Euro PM2011 – Powder Injection Moulding: Feedstock Development Manuscript refereed by Professor José Torralba, Madrid University Carlos III

Behaviours of Feedstock Including Not-Tailored Powder for Powder Injection Moulding a,b*

b

a

a

BRICOUT Julien , GELIN Jean-Claude , ABLITZER Carine , MATHERON Pierre and BROTHIER a Méryl a CEA, DEN, DEC, SPUA, LCU, F-13108 Saint-Paul-lez-Durance, France. b Université de Franche-Comté, FEMTO-ST, DMA, F-25044 Besançon Cedex.

Abstract This study demonstrates the preponderant influence of powder characteristics in the powder injection moulding process. The originality is to use agglomerated powders with particle characteristics not tailored for the process. Tests were conducted, on the same binders, including different kinds of alumina powders offering a panel of size, shape, size distribution, specific surface area and agglomeration state. The mixing step significantly changes the agglomeration state allowing the achievement of quite high critical solid loading value. Among the characteristics of powders, the deagglomeration has been particularly studied. The rheological analysis carried out for each feedstock at optimal solid loading highlighted a pseudoplastic behaviour and satisfied rheological criteria imposed. It is also shown that the specific surface area of powders has a strong influence on the implementation and the behaviour of the feedstocks. The study demonstrates that coupling between rheological results and mixing torques provide an accurate determination of optimal solid loading. Key-words: Alumina powder, powder injection moulding, feedstock, mixing, deagglomeration, rheology.

1.Introduction Powder Injection Moulding (PIM) is a near-net shape processing technique allowing a mass production of complex components. Ceramic or metal powders can be used in this process. PIM consists in mixing powder and binders to obtain a mixture called feedstock. The resulting feedstock is injected into a mould (green part). The unwanted binders are then removed (brown part) while retaining the original shape and finally the brown part is sintered to near the full-density in a controlled atmosphere furnace at high temperature [1]. A judicious selection of powder and binders is needed to obtain a feedstock suitable for injection. Generally, the binder system is a multi-component polymer system allowing, on the one hand, the flow of feedstock during injection and the other hand a gradual removal of the binders in a wide temperature range, improving the debinding process. The selection of a polymer system has been widely studied by different authors offering an important range of choice [1,2,3,4]. The choice of powder content and morphology has a huge importance on the feedstock properties, especially on the rheological behaviour and the critical powder volume concentration. German et al. [1] have defined the characteristics of an “ideal” powder for the process among which the main ones are a particle size between 0.5 and 20 microns with a d50 in the range from 4 to 8 μm, a tap density greater than 50% of the theoretical density and no agglomeration. However, in some cases, the choice of the powder is imposed and does not meet the criteria outlined above. The use of submicron powder is tricky due to the agglomeration phenomena. Indeed, the presence of agglomerates can alter the flow [5] and cause defects and heterogeneities after sintering [6]. To allow the production of parts free of defects when using this kind of powder, the mixing operation must allow the deagglomeration of the powder thanks to the shear stresses generated in the system. The deagglomeration comes from two different mechanisms. The first is the breakdown of agglomerates into different fragments, while the second is the erosion characterized by a progressive and continuous detachment of small fragments occurring at a lower shear stress than rupture one [7]. In many cases, the deagglomeration comes from a combination of both mechanisms [8]. The purpose of this paper is to show that the deagglomeration allows using powders that are not so well appropriated for the injection process. Different alumina powders, not tailored for PIM, have been selected and implemented with the same binders system in the same mixing conditions. The innovative aspect of this work is to directly implement raw powders on the process by using quite different types of powders in terms of size, shape, specific surface area and agglomeration state.

*

[email protected]

Euro PM2011 – Powder Injection Moulding: Feedstock Development

2.Experiments 2.1Powder features Alumina powders tested in the experiments, directly come from the supplier without modifications. Eight high purity alumina powders offering a large panel for size, shape, size distribution, specific surface area (SSA) and agglomeration state were selected. Some of the characteristics of these powders such as particle size distribution (measured with a Beckman Coulter LST 13 320 particle size analyser), particles density (measured with Micromeritics AccuPyc II 1340 pycnometer), specific surface area (measured with Micromeritics ASAP2020), apparent and tap density (NF EN ISO 3953) are presented in Table 1. Table 1. The characteristics of the eight alumina powders that are used in this work

Density (g/cm³) D50 SSA

Powder ID : CR1 CR6 CR15 Pycnometer 4 4 3.9 Apparent 0.72 0.59 0.46 Tap 0.84 0.72 0.53 µm 1.5 0.7 0.4 m²/g 3.5 5.8 14.6

CR30 3.8 0.34 0.45 a 1.1 23.8

GE15 4 0.35 0.56 15.5 13.6

BA15 BRA15 BA15W 4 4 3.9 0.31 0.31 1.23 0.45 0.41 1.33 18.7 0.42 1.2 15.4 15.7 18.7

a. SEM observations show smaller elementary particles. This high value (in comparison with what we could expect) is certainly an artefact due to the very little size of the elementary particles which compose this powder

The particle size distributions coupled with SEM observations confirm that elementary particles of these eight powders have a submicronic size and non spherical shape. The manufacturing method and the size of elementary particles of the powders generate phenomena of agglomeration and aggregation with agglomerates which are different in terms of size, shape and cohesion (Figure 1). 3 The very low tapped density of these powders, inferior to 20% of theoretical density (3.96 g/cm ), comes from the presence of agglomerates. This work, using the powders selected, allows first to know the influence of specific surface area during mixing for the same processing conditions, and secondly the influence of the manufacturing method for the same specific surface area. Indeed, powders can be grouped into three manufacturing methods: un-milled (BA15, GE15), milled (CR1, CR6, CR15, BRA15, CR30), and atomized (BA15W). Furthermore, powder CR15 comes from the milling of GE15, as wall as for BRA15 and BA15. Characterizations of powder prove that the selected powders are not in agreement with a classical PIM powder.

