Journal of Materials Processing Technology 166 (2004) 430–439

Development and assessment of a new quick quench stir caster design for the production of metal matrix composites

S. Naher∗, D. Brabazon, L. Looney

Materials Processing Research Centre, Dublin City University, Dublin 9, Ireland

Received 11 August 2004; received in revised form 7 September 2004; accepted 8 September 2004

Abstract

Examination of the liquid and semi-solid stir casting method to produce Al–SiC composites was the focus of this study. A significant part of the work consisted of the design, construction and validation of a specialised quick quench compocaster for this high temperature processing method. Stainless steel was chosen as the main crucible and stirrer material. The machine consisted of a four 45◦ flat bladed stirrer and a crucible in a resistance heated furnace chamber. A linear actuator was integrated to this rig to allow the crucible to be quickly extracted from the furnace for quenching. Stirring speed ranging from 200 to 500 rpm and different shear periods were investigated. Ten percentage volume of 30 J.m sized SiC particles was used. The main research challenge was to get a uniform distribution of SiC in the aluminium matrix. In the compocasting experiments it was found that the uniformity of SiC particles in the aluminium matrix were dependent on shear rate, shear period, cooling rate and volume fraction of primary solid. The quick quench compocaster system was successful in producing cast MMC samples. The use of clean heat-treated SiC particles and the quick quench method was sufficient to produce homogeneous composites. Castings from the liquidus condition were found to result in poor incorporation of SiC particles whereas castings from the semi-solid condition were found to produce a uniform distribution of SiC particles. However, quicker solidification, after cesation of mixing, was found to improve the uniformity of the SiC distribution significantly. Characterization of the MMC samples produced included microstructure recording and image analysis thereof. The matrix phase size, morphology and distribution of SiC particles throughout the stir castings were examined.

Keywords: Stir casting; Liquid and semi-solid processing; Al–SiC MMC; Microstructure; Image analysis

1. Introduction

There are several fabrication techniques available to man- ufacture MMC materials. Depending on the choice of matrix and reinforcement material, the fabrication techniques can vary considerably. Fabrication methods can be divided into three types. These are solid phase processes, liquid phase process and semi-solid fabrication process. Solid state pro- cesses are generally used to obtain the best mechanical prop- erties in MMCs, particularly in discontinuous MMCs. This is because segregation effects and intermetallic phase forma- tions are less for these processes, when compared with liq- uid state processes [1]. Among the variety of manufacturing

processes available for discontinuous MMC production, stir casting is generally accepted, and currently practiced com- mercially [2,3]. Stir casting of MMCs generally involves pro- ducing a melt of the selected matrix material, followed by the introduction of a reinforcing material into the melt and ob- taining a suitable dispersion through stirring. Vogel et al. [4] gave the term ‘stir-casting’ to the production of metals with spheroid like microstructure by a shearing action induced by stirring. The term stir-casting and compocasting are used in- terchangeably in this work. Its advantages lie in its simplicity, flexibility and applicability to large scale production. It also, in principle, allows a conventional casting route to be used.

This semi-solid metallurgy technique is the most econom- ical of all the available routes for MMC production. It al- lows very large sized components to be fabricated, and is able to sustain high productivity rates. According to Skibo

S. Naher et al. / Journal of Materials Processing Technology 166 (2004) 430–439

et al. [5], the cost of preparing composite materials using a casting method is about one third to one half that of com- peting methods, and for high volume production, it is pro- jected that costs will fall to one tenth [6]. Potentially, this is a way of making a broad range of MMCs whereby the dispersoid is added to the surface of the melt and then be- comes entrained in the melt by agitation and/or mechani- cal work [6,7]. This process can take the form of various semi-solid metal processing (SSP) routes e.g. compocast- ing/rheocasting/thixocasting/thixoforging [8]. The rheocast- ing technique was initially extended to produce metal matrix composites in the early 1970s [9].

