英文原文1.pdf

J. C. Gelin1 Professor e-mail: jean-claude.gelinens2m.fr Th. Barriere J. Song Department of Applied Mechanics, FEMTO-ST Institute, ENSMM Besanon, 26 Rue de lEpitaphe, 25030 Besanon, France Processing Defects and Resulting Mechanical Properties After Metal Injection Molding The paper is concerned with occurrence of processing defects and resulting mechanical properties associated with material processing by metal injection molding (MIM). MIM process is a multistep one that consists fi rst in the injection of metallic powders mixed with a thermoplastic binder, followed by a debinding stage that permits to evacuate the polymeric binder, and then followed by a sintering stage by solid state diffusion that normally leads to a nearly dense component. The main defects arising during MIM processing are associated with powder segregation during injection molding, and uncom- pleted or heterogeneous mechanical properties resulting from solid state diffusion. The paper fi rst describes a biphasic fl uid fl ow approach that can accurately predict powder volume fraction after injection molding and consequently the associated segregation defects. This analysis is followed and continued by a proper sintering model based on an elastic-viscous analogy that predicts the resulting local densities after sintering and also associated defects. So, from the two subsequent models, it becomes possible to get the fi nal powder densities after processing and to localize the possible resulting defects. This analysis is completed by an analysis using a porous material model to get the fi nal resultant mechanical properties after processing. ?DOI: 10.1115/1.2931155? Keywords: metal injection molding, defects, segregation, sintering, modeling 1Introduction Metal injection molding ?MIM? is relatively new processing technology used in powder metallurgy industries, which is espe- cially effi cient and benefi cial for manufacturing small and intri- cate metallic components in large quantities. It includes four basic steps consisting in mixing the powders and binders, injection molding, debinding, and fi nally sintering the powder skeleton ?1?. The defects arising in each MIM steps should be properly con- trolled in order to obtain the fi nal components with the required properties. Both injection molding and sintering are the most im- portant steps related to get the green parts and the fi nal ones, respectively. As in the injection molding process for thermoplastic feedstock, the defects such as jetting, air trap, dead zones, welding lines, etc., can also occur in MIM. However, the power-binder separation, called phase segregation, happens during the high speed and high pressure injection molding process due to different densities associated with metallic powders and thermoplastic binders. It can induce the inhomogeneity of the green compo- nents. All these effects appearing in the injection step are certainly amplifi ed in all next following ones ?2?. After the debinding step, the binder is removed and the remained component results in a porous one including both powders and pores. In the following sintering step, the debinded components are treated at a tempera- ture just below the melting point of the main constituent in order to obtain the required fi nal density by bonding the powder par- ticles together through diffusion. The shrinkage between the green component resulting after injection molding, debinding and then sintered, is typically in the range 1020% and the fi nal density is in the range 95100%. In order to get the fi nal components with the required dimensional accuracy and specifi ed mechanical prop- erties, it is thus necessary to control the defects such as inhomo- geneous shrinkages, distortions, cracks, etc. These defects are in- fl uenced by the material and processing factors including initial density, heating rate, sintering temperature and atmosphere, fric- tion, gravity, etc. ?3?. The experimental investigations concerning the defects arising in MIM process are presented in this paper. The conventional trial and error methods are widely used in the MIM industries to obtain the qualifi ed products by adapting and adjusting tools and processing parameters iteratively. The numeri- cal simulations for MIM are now in development and are well expected in order to provide a cost effectively alternative to trial and error methods. This paper focuses on the modeling and nu- merical simulation of the injection molding and sintering steps in MIM. A biphasic model is presented for the fl uid-particle fl ows arising in the injection process ?4,5?. Each phase is characterized by its own density, velocity fi eld, and volume fraction. An inter- action term between the powders and the thermoplastic polymers accounted for the momentum exchange between both phases. A new effi cient explicit algorithm has been implemented in the fi nite element ?FE? software developed in our research team. This newly developed algorithm can solve the biphasic incompressible fl ow problem explicitly with an excellent effi ciency ?6,7?. The phenom- enological sintering models based on the continuum mechanics concepts are employed to predict the fi nal component dimensions ?810?. The material and process parameters in the used visco- plastic constitutive law are identifi ed by the bending tests carried out in sintering conditions and by the dilatometry tests ?11,12?. The model and the identifi ed material parameters are implemented in theABAQUSFE solver in order to perform numerical simula- tions of the sintering step associated with MIM. The powder vol- ume fraction contours resulting from the injection molded compo- nent issued from biphasic injection simulation are then accounted in the following sintering simulations. Based on the simulation results, the fi nal mechanical properties as mechanical strength are then predicted. The experimental investigations on 316L stainless steel have been carried out to verify the proposed modeling and simulations. 