BA15

20 µm

BA15

20 µm

BA15W

50 µm

Fig1. SEM observations of BA15 (left), CR30 (middle) and BA15W (right)

2.2Experimental procedure The binder system chosen, which already prove its efficiency for high-pressure injection of small parts 3 (~ 3 cm ), is based on paraffin wax [2]. This binders system is stable under mixing and moulding conditions and allows an independent and progressive departure of each component, getting theoretically part without defects after debinding. ® ® The mixing of powder and binders was carried out on a BRABENDER Plasti-Corder two screw 3 mixer. The pair of rotor blades used offers a capacity tank equal to 55 cm . The torque value is related to the resistance on the rotor blade. The analysis of the torque variation during mixing is used to determine the Critical Powder Volume Concentration (CPVC) or critical solid loading, the optimal solid loading and to rule about the feedstock homogeneity (steady state of the mixing torque). ® The rheology of the alumina feedstocks at the optimal solid loading was evaluated using a MALVERN RH2000 capillary rheometer equipped with a 32 mm long and 2 mm diameter capillary die at 200°C. This temperature corresponds to a possible injection temperature for which no degradation of binders occurs. Raw rheological data will be compared without applied the Bagley correction. To determine the powder characteristics after mixing, powder samples were obtained from feedstocks debounded at 600°C. SEM observations and particle size distribution have been implemented.

Euro PM2011 – Powder Injection Moulding: Feedstock Development

3. Results and discussion 3.1Critical powder volume concentration The theoretical CPVC can be calculated directly from the characteristics of the raw powder. Indeed, it corresponds to the state where every interstitial void left by the powder is filled by binders, when powder is packed as tightly as possible [9]. Ratio of tap to pycnometer density gives the theoretical CPVC, without taking account an eventual reorganization or deagglomeration of the powder. The other way to determine the critical solid loading is the torque rheometry. By studying the torque behaviour at each solid loading, the experimental CPVC can be found. Indeed, just before this value, a sharp increase of the torque appears ant it becomes very erratic and unstable. Just after CPVC, the occurrence of "free powder" that no longer belongs to feedstock leads to a loss of cohesion of the system, so a decrease of the torque. Table 2 regroups calculated / experimental CPVC and optimal solid loading deducted from the experimental CPVC. Table2. Critical and optimal solid loading of alumina feedstocks determined by calculations and torque rheometry CR1 20,95 60-62 58

Calculated CPVC Experimental CPVC Optimal solid loading

CR15 13,32 58-60 54

BA15W 34,02 56-58 54

CR6 17,96 56-58 54

GE15 13,98 56-58 54

CR30 11,66 56-58 54

BRA15 10,2 56-58 54

BA15 11,06 48-50 46

The comparison between experimental and theoretical value of CPVC highlights an important change of the characteristics of powders during the mixing stage. Indeed, it is well accepted that the range for the real CPVC in PIM feedstock is usually from 10 to 20 %vol. higher than the value determined by powder densities [10] due to the reorganization of the powder during mixing. In our case, the difference is over 40 % which suggests not only a powder rearrangement, but a possible deagglomeration. Despite the very different characteristics of powders, CPVC are quite similar and higher for 7 among the 8 powders tested (between 58 and 62 %vol.). Mixing process significantly alters the original powder characteristics resulting in an increase of the packing density. 3.2Mixing impact on the powder characteristics The deagglomeration of the powders inside the feedstock was seen as a macroscopic parameter with CPVC investigations. In the following section, the analysis of the powders after mixing and feedstock debinding allows to understand the impact of mixing on the microscopic characteristics of the powders. At this scale, the deagglomeration occurs differently depending on the powder used. The different cases encountered are grouped and discussed below. For the non-milled GE15 powder and atomized powder (BA15W), the deagglomeration is characterized by a breakdown of agglomerates. As example the particle size distribution of GE15 shows the breakage of agglomerates of 20 microns initially present. This change leads to a surge of elementary particles (0.3 microns) and the creation of new granulometric classes with size of 1, 5 and 10 microns (Figure 2). The deagglomeration causes an increase in the width of particle size distribution facilitating and improving the packing of powder within the feedstock. CPVC is thereby naturally increased. The SEM micrographs on figure 3 corroborate the size reduction of agglomerates. Particle size distribution of GE15 5

4

Before mixing

Before

4,5

After mixing

Volume (%)

3,5 3 2,5

GE15

20 µm

BA15W

100 µm

GE15

20 µm

BA15W

100 µm

2

After

1,5 1 0,5 0 0,01

0,1

1

10

Particle diameter (µm)

100

1000

Fig2. (left) Particle size distribution of GE15 (middle) SEM micrograph of GE15 and (right) BA15W powder before and after mixing

The milled powders (CR1, CR6, CR15, CR30 and BRA15) exhibit the same behaviour in the modification of the powder characteristics, more or less exacerbated depending on the powder. Initially, these powders contain 3 different set of particle size: the elementary particles corresponding to about 0.5 microns, a main population to 1 µm and another to 4 μm. It is observed for these powders an increase in the volume of elementary particles at the expense of 1 µm agglomerates causing a

Euro PM2011 – Powder Injection Moulding: Feedstock Development decrease in d50 and a wider particle size distribution. The deagglomeration time during the mixing step depends on the powder. The deagglomeration mechanism seems to be erosion because of the mixing time required to obtain a constant and stabilized torque. The deagglomeration is quite long; the time and the way for powder introduction have to be adapted to the powders used. The higher the specific surface area is, the longer is the mixing time and the slower is the filling speed. In alumina powders, agglomerates are linked by physical forces, especially those of Van der Walls [11]. The Van der Waals attractive force between particles becomes far greater than the gravitational force, as the size of particles, reduces. The higher the specific surface area of the powder is, the greater the cohesion of agglomerates is, increasing the mixing iterations needed and the residence time required in the mixer to get a proper deagglomeration. Figure 3 shows the two types of behaviour: CR1 with a rapid reorganization/deagglomeration and CR15 with a slow one. Plastogram of CR15 Plastogram of CR1 6