Since reinforcements such as ceramic particulates, short fibers, or whiskers have poor wettability to molten metals, it is very difficult to fabricate MMCs by mixing reinforce- ments and liquid metal. However, the reinforcement may be more easily incorporated into a semi-solid alloy slurry of a matrix formed by rheocasting. Once the reinforcements are introduced into the semi-solid slurry, they are entrapped me- chanically by primary solid particles. Chemical interactions between the reinforcements and liquid matrix also proceeds with time, and finally the reinforcements are trapped in the composite slurry. Incorporation of the reinforcement particles within a semi-solid alloy is claimed to be advantageous, be- cause the solid mechanically entraps the reinforcement thus avoiding particle agglomeration, and settling or flotation [10]. This mechanical entrapment is promoted by a vigorous agi- tation, which has also been seen to promote wetting. As the reinforcement is forced into the matrix, it is essential to con- tinue mixing after the particulate addition, in order to ensure proper interface bonding between the matrix and particulate [11].

Although compocasting is generally accepted as a com- mercial route for the production of MMCs [12], there are however technical challenges associates with producing a ho- mogeneous high density composite. In order to achieve the optimum MMC properties, the distribution of the reinforce- ment materials in the matrix alloy must be uniform, and the wettability of bonding between these two substances should be optimized [13]. The mechanical properties of MMCs are controlled to a large extent by the structure and properties of the reinforcement metal interface [13,14]. It is believed that a strong interface permits transfer and distribution of load from the matrix to the reinforcement, resulting in an increased elas- tic modulus and strength. Porosity, as is well known, is one of the biggest problems in the production of premium quality aluminium castings; it is always a case for concern because, in addition to affecting the surface finish, its presence can be detrimental to the mechanical properties [15] and corrosion resistance [16] of the casting. Porosity levels must, therefore, be kept to a minimum in order to produce sound casting with optimum properties. Porosity cannot be fully avoided during the casting process, and so the mechanical properties of cast materials are commonly correlated to the volume fraction of its porosity [17]. For MMCs, excessive cavities are expected to be formed at the interfaces and to give rise to larger dam-

age. The ability of pores to form is, however, much reduced in the semi-solid alloy compared to the liquid processed alloy.

Particle distribution in the matrix material during the melt stage of the casting process depends strongly on the viscos- ity of the slurry, particle wetting, how the characteristics of the reinforcement particles influence settling rate, the effec- tiveness of the mixing and breaking up of agglomerates, and the minimizing of gas entrapment. The uniformity of particle dispersion in a melt before solidification is also controlled by the dynamics of the particle movement in an agitated ves- sel. One significant requirement when using a stir casting technique is the continuous stirring of the melt with a motor driven agitator to prevent settling of particles. If the particles are more dense than the host alloy, they will naturally sink to the bottom of the melt [18]. The vortex method for partic- ulate entrapment was the most frequently used [3,10,15] in previous studies since any stirring of a melt naturally results in a formation of a vortex. During these works stirrer speeds ranging from 100 to 1500 rpm were investigated [1,3,10,19]. Ceramic particles were introduced into the vortex which was created in the melt with a mechanical impeller.

The reinforcement particles can also be pre-treated by heat-treating (artificially oxidized). For SiC oxidation, dif- ferent workers have used varying temperatures and times: 1000 ◦C for 1.5 h in air [3], 1100 ◦C for 1–3 h [20], or 1.5 h [21], 850 ◦C for 8 h and 1200 ◦C for 1 h [22]. Additionally, gas absorbed on the surface of SiC, which was prepared in air, can be removed by preheating at a certain temperature and for a certain period of time. For example, particles have been heated to 554 ◦C for 1 h, or at the temperatures of 900 ◦C [23], 799 ◦C [1,15] and 1100 ◦C [24] for 8 h. 2. Experimental 2.1. Stir caster design

In this study, a new quick quenched stir caster was de- signed to fabricate MMC ingot. This device was developed from a previous compocaster unit [25]. The schematic draw- ing, of the stir caster used in this work is shown in Fig. 1. The stir-caster furnace was mounted on four legs. This in turn was attached to a welded steel table to which a screw driven actu- ator lift was bolted vertically underneath. A ceramic spacer was used to separate the crucible from the actuator. This ar- rangement allowed rapid extraction of the crