sensitivities suggest a more complex mechanism. A convenient way of comparing the stability of the various compounds is through a ‘working windows’ diagram, as shown in Fig. 7. The working window at the maximum volume fraction should be of the order of hours to a few days and the figure demonstrates that it is considerations of powder surface chemistry that dominate this variable. However, it is clear that manageable moulding compounds can be achieved with all three powders and this is considered to be an important observation. The flow properties of the moulding compounds are also clearly of great importance as the material has to be injection moulded, compression formed or perhaps extruded. The relative flow indices for the three types of compound as functions of volume fraction, shown in Fig. 2, are useful comparators of behaviour, demonstrating that the silicon nitride powder presents the greatest resistance to flow, an observation consistent with the previous remarks about particle size, surface area and maximum achievable volume fraction. Comparison of the data in Figs 3 and 4 shows the marked shear sensitivity of the moulding compounds, and Table 4 gives the derived power law relationships from the cone and plate rheometry. All the compounds are strongly pseudoplastic with very low flow behaviour indices at high volume fractions. It is presumed that following mixing the moulding compound consists of powder particles with adherent short polymer chains, initiated by the surface and killed by the inhibitor, a small quantity of dissolved low molecular weight polymer and a larger quantity of un-reacted monomer. All polymer solutions are pseudoplastic, but the extreme shear sensitivity demonstrated here is presumed to arise substantially from particle rearrangement processes rather than binder properties. A comparison of the relative viscosity values given in Figs 3 and 4 indicates that particle rearrangement occurs at low shear rates and is presumably complete at high shear rates. Whether this process could lead to particle anisotropy in fabricated components would require investigation. A number of models exist to predict the viscosity of suspensions of particles in fluids of known viscosity and these can be divided into two groups. The first group, detailed in Table 7, does not take the maximum volume fraction into account and are derived for suspensions of spheres in Newtonian liquids. Application of these models to the data for stainless steel shows that the experimental data lies within the range predicted by the models. Application of the same models to the two ceramic powders c 2008 Society of Chemical Industry J Chem Technol Biotechnol 2009; 84: 454–460 Development of the powder reaction moulding process www.soci.org Table 7. Semi-empirical models indicating the relationship between the volume fraction, φ, and viscosity, ηr Model Equation Comment ηr = 1 + 2.5φ + 10.05φ 2 + 0.0273 exp(16.6 φ) 1 5 7 ηr = 1 − 2.5φ + 11.0φ − 11.5φ ηr = 1 + 2.5φ + 14.1φ 2 1 2 2.5 η = (1 − 1.10φ − 0.97φ ) Thomas (1965)11 Ford (1960)12 Guth (1938)13 Vand (1948)14 φ ≤ 0.6 φ ≤ 0.52 Non-dilute suspensions of spheres φ ≤ 0.59 r ηr = (1 − 1.35φ)−2.5 Roscoe and Brinkman (1952)15,16 Table 8. Models indicating the relationship between the maximum volume fraction, φ, and relative viscosity ηr Model Equation 1.25φ 2 φ ) φm φ φm 2 ηr = [1 + 0.75( φ )] φm = 0.65 for glass beads of a uniform size 1− φm 1.25 φ ηr = (1 + φ − φ )2 ηr = (1 + Eliers (1941)17 Chong et al. (1971)18 Ferdos (1975)19 1− m Frankel and Acrivos (1967)20 Quemada (1977)21 Mooney (1951)22 For a system of permanent aggregates in Newtonian liquid φ 1 )3 ( 9 [ φm ηr = 8 φ 1 1−( )3 φm For concentraed suspensions of spheres in Newtonian fluids φ ηr = (1 − φ )−2 m Limited to shear large enough to disperse all aggregates but small enough to avoid any unstable effects or turbulence in the flow and any damage to the particles KE φ ln ηr = 1−Sφ kE = 2.5, s = 1/ m , for suspensions of rigid spherical particles in Newtonian fluids Table 9. Average model-predicted and measured maximum volume fractions Powder Silicon nitride Alumina Stainless steel Average of maximum volume fraction from prediction by different models