Particle size distribution CR1

Volume (%)

5 4

Particle size distribution CR15

5

6

Before mixing After mixing

Volume (%)

7

Before mixing After mixing

4

0

15

30

3

45

0

3

15

30

45

60

2

2

1

1 0 0,01

0,1

1

10

100

0 0,01

0,1

1

10

100

Particle diameter (µm)

Particle diameter (µm)

Fig3. Particle size distribution before and after mixing and the plastogram of (left) CR1 and (right) CR15

These results are in concordance with those obtained by Contreras et al.[12], who studied the influence of particle size distribution on the CPVC of Inconel 718 feedstocks. Indeed, the mixing causes a reduction in the size of agglomerates with creation of elementary particles. The average diameter of the system is reduced and the width of the particle size distribution is increased; these two parameters controlling strongly the value of CPVC. 3.3Rheological behaviour of feedstock A feedstock charged at the optimal solid loading contains a greater proportion of binders than into a critical solid loading feedstock. The feedstock must have, in theory, between 2 and 5 %vol of powder less than the critical powder volume concentration. This allows a better flow behaviour and better flexibility with respect to variations of the characteristics of powders from different batches [1]. The rheological behaviour of feedstock charged at the optimal solid loading (= CPVC – 4 %vol) has been investigated. The aim is to confirm that the flow is possible at the optimal solid loading and to check if the rheological properties of feedstocks meet the injectability criterion. The results show two typical behaviours. The first category of feedstock, comprising the powders CR15, CR30, BRA15 and GE15 does not flow through the capillary rheometer die, due to the too high viscosity at optimal solid loading. The optimal solid loading should be reduced and it is determined by the maximum solid loading feedstock which is able to flow through the capillary die (Table3). The second category of feedstock, comprising CR1, CR6, BA15 and BA15W flows correctly through the capillary die. 100000 The resulting curves can be fitted by a power law like , n representing the power law or flow index. It provides the sensitivity degree of the viscosity ( ) with respect to shear rates ( ) variations. Results10000 are summarized in figure 4 and table 3. 100000

CR1 58%vol

CR1 - 58%vol CR6 - 54%vol

1000

CR15 - 50%vol

Shear viscosity (Pa,s)

10000

CR6 54%vol

CR30 - 46%vol BA15 - 46%vol 100

1000

BA15W - 54%vol

CR15 50%vol

GE15 - 52 %vol

10010 CR30 46%vol 10 1 1

10

100

1000

BA15 46%vol

10000

Shear rate (/s)

1 1

10

100

1000

Fig4. Rheological behaviours of feedstocks charged at optimal solid loading at 200°C

10000

BA15W 54%vol GE15 52%vol CR30 -

Euro PM2011 – Powder Injection Moulding: Feedstock Development Table 3. Rheological characteristics of alumina feedstocks Optimal solid loading Fitted curve equation n Viscosity at -1 100s (Pa.s)

CR1

CR6

BA15

CR15

GE15

CR30

BA15W

BRA15

58

54

46

50

52

46

54

-

η = 180γ-0,25

η = 286γ -0,41

η = 2424γ-0,50

η = 6034γ-0,54

0,75

0,59

0,5

0,46

0,36

0,34

0,26

-

56

44

237

509

933

1317

135

-

η = 17536γ-0,64 η = 27167γ-0,66 η = 4088,2γ-0,74

-

The optimal solid loading had to be re-estimated downwards for powders CR15, CR30, GE15 and BRA15, as the feedstock at the optimal solid loading deducted from experimental CPVC does not flow. Optimal solid loading for all alumina feedstocks are given in Table 3. The values of the flow index are all lower than the unity, showing the pseudoplastic behaviour of the feedstock. In addition, the shear -1 viscosity at the shear rate equal to 100 s is less than 1000 Pa.s (except for the feedstock containing CR30, where the viscosity is slightly higher). Thus, the feedstocks fulfill the rheological injectability criteria. The flow index values, ranging from 0.26 to 0.75, strongly depend on the powders used. Flow index low values correspond to a marked pseudoplastic behaviour. The flow index could be assimilated like the measurement of the disturbance produced by the solid loading on the polymeric binders [13]. The disturbance is much higher when the power law index n is low; this latter increasing with a temperature increase or decreasing with a solid loading increase. Table 3 shows that the solid loading has a slight influence on the flow index. Indeed, the CR1 powder that has the higher optimal solid loading also causes the least disturbance into the polymer system with a high flow index. Contrariwise, the specific surface area of powders has a major impact on pseudoplastic behaviour of feedstocks; the greater the specific surface area of the powder used is, the greater the disturbance increases, decreasing the flow index. For example, CR30, with the largest specific surface area and the lowest optimum charge, has a very low flow index. On the other hand, the shape of powders is also very important, especially the shape of residual agglomerate after mixing. The atomized powder (BA15W) has a low flow index (n = 0.26) compared to its specific surface area. Indeed, the powders with a surface area of 15 m² / g have normally an index between 0.35 and 0.50. -1 The comparison of shear viscosities at optimal solid loading and shear rate equal to 10 and 100 s also shows the influence of specific surface area powders on this parameter (Figure 5). For the same fabrication, the higher the specific surface area of powders is, the higher the shear viscosity is, irrespective of the solid loading used. The friction between inter-particles agglomerates during the flow is responsible of the viscosity of the system. When the specific surface of the powder is high (CR30), the size of elementary particles composing the agglomerates is very fine, increasing contact and friction between agglomerates. The viscosity of the system is therefore very important. For the same specific surface area, the shape has a slight influence on the shear viscosity. The microscopic characteristics of powders have therefore an important responsibility on the rheological behaviour. 10000