1Corresponding author Contributed by the Materials Division of ASME for publication in the JOURNAL OF ENGINEERINGMATERIALS ANDTECHNOLOGY . Manuscript received August 23, 2007; fi - nal manuscript received February 8, 2008; published online December 9, 2009. As- soc. Editor: Matthew P. Miller. Paper presented at the Material Processing Defects MPD 5 held in Ithaca ?Cornell University? on July 2007. Journal of Engineering Materials and TechnologyJANUARY 2010, Vol. 132 / 011017-1 Copyright 2010 by ASME Downloaded From: / on 04/10/2013 Terms of Use: /terms 2Experimental Investigations 2.1Materials and Procedures. A 316L stainless steel feed- stock was provided by Advanced Metal Working Particles, LLC., Carmel, IN, and wax-based binders were used for the feedstock. The powder volume fraction of the feedstock is 62%. A micro- structural photograph of the feedstock observed by scanning elec- tronic microscope ?SEM? is shown in Fig. 1. The powders are of spherical shape with a particle size smaller than 45?m and D80 =16?m. It can be observed that the powders and binders are well mixed and one has to underline that the homogeneity of the feed- stock is important in the following processing steps. The injection molding is the step that consists to shape the feedstock into the desired geometries in MIM processing. The step includes heating the feedstock at a suffi cient temperature to make it as a melt, forcing melt to fl ow into the mold cavities, then packing at high pressure, and fi nally cooling and ejecting the molded parts out of the mold cavities. In the experimental work, a 22 ton injection machine was used. The thermal debinding was used to remove the wax-based binders from the 316L stainless steel powder molded parts. The presintering and sintering pro- cesses were conducted in a batch furnace under vacuum condi- tions. Various injection molding, debinding, and sintering process- ingweretestedtoinvestigatetheireffectsonthe fi nal components. 2.2Jetting Defects During Injection Molding. Jetting refers to the phenomena that occur when the melt does not form a uni- form fl ow front, but rather proceeds as a fi ngerlike steam, main- taining the geometry of the gate as it enters the die cavity ?13?. There are two forms of jetting in MIM described in literature: conventional jetting ?in liquid stage? and solid-phase jetting. For conventional jetting, a single liquid fl ow steam moves to the far face of the cavity, then upon fl ow reversal, fi nally forms a fl ow front that fi lls the cavity backward. Conventional jetting results in defects in the fi nal molded part. In the case of solid-phase jetting, one solid fi ngerlike fl ow steam piles up upon itself instead of forming a backward fl ow. The consequence of solid-phase jetting is surface irregularities including weld lines and cracking. The glass windows of the mold cavity allow the quasicontinu- ous monitoring of mold fi lling using a fast charge coupled device ?CCD? camera that records the front mixture advances during fi ll- ing, see Fig. 2 ?14?. Then, image processing software provides a quasicontinuous view of the fi lling stage. In MIM, the rheological behavior of the powder/binder mixture is largely different from the rheological behaviors of thermoplastic polymers due to the fact that the amount of powder is very large, e.g., 60% in volume. We focused here on the analysis of incidents associated with jet- ting in the cavity with the same thickness as the cross section of the gate. For this purpose, different runners and lengths are used. The dimensions of the die cavity were also different. The injection parameters used to obtain the components are described in Table 1. These parameters are in agreement with the ones proposed by the feedstock provider ?15?. Different fi lling stages are recorded by a CCD camera during injection molding with a frame interval equal to 0.04 s. These records provide an accurate description of the fi lling patterns dur- ing injection process, as shown in Fig. 2. The injection of original feedstock is shown in Fig. 2?a?. One can observe that the 316L melt inserts into the cavity like a fi nger steam at the beginning of the injection stage. Then, the steam reaches the opposite wall in cavity. Afterward, the cavity is fi lled, mainly along the injection direction, until the fi lling is completed. It should be mentioned that the melt steam is extremely curved in the middle of the cavity. The steam overlaps cause the problems for continuation of the injection process. Jetting phenomenon is a conventional one in loaded polymer injection molding; this phenomenon is undesir- able as defects may result in the fi nal components. The second column presents a jetting under control with the recycled feed- stock after the fi rst injection stage. It indicates that an initial jet is formed, but it is followed by an acceptable fi lling result, see Fig. 2?b?. This phenomenon is called solid-phase jetting in literature. Melt begins to fully fi ll cross sections of the cavity since middle of the injection process. The frozen tail sticks on the down part of the cavity. The frames also indicate that the components injected with recycled feedstock are more homogeneous compared to com- ponents injected with the original one, due to the fact that the recycled feedstock is more homogeneous than the original one. 2.3Segregation Defects During Injection Molding. In order to investigate the segregation between the powders and binder during the injection molding, a fi ve-cavity mold has been de- signed and realized for the experiments. With the injection pro- cessing parameters as presented in Table 1, the obtained molded components are shown in Fig. 3?a?. The molded tensile test speci- men and the wheel component were cut into small segments. A helium pycnometer and an accurate balance were used to obtain Fig. 1SEM photograph of the MIM feedstock composed of gas-atomized 316L stainless steel powders and wax-based thermoplastic binders Fig. 2Jetting phenomena arising in mold cavity during the fi lling stage with 316L stainless steel based feedstock: a original feedstock and b recycled feedstock Table 1Processing parameters for the injection of 316L based feedstock ParametersUsed Injection pressure ?bar?160 Injection velocity ?mm/s?160 Mold temperature ?C?50 Melt temperature ?C?185200 Packing pressure ?bar?45 Injection time ?s?0.18 011017-2 / Vol. 132, JANUARY 2010Transactions of the ASME Downloaded From: / on 04/10/2013 Terms of Use: /terms the local apparent density of the molded components, as shown in Figs. 3?b? and 3?c?. The segregation effect depends on the feed- stock properties, mold design, and the processing parameters. 