Experimental maximum volume fraction 0.365 0.655 0.664 0.42 0.68 0.70 J Chem Technol Biotechnol 2009; 84: 454–460 c 2008 Society of Chemical Industry www.interscience.wiley.com/jctb 459 gave theoretical relative viscosities well below those obtained by experimental measurement and this is considered to arise from the non-spherical nature of the particles (relative to stainless steel) and particle surface initiation of binder. The second group of mathematical models, shown in Table 8, includes the maximum volume fraction, and has been used here by inputting measured viscosity data to predict the maximum powder loading, m , for a feedstock. Each of the models was developed for particular sets of circumstances, which will only be approximated here, and so the procedure adopted has been to calculate m values using each model and then take the average result. This approach allows estimation of the theoretical maximum powder loadings and in Table 9 these are compared with the maximum loadings obtained by experiment. The closeness of the experimental maxima and the theoretical maxima suggests that this approach is valid and that experimental values are close to optimum. The de-binding behaviour, shown as examples in Figs 5 and 6 for silicon nitride mouldings, is typical of that generally observed, and is consistent with a rapid depolymerisation of the binder, which is then able to diffuse from the moulding. In an industrial process this would be recovered for re-use. Comparing the experimental and theoretical mass losses suggests two points of note. First, the compounds are not completely homogeneous, and this can be explained by the hand mixing techniques used. Secondly, taken overall, mass losses are slightly less than the theoretical values. This merits further investigation as even small amounts of residual organic material could have consequences for a system such as stainless steel. It is, however, possible that any residual material would be lost during early stage sintering. Considering the density and porosity data in Table 5 it is clear that the sintering of the ceramic materials is far from satisfactory. With the alumina this arises in part from the use of a relatively coarse powder, chosen for ease of mixing. Although there is some use for porous ceramics in applications such as filters, such an outcome was not the intention of this work. It is considered that significant improvement could come from increasing moulding pressure during forming of the compacts and also from optimising the sintering heating programme. The results with the stainless steel are much more encouraging and show the benefit of working with a feedstock optimised for powder processing. www.soci.org L Zhao, MN Collins, C Birkinshaw CONCLUSIONS REFERENCES Essential aspects of the powder reaction moulding process have been demonstrated using three different types of powder, silicon nitride, alumina and stainless steel, and n-butyl cyanoacrylate as carrier and binder. Compounds were prepared for each system at maximum volume fractions, working windows were established, and compounds cured, de-bound and sintered. It is appreciated that the results generated are unique to particular combinations of powder and monomer, but nevertheless they demonstrate the validity of the underlying ideas. In the work reported here the approach has been to deal with premature reaction of the carrier-binder by dissolving appropriate quantities of inhibitor in the monomer. An alternative approach would be prior modification of the powder surfaces, perhaps by acid washing, and it is possible that this route may give manageable compounds at lower overall inhibitor levels. It is desirable that the inhibitor concentration be minimised to reduce the possibility of un-wanted residue and side reactions with the powders. Most of the results under discussion here were obtained using powder and monomer only and it is appreciated that with ceramic powders the inclusion of sintering aids, such as magnesia, may influence compound stability through reaction with inhibitor. Although sintering studies did not form a major part of the work described here it is appreciated that this is an important area to be