25

Shear viscosity (Pa.s)

Shear viscosity at 10s-1

20

Specific surface area

1000

15 100 10

10

Specific surface area (m²/g)

Shear viscosity at 100s-1

5

1

0

CR1

CR6

GE15

CR15

BA15

BA15W

CR30F

-1

Fig5. Specific surface area and viscosity at 10 and 100 s versus powders

3.4Determination of optimal solid loading by analyses of torque evolution – Coupling with rheological results. Given the results of flow, get the optimal solid loading from experimental CPVC is very rough for agglomerated powders. Indeed, the optimal solid loading has been re-estimated for the half of the powders used in this work. The use of rheological measurements to determine the optimal solid loading is efficient, but is very long and tedious. It is possible to use the torque values during mixing to

Euro PM2011 – Powder Injection Moulding: Feedstock Development

12

Mixing torque (N.m)

Mixing torque (N.m)

get a rapid and accurate value for the optimal solid loading. The analysis of the evolution of mixing torque stabilized at different solid loading (every 2% for example) allows this determination. This analysis leads to identify three distinct phases. At low solid loading, the torque of the system corresponds to this of the binder system, so in our case, a too low couple to be detected. At a given solid loading, the torque increases slightly at each solid loading. This phase can be regarded as pseudo-stable. The third phase corresponds to a sharp increase of the mixing torque leading to the critical solid loading. The coupling of rheological studies with the analysis of torque rheometry indicates that the optimal solid loading is at the border between the pseudo-stable phase and the sharp rise of the mixing torque (Figure 6). Note that it confirms the relationship CPVC - 4% for the case of powders CR1, CR6 and BA15W. Optimal solid loading

10

Zone 2

8

Zone 3

CR1 CR6

6 4 2

Zone 2

0 50

52

54 56 Solid loading (%vol)

Zone 3

58

60

16 14 12 10 8 6 4 2 0

Zone 2

Zone 3 CR15 CR30

38

40

42

Zone 2

Zone 3

44 46 48 50 52 Solid loading (%vol)

54

56

58

Fig6. Mixing torque according to solid loading for milled powder

4.Conclusion The choice of a powder for the PIM process must not only be based on the initial characteristics of the powder. The use of agglomerated powder having particle characteristics (submicronic particles) not intended for the PIM process, is possible. The mixing step allows reducing the size of the agglomerates and the average diameter of powders causing an increase of the width of the particle size distribution. These changes in the powder characteristics allow the achievement of high critical solid loadings equal or greater than 58%vol, for 7 of the 8 powders studied. The analysis of the feedstocks rheology shows a pseudoplastic behaviour with a shear viscosity of less than 1000 Pa.s at -1 a shear rate equal to 100 s , satisfying injectability conditions. The rheological behaviour is strongly dependent on the characteristics of the powders. More the specific surface area increases, more the pseudoplastic character is marked with high viscosity value. The presence of agglomerates which have different shapes and numbers of elementary particles is an explanation of the different behaviours. The determination of the optimal solid loading from the critical solid loading is not immediate for agglomerated powders. Variations of mixing torques for several solid loadings show a sharp increase beyond a given solid loading. By linking with the rheological results, it is clear that this given solid loading corresponds to the optimal solid loading. The method for determining the optimal solid loading by torque rheometry is very effective for this kind of powder. References [1] German R.M and Bose A., 1997, “Injection molding of metals and ceramics”, Princeton, New Jersey, USA, MPIF. [2] Quinard C., Barriere T. and Gelin J.C., 2009, Powder Technology, vol.190, pp. 123-128. [3] Mutsuddy B.C. and Ford R. G., 1995, “Ceramic Injection Molding”. [4] Ahn S., Park S., Lee S., Atre S. and German R.M.,2009, Powder Technology, vol.193, pp.162-169. [5] Lewis T.B. and Nielsen L.E., 1968, Transactions of the society of rheology, vol.12, pp. 421. [6] Lange F.F. and Metcalf M., 1983, Journal of American Ceramic society, vol.166, pp. 398-406. [7] Rwei S.P. and Manas-Zloczower I., 1990, Polymer engineering and science, vol.30, pp701-706. [8] Suri P., Atre S.V., German R.M. and Souza J.P., 2003, Material science & engineering, vol.A356, pp. 337-344. [9] Warren J., German R.M., 1989, Modern Developments in Powder Metallurgy, Vol. 18, pp. 391-402. [10] Dihoru L.V., Smith L.N. and German R.M., 2000, Powder Metallurgy, vol.43, pp. 31-36. [11] Rao S.A, 1987, Ceramics International, Vol.13, pp. 139-143. [12] Contreras J.M, Jiménes-Morales A. and Torralba J.M., 2010, Powder injection moulding international, vol.4, pp. 67-70. [13] Contreras J.M, Jiménes-Morales A. and Torralba J.M., 2007, Powder injection moulding international, vol.1, pp. 59-62. Acknowledgment The authors would like to thank Jérémy Raguin, Mohamed Sahli and Thierry Barrière from FEMTO-ST for their technical and scientific support.