2.4Cracks and Distortions Occurring During the Debind- ing Stage. In the debinding process, the components become gradually fragile due to the removal of the binder in the molded parts. The defects such as cracks and distortion are prone to occur during debinding. The debinding process should be designed properly accordingly to the feedstock and shape of the compo- nents. The proposed thermal cycle for the tensile and bending test specimens as shown in Fig. 3?a? consists in heating up to 130C with a rate equal to 0.625C/min, then heating up to 220C with a slower rate equal to 0.1C/min and then holding for 1 h ?7?. However, when this cycle was employed for the debinding of hip implant prototype made with the same feedstock as Fig. 4?a?, the cracks occur, as shown in Fig. 4?b? ?16?. Simultaneously, when the hip implant is debinded on the plate support, there is obvious distortion at the contact position due to the gravity, as shown in Fig. 4?c?. In order to avoid these defects, slower heating rates for the thermal debinding cycles and a proper support with the half cavity as the injection mold were employed ?16?. 2.5Uneven Shrinkage and Distortion in Sintering. The shrinkage of the parts during sintering should be uniform to render possible the design of the cavity of the injection mold. Unfortu- nately, the factors such as green inhomogeneity, gravity, friction, temperature gradient, etc., make the shrinkage uneven. A sintering test has been conducted for the tensile test specimen, as shown in Fig. 5. The mean shrinkages of the sintered tensile specimen mea- sured by experiments correspond to 13.11%, 14.09%, and 14.55% in the length, width, and thickness directions, respectively. The defects associated with injection molding and debinding cannot be removed, but are amplifi ed in the sintering process. As an ex- ample, the distortion of the bending test specimen after sintering at 1150C is shown in Fig. 6. It is due to the green inhomogeneity induced by short shot or under packing in the injection molding. 3Biphasic Injection Simulation for MIM 3.1Biphasic Model for Injection Molding. The simulation of the injection stage in MIM is carried out under the frame of the Eulerian description. The injection fl ow of feedstock mixture is expressed by the fl ows of two distinct phases, namely, the solid one to express the fl ow of metallic powder and the fl uid one for the fl ow of polymer binder. Both distinct fl ows are described by their proper NavierStokes equations that are coupled through the momentum exchange terms. At each instant t, the volume fractions of each phase in the fi lled portion of the mold cavity are defi ned by two variables?s and?f , named solid and fl uid volume fractions, respectively. Due to the mass conservation,?sand?fshould continuously satisfy the following saturation conditions: ?s+?f= 1 and ? ?t?s +?f? = 0?1? The fl ows of solid and fl uid phases are described by two distinct velocity fi elds Vsand Vf, respectively, and the effective velocity Veff for the mixture is defi ned as Veff=?sVs+?fVf?2? The fi lling front is tracked by the advection effect associated with the effective velocity fi eld. A fi lling state fi eld variable F is used Fig. 3Inhomogeneous green density in the molded components due to segregation defects occurring during injection step of MIM for 316L stain- less steel: a injection molded components, b density contours of tensile test specimen, and c density contours of the wheel component through segmented parts Fig. 4Cracks and distortion defects occurring during the debinding of 316L stainless steel MIM hip implants: a injection molded and debinded components, b cracks occurring during debinding, and c distortion oc- curring during debinding Journal of Engineering Materials and TechnologyJANUARY 2010, Vol. 132 / 011017-3 Downloaded From: / on 04/10/2013 Terms of Use: /terms to express the fi lling stage and corresponds to a value equal to 1.0 in the fi lled part of the mould cavity, while the value in the un- fi lled part of the mold cavity is zero. This fi eld variable permits to express the front position versus time. A TaylorGalerkin method is used for the treatment of the associated advection equation for the fl ow of the mixture: ?F ?t + ? ?VeffF? = 0?3? So, the mass conservation for the fl ows of each phase results in the following equations that can be used directly to evaluate the evolution of volume fractions: ?s ?t + ? ?sVs? = 0 and ?f ?t + ? ?fVf? = 0?4? The solution of Eq. ?4? directly measures the segregation effects at the prescribed instant. Incompressibility of the mixture is equivalent to the mass con- versation constraints for each phase resulting from the associated saturation for their volume fractions, which leads to a single equa- tion for the mixture expressed as the incompressibility condition: ? ?Veff? = 0?5? In the MIM injection stage, as the Reynolds number is gener- ally small, it permits to neglect the advection terms in Navier Stokes equations. Then, momentum conservation equations f