addressed in the future. Relative viscosities were much higher than those predicted by simple models and all compounds were strongly pseudoplastic, and this behaviour is considered to arise from partial polymerisation of the monomer. Although compound viscosities are high compared with those encountered in conventional melt processing of polymers, the moulding compounds are capable of flow, although again this may occur in part through particle–particle contact as well as through hydrostatic pressure. Sintered porosity values obtained were high, particularly with the ceramic materials and this probably arises from the hand mixing and moulding and from the choice of powder used. It is very likely that machine mixing and application of compaction pressures such as those experienced in injection moulding would offer significant improvement when used in conjunction with optimised heating regimes. It is considered that the data presented suggest that the powder reaction moulding process has the potential to be developed into a viable manufacturing technology. Future work should utilise machine mixing and moulding, address powder–monomer reactivity through powder treatment and develop practical moulding technology. 1 Birkinshaw C and Hull B, A novel carrier/binder system for powder reaction injection moulding and extrusion. British Patent 94015914 (1994). 2 Birkinshaw C, Buggy M and O’Neill A, Reaction moulding of metal and ceramic powders. J Chem Technol Biotechnol 66:19–24 (1996). 3 Pepper DC, Kinetics and mechanisms of zwitterionic polymerizations of alkyl cyanoacrylates. Polymer J 12:629–637 (1980). 4 Pepper DC and Ryan B, Kinetics of polymerisation of alkyl cyanoacrylates by tertiary amines and phosphines. Makromol Chemie – Macromol Chem Phys 184:395–410 (1983). 5 Pepper DC and Ryan B, Initiation processes in polymerizations of alkyl cyanoacrylates by tertiary amines. MakromolChemie-Macromol Chem Phys 184:383–394 (1983). 6 Costa G, Cronin JP, Pepper DC and Loonan C, Termination and transfer by acid in the pyridine-initiated polymerization of butyl cyanoacrylate. Eur Polym J 19:939–945 (1983). 7 Birkinshaw C and Pepper DC, The thermal degradation of polymers of n-butylcyanoacrylate prepared using tertiary phosphine and amine initiators. Polym Degrad Stab 16:241–259 (1986). 8 Ridgway JS, Hull JB and Gentle CR, A PRIME approach for the moulding of conduit ceramic parts. J Mater Process Technol 133:181–188 (2003). 9 Ridgway JS, Hull JB and Gentle CR, Development of a novel binder system for the manufacture of ceramic heart valve prosthesis. J Mater Process Technol Spec Issue 109:161–167 (2001). 10 Ng SH, Hull JB and Henshall JL, Machining of novel alumina/cyanoacrylate green ceramic compacts. J Mater Process Technol 175:299–305 (2006). 11 Thomas DG, Transport characteristics of suspensions. J Colloid Sci 20:267–277 (1965). 12 Ford TF, Viscosity-concentration and fluidity-concentration relationships for suspensions of spherical particles in Newtonian liquids. J Phys Chem 64:1168–1174 (1960). 13 Guth E, On the hydrodynamical theory of the viscosity of suspensions. Phys Rev 53:322 (1938). 14 Vand V, Viscosity of solutions and suspensions I Theory. J Phys Colloid Chem 52:277–299 (1948). 15 Roscoe R, The viscosity of suspensions of rigid spheres. Br J Appl Phys 3:267–269 (1952). 16 Brinkman HC, The viscosity of concentrated suspensions and solutions. J Chem Phys 20:571 (1952). 17 Eilers H, The viscosity of emulsions made from highly viscous materials as a function of concentration. Kolloid – Z 97:313 (1941). 18 Chong JS, Christiansen EB and Baer AD, Rheology of concentrated suspensions. J Appl Polym Sci 15:361 (1971). 19 Fedors RF, Viscosity of Newtonian suspensions. Polymer 16:305–306 (1975). 20 Frankel NA and Acrivos A, On the viscosity of concentrated suspension of solid spheres. Chem Eng Sci 22:847–885 (1967). 21 Quemada D, Rheology of concentrated dispersed systems and minimum energy dissipation principle. Rheol Acta 16:82–94 (1977). 22 Mooney M, The viscosity of a concentrated suspension of spherical particles. J Colloid Sci 6:162–170 (1951). 460 www.interscience.wiley.com/jctb c 2008 Society of Chemical Industry J Chem Technol Biotechnol 2009; 84: 454–460