Euro PM2011 – Powder Injection Moulding: Feedstock Development Manuscript refereed by Professor José Manuel Torralba, Madrid University Carlos III

Powder Injection Moulding of Light Alloys Powder 1

2

Engin Ergul , Levent Acar , H. Ozkan Gulsoy

2,3

1

Dokuz Eylül University, Izmir Vocational School, Department of Technical Programs, Buca, Izmir, 35160, Turkey 2 Marmara University, Technology Faculty, Metal. and Mater. Dep., Göztepe, Istanbul, 34722, Turkey 3 TUBITAK-MRC, Materials Institute, Gebze - Kocaeli, 41470, Turkey

ABSTRACT This paper describes the microstructural and mechanical properties of injection molded of Al and Ti6Al4V (Ti64) materials. Al and Ti64 powders were injection molded with wax-based binder. The critical powder loading for injection moulding was 62.5 - 67.5 vol. %.Binder debinding was performed in solvent and thermal method under high purity argon atmosphere. After debinding the samples were -4 sintered at different temperatures and times in high purity argon and vacuum atmosphere (10 torr). The performances of the sintered parts were characterized using tensile testing, hardness testing, optical microcopy (OM) and scanning electron microscopy (SEM). The strengths and weaknesses of the test conditions have been analyzed from the densification behavior, microstructure, and mechanical properties. The results show that Al and Ti64 powders could be sintered to a maximum 96.2 % and 99 % of theoretical density. Maximum ultimate tensile strength and hardness were obtained 126 MPa, 46 HB at 650 ºC for 2 hours for Al powder and 704 MPa, 38.6 HRC at 1275 ºC for 10 hours for Ti64 alloys. INTRODUCTION Powder injection moulding (PIM) is a process, which combines the advantages of plastic injection moulding and conventional powder metallurgy (PM) technologies. This technique combines the benefits of the plastic injection moulding with the material versatility of the traditional powder metallurgy, producing highly complex part of small size, tight tolerance, and low production cost. The process overcomes the shape limitation of traditional powder compaction, the cost of machining, the productivity limits of isostatic pressing and slip casting, and the defect and tolerance limitations of conventional casting. Mechanical properties of a well-processed powder injection molded material are indistinguishable from cast and wrought material. The PIM process is composed of four sequential steps; mixing of the powder and organic binder, injection moulding, debinding (binder removal), and sintering [1]. During the last decade, interest in powder injection moulding (PIM) of light alloys such as aluminum, titanium and their alloys parts for the general purpose (household, automobile, dental, biomedical, military and aerospace industry) has dramatically increased [2], as evidenced by the number of scientific papers and successful use-oriented developments. Lightweight structural components formed from aluminum and titanium is increasingly attractive in view of increasing energy costs and requirements to reduce atmospheric carbon emissions. Aluminium and aluminum alloys have been investigated due to its attractive properties, such as high specific strength with good ductility, high thermal conductivity, excellent electric conductivity, corrosion resistance and low density compared to other materials [3-7]. Titanium and titanium alloys have a low density, high strength and excellent corrosion resistance in many media and is known to be biocompatible [8-10]. Aluminum and aluminum alloys have relatively low cost and strength, but titanium and titanium alloys have high cost and high strength. Both materials can be easily produced by conventional powder metallurgy processing, but major problem to be overcome in the conventional powder processing is the oxide film and redisual radical elements such as C, N and N [3, 8]. These problems are big barrier for sintering behaviours and final properties of Al and Ti alloys. Earlier investigations on injected moulded Al and Ti alloys focused on the effect of powder characteristics (gas and water atomised mixed powder), development of binder system and debinding properties, sintering temperature, sintering time, heat treatment, residual carbon and oxygen content on microstructure and small additive such as tin to enhance sintering behavior and mechanical

Euro PM2011 – Powder Injection Moulding: Feedstock Development properties and particle reinforced Al and Ti parts [3,4,8,10 -13]. Sintering densification, microstructural evaluation and mechanical and thermal properties of injection moulded Al and Ti powders were studied by several investigators [3,10].However, microstructural and mechanical properties of Al and Ti powder have not been compared and explained. This present work was aimed to investigate and compare the effect of sintering parameters and final properties of powder injection moulded Al and Ti64 alloys. Metallographic techniques were employed to sinter the tensile bars to investigate the sintering behaviour. Tensile and hardness properties of the sintered products were evaluated in sintered condition. Powder morphology, fracture surfaces of moulding samples were analyzed under a scanning electron microscope. EXPERIMENTAL PROCEDURES In this research, gas atomized aluminum powders (99.5 purity %, 0.45 O %, 0.040 Si %, 0.010 Fe %) provided by Ecka Granulate GmbH & Co Ldt. and gas atomized Ti64 powders (Ti-5.9Al-3.9V-0.19Fe0.12O-0.01C-0.01N-0.004H) provided by SOLEA Corp. (France) were used. Al powder has particle size distribution of D10=3.88 m, D50=7.35 m and D90=13.62 m. and Ti64 powder has particle size distribution of D10 = 10.32 μm, D50 = 24.61 μm, D90 = 45.61 μm. Particle size distributions of Al and Ti64 powder were determined on Malvern Mastersizer (UK) equipment and given in Fig. 1. Morphology of the Al and Ti64 powders observed using scanning electron microscopy (Jeol- JSM 6335F- Japan) is given in Fig. 2.

Fig. 1. Cumulative particle size distributions for Al and Ti64 powder.

Fig. 2. Scanning electron micrograph of Al (left) and Ti64 (right) powder. A multiple-component binder system consisting of paraffin wax (PW), polypropylene (PP), carnauba wax (CW), and stearic acid (SA) was used. Feedstock was prepared at 175 ºC with the binder melted

Euro PM2011 – Powder Injection Moulding: Feedstock Development first and then powder blend added incrementally. The powder loading in this mixture was 62.5 vol%. for Al and 67.5 vol% for Ti64 powder. After cooling, the feedstock was pelletised by hand. These feedstocks were injected using a 12.5 MPa specially made injection-moulding machine to produce tensile (MPIF 50) test specimens. Debinding was conducted in a two-step solvent/thermal operation. Green parts were solvent debound at 70 ºC for 7 h. for Al samples and 60 ºC for 4 h. for Ti64 samples in heptane; followed by thermal debinding step at 1-2 ºC/min to 540 ºC for 1 h in Ar atmosphere for Al samples and 600 ºC for 1 h and -4 pre-sintered at 4 ºC/min to 900 ºC for 1 h under the vacuum (10 torr) for Ti64 samples. Thermo gravimetric analysis (TGA) is measured as a function of temperature while the sample is subjected to a controlled temperature program. This can be achieved as a function of the increasing temperature or isothermally as a function of time with an Ar atmosphere. The TGA measurements were done on a SII 6300 TGA-DTA (SII Nanaotechnology Inc., Tokyo, Japan) with heating rates, 10 ºC /min, from room temperature up to 600 ºC for feedstock. The sintering cycle applied to the Al samples was as follows; samples were heated to 580 ºC - 650 ºC sintering temperatures at a rate of 10 ºC/min. and they were held at 650 ºC for 60 min. in high purity N2 atmosphere. The sintering schedule for Ti64 samples was as follows; samples were heated to 1100 ºC at a rate of 10 ºC/min., then the samples were heated to different sintering temperatures of 1150 ºC, 1200ºC, 1260ºC and 1275 ºC at a rate of 10 ºC/min and -4 they were held at each temperature for 120 min. under vacuum (10 torr). The densities of the sintered samples were measured by means of the Archimedes water-immersion method. For metallographic examination, samples were cut from the center of the each sintered tensile test bar. Keller’s and Kroll reagent were used to etch the samples for optical metallography. All tensile tests were performed using Zwick 2010 mechanical tester at a constant crosshead speed of 1 mm/min (25 mm gauge length). The hardness tests were performed using an Instron-Wolpert Dia Testor 7551 at Brinell and Rockwell C scale. At least three specimens were tested under the same conditions to guarantee the reliability of the results. The powder morphologies and fracture surfaces of sintering samples were examined using a scanning electron microscope. RESULTS AND DISCUSSION Fig. 3. is a typical fracture surface of a green part shows that a good homogeneity was achieved for Al and Ti64 samples. Scanning electron micrographs taken at the center of the fractured green molded part show a uniform distribution of the binder throughout the part. A thin layer of the binder around almost all the particles can be seen which is considered useful for facilitating flow during the moulding stage. Spherical shape of powders can be seen in these fractures of green parts.

(a)

(b)

Fig. 3. SEM fractographs in centre area of molded Al (a) and Ti64 (b) sample. The TGA of feedstocks helps to design the thermal debinding cycles. Fig. 4 shows the weight loss– temperature plot with 10 ºC/min. heating rate for the Al and Ti64 feedstock. Below 200°C, no materials decomposes, but at 200°C, wax starts to decompose, and it makes many paths for degassing around aluminum particles. Then, at 300°C, all binders decompose with increasing temperature. Finally, over

Euro PM2011 – Powder Injection Moulding: Feedstock Development 475°C, all of binder constituents have been decomposed. This is a basic debinding mechanism in binder system. The effect of sintering temperature and times on the theoretical density of Al and Ti64 sample is shown in Fig. 5. Fig. 5a shows that theoretical densities of samples increase with increasing sintering temperatures. The Ti64 samples were sintered at 1150 ºC for 2h; a maximum theoretical density of only 85 % was achieved. Samples attained a maximum theoretical density of 96.9 % at sintering temperature of 1275 ºC for 2h. Fig. 5b shows that theoretical densities of samples increase with increasing sintering time [8-10]. From Fig. 5b it can be seen that at sintering temperature of 1275 ºC for 1h., maximum theoretical density of 93%. Samples attained a maximum theoretical density of 99 % at sintering temperature of 1275 ºC for 10h. Similarly, Al samples were sintered at 580 °C for 60 min.; a maximum theoretical density of only 88.1 % was achieved. Theoretical density of Al sample is low, because of more porosity [3,4]. Samples attained a maximum theoretical density of 96.5 % at sintering temperature of 650 °C for 60 min. Fig. 5d shows that theoretical density of samples increase with increasing sintering time. From Fig. 5b, it can be seen that at sintering temperature of 650 °C for 30 min, maximum theoretical density of 96 %. Samples attained a maximum theoretical density of 96.5 % at sintering temperature of 650 °C for 120 min. This result shows that the sintering temperature and time improve the theoretical density of Al and Ti64 samples [3,5,6,7]. CW

PP

PW

120

Weight, %

100 80 60 40 20 0 0

100

200 300 400 Temperature, oC

500

600

Fig. 4. TGA curves of binders and feedstock. Fig. 6 shows the optical micrographs of injection molded Al and Ti64 samples sintered at 650 °C for 2 hours and 1275 °C for 10 hours. The porosity decreases and sintered density increases with increasing sintering temperature for Al and Ti64 samples. Microstructures at low sintering temperature have more porosity. As a result, the value of sintered density is in very low level. Porosity decreased and sintered density increased with increasing sintering temperature and time [1,3,8]. Atmosphere of nitrogen in sintering stages increases the sintered density. In particular, the nitrogen atmosphere during Al sintering is known to improve the sintering behavior [3,4]. Fig. 6b. shows microstructure of Ti64 sample sintered 1275 °C for 10 hours. The matrix is the small second phase particles are in all samples. The Ti64 was found to have a coarse acicular microstructure, revealing grains with intergranular -phase. The percentage of -phase in the alloys depend on the sintering conditions. A lower sintering temperature affects porosity and the amount of -phase, which appears precipitated at the grain boundaries, darker than -phase. Samples sintered for high temperatures and times possess Widmanstatten microstructure, which consisted of and lamellae. The grains were outlined by the phase and the contained several colonies of and lamellae. The colonies had and lamellae aligned in the same orientation. The grains, colonies and phase thickness in the grains had much effect on properties [8,10].

Euro PM2011 – Powder Injection Moulding: Feedstock Development

(a)

Sintering time : 2 h.

Theoretical Density, %

Theoretical Density, %

100 95 90 85 80 1100

1150

1200

1250

100 98 96 94 92 90

1300

0

2

Sintering Temperature, ºC

100

4

6

8

10

12

Sintering Time, h.

(c)

Sintering Time : 60 min.

(b)

Sintering temperature : 1275 ºC

98

100

(d)

Sintering Temperature : 650 oC

99

Theoretical Density, g/cm 3

Theoretical Density, g/cm 3

96 98

94 92

97

90

96

88 86 560

95

580

600

620

640

Sintering Tem perature, oC

660

0

30

60

90

120

150

Sintering Tim e, m in.

Fig. 5. Effect of sintering temperature and time on theoretical density of injection molded (a, b )Ti64 samples and (c, d) Al samples.

(a) (b) Fig. 6. Optical micrographs of injection molded (a) Al sample sintered at 650 ºC for 2 h., (b) Ti64 sample sintered at 1275 ºC for 10 h (black points indicate pores). Table 2. gives the overall mechanical properties of Al and Ti64 samples. It can be seen that all mechanical properties, sintered and theoretical density of injection molded Al and Ti64 samples. Al and Ti64 sintered samples were contained amount of oxygen, nitrogen and carbon more than starting powders. It could be found that the residual carbon content, nitrogen and oxygen in debinding stages [8].

Euro PM2011 – Powder Injection Moulding: Feedstock Development Table 1. Mechanical properties and densities of Al and Ti64 samples Sample Sintering temp., ºC Sintering time, h Theoretical density,% UTS, MPa Elongation, % Hardness C, % O, % N, %

Al 650 1 96.2 126 26.8 46 HB 0.19 0.72 0.95

Ti64 1275 10 99 704 6.37 38.6 HRC 0.10 0.25 NA

CONCLUSIONS Experimental results show that the sintering temperature and time increased sintered density, ultimate tensile strength, elongation and hardness of injection molded Al and Ti64 samples. For Al samples, the theoretical density, ultimate tensile strength, elongation, and hardness were 96.2 %, 126 MPa, 20.8 % and 46 HB, respectively. For Ti64 samples, the theoretical density, ultimate tensile strength, elongation, and hardness were 99 %, 704 MPa, 6.37 % and 38.6 HRC, respectively. The conditions used for manufacturing these materials lead to good mechanical properties. Light materials such as aluminum, titanium and their alloys can be easily produced by powder injection moulding process. ACKNOWLEDGEMENTS This work was supported by the Scientific Research Project Program of Marmara University (Project FEN-YLS-290107-0047, FEN-C-YLP-181208-0284 and FEN-C-YLP-210311-0055). The authors are grateful to Marmara University for their financial support and the provision of laboratory facilities. REFERENCES [1] R.M. German and A. Bose: Injection Moulding of Metals and Ceramics. 1997. p 259. MPIF, NJ. [2] ] R.M. German: PIM International, Vol.3 No. 4 December 2009, p 21-37. [3] Z. Y. Liu, T. B. Sercombe and G. B. Schaffer: Powder Metall.,2008, 51, (1), 78–83. [4] Z. Y. Liu, D. Kent and G. B. Schaffer: Mater. Sci. Eng. A, 2009, A513–A514, 352–356. [5] H. Kwon, D. H. Park, Y. Park, J. F. Silvain, A. Kawasaki and Y. Park: Met. Mater. Int., 2010, 16, (1), 71–75. [6] L. K. Tan and J. Ma: Proc. 9th Intersociety Conf. on ‘Thermal and thermomechanical phenomena in electronic systems’, Vol. 1; 2004,Las Vegas, NV, IEEE. [7] K. Katou, T. Sonoda, A. Watazu, Y. Yamada and T. Asahina: J. Jpn Soc. Powder Metall., 2003, 51, (7), 492–498. [8] G. Shibo, Q. Xhuanhui, H. Xinbo and D. Bo-hua: Trans. Nonferrous Met., 2004, 14, (6), 1055. [9] Y. Wu, R. Wang, Y. Kwon, S. Park and R. M. German: Int. J. Powder Metall. 2006, 42, (3), 59. [10] G. Shibo, Q. Xuanhui, H. Xinbo, Z. Ting and D. Bohua: J. Mater.Proc. Tech., 2006, 173, (3), 310. [11] A. T. Sidambe, I. A. Figueroa, H. Hamilton, I. Todd: PIM International, 2010, Vol.4 No.4, p 54-62. [12] E. Baril: International, 2010, Vol.4 No.4, p. 22-33. [13] R. Zhang, J. Kruszewski and J. Lo: PIM International, 2008, Vol.2 No. 2, p.74-78.

Euro PM2011 – Powder Injection Moulding: Feedstock Development Manuscript refereed by Dr Gemma Herranz, Castilla La Mancha University

Viscosity and Thermal Conductivity Study of Feedstocks for PIM Process: Influence of Powder Size and Chemical Nature C. Rigollet (1), D. Checot-Moinard (1,2), P. Lourdin (1) Laboratoire de Sciences des Matériaux (1) Ecole Catholique d’Arts et Métiers, 69321 Lyon, France, [email protected] (2) Arts et Métiers ParisTech, LaBoMaP, 71250 Cluny, France, [email protected] Abstract Powder Injection Moulding PIM process is based on injection of mixtures defined as “feedstocks” made of polymeric binder and metallic or ceramic powder. Feedstock viscosities are key parameter to control the injection step having strong impact on final parts in particular regarding complex parts with small structural details. On the same manner, thermal conductivity is a crucial feedstock property. Nevertheless, besides numerous studies on feedstocks, only a few experimental works have addressed their thermal conductivity at temperature and pressure close to those encountered in the injection moulding process. However, these data are searched for in order to perform numerical simulation of PIM process. In this contribution, we study thermal conductivity and viscosity of some commercial and "self-made" feedstocks. The influence on thermal conductivity and viscosity of parameters such as the solid loading, the powder size and chemical nature is highlighted. Introduction Powder and Micro-Powder Injection Moulding (PIM, µPIM) combine plastic injection process and powder metallurgy. These processes are based on the same principle which can be divided in 4 steps respectively denoted as: feedstock preparation, moulding, debinding and sintering. The first step is very important, because feedstock quality will affect all subsequent processing operations. Indeed different parameters such as powder characteristics, binder composition, powder/binder ratio, mixing process are crucial [[1], [2]]. In order to manufacture moulded part without defects (density gradient, cracks, porosity), feedstock characteristics have to be taken into account, to determine injection temperature and pressure, and the best mould temperature [3]. In order to obtain good injection conditions, feedstock viscosity has to enable optimum mould filling. That is why knowledge and understanding of rheological behaviour of feedstock have to be highlighted. Then much attention has been paid to study feedstock viscosities. R. M. German showed viscosity measurement is appropriate tool to depict and characterize feedstock homogeneity [[4]]. In the same manner, a lot of scientific papers dealing with feedstock viscosity and rheological behaviour were published to further characterize feedstock homogeneity [[5]]. Influence of solid loading close to critical solid loading on rheological behaviour was also studied [[6], [7], [8]]. Moreover the rheological properties knowledge is useful to indicate the feedstock ability to be injected. It is generally accepted that viscosity has to be -1 lower than 1000Pa.s in shear rate range from 100 - 1000 s to be considered suitable for injection moulding [9]. Thermal conductivity is another property which is essential for optimization of the injection process. During cooling, but also during mould filling, thermal conductivity knowledge is crucial to avoid non uniformity, shrinkage, distortions and every defect possibly caused by a rapid cooling or/and solidification [10]. Feedstocks thermal conductivity depends not only on binder thermal conductivity, which is often very low (thermal conductivity of polymers is low), but also on ceramic or metallic powder thermal conductivity which can be very high. Solid loading is of course the first feedstock characteristic to take into account regarding thermal conductivity [11-13]. Binder composition, and particles sizes and shapes are also important properties[11,13,14]. After characterizing some commercial feedstocks we focused in this contribution on the solid loading and powder size influences on feedstock rheological properties and thermal conductivity. To better understand the viscosity of such system, we decided to increase very gradually the solid loading from 10%vol to almost the critical solid loading and to perform experiments with two powders having the same chemical nature, roughly the same morphology but two different size distributions. Since typical 2 5 shear rate encountered in PIM process varies between 10 to 10 s 1, feedstock viscosity was studied 3 -1 as a function of shear rate ranging from 100 to 5.10 s . Thermal conductivity was also tested on feedstocks with different powders sizes and chemical natures using the same binder composition, in temperature and pressure range close to injection conditions.

Euro PM2011 – Powder Injection Moulding: Feedstock Development Materials and Experimental section Materials High density polyethylene (HDPE), paraffin (melt temperature: 64-68°C) and stearic acid SA were purchased from Sigma Aldrich. BMA15 (d50=0.17µm) and SMA3 (d50=0.33µm) alumina powders come from Baikowski. They are different size, hence they present specific surface area measured by BET method equal to 14m²/g for BMA15 and 6m²/g for SMA3. WC-10%mass.Co powder (d50=0.45µm) was purchased from Hexametal. Studied commercial feedstocks are: Advamet 316L (stainless steel feedstock) from Advanced Metalworking, PXA-321 (zircona feedstock) from Tosoh, PolyMIM FN02 (low alloy steel feedstock) and PolyMIM WC0,8Co10 from PolymerChemie. Feedstock preparation Feedstocks based on SMA3, BMA15 and WC-10%mass.Co were prepared using Mixer W50 from Brabender (binder : 65%mass HDPE/ 30%mass paraffin/ 5%mass stearic acid). Mixing was carried out in the following manner: PEHD is first placed in the mixer at 180°C. Once PEHD melted, paraffin, SA and alumina powder were added. The mixing speed is 30rpm. Rheological tests were performed on capillary rheometer Smart RHEO 5000, twin bore apparatus from CEAST. Tests were performed with one or two dies (diameter 1mm) and length 20 and 5 mm. Thermal conductivity Thermal conductivity measurements were performed on capillary rheometer Smart RHEO 5000, twin bore apparatus from CEAST, with a system developed by CEAST according to ASTM D5930 and ISO 22007. The measure is made through special piston with needle at different temperatures and pressures of the sample. The needle generates heat through an heating wire and measures the temperature with a thermocouple (Figure 1).

Figure 1 : Thermal conductivity measurement principle

Results and Discussion Rheological behaviour of commercial feedstocks Powder sizes are very different from one commercial feestock to another. In Advamet 316L feedstock, powder size is around 15-20µm, while in Tosoh PXA321 zircona feedstock it’s only 400nm, 2-3µm in polyMIM FN02 low carbon steel feedstock and about 1µm in polyMIM WC feedstock. Due to these different sizes, solid loadings are also quite different, that is about 67% in Advamet feedstock, and 5052% in polyMIM and Tosoh PXA321 feedstocks. Even if binders are different, injection temperatures are in the same range (170°C-200°C) for all these commercial feedstocks. Feedstocks are considered as non Newtonian fluid since the viscosity decreases as the shear rate increases, consequently the Rabinowitsch correction has to be applied in order to correct the shear rate at the wall following the equation eq(1). w

app 4

3

d ln app d ln

eq (1) with

 w the actual wall shear rate,

 app the apparent shear rate and

w

wall shear stress. Then, the actual viscosity

can be expressed as :

w

w

eq(2)

w

the

Euro PM2011 – Powder Injection Moulding: Feedstock Development

Figure 2 : Viscosity of commercial feedstocks measured at 170°C Viscosity of these feedstocks was measured at temperature close to injection temperature as function of shear rate and results of actual viscosity (taking Rabinowitch and Bagley corrections into account) can be observed on Figure 2. All commercial feedstocks present viscosity ranging from 1000 to 10 Pa.s in the studied shear rate domain, and follow a power law model (Ostwald de Waelle model), according to equation

K  n eq(3) with  the shear rate,

the shear stress, K(Pa.sn) is the consistency index and n is

the dimensionless flow behavior index. For all commercial feedstocks, n is lower than 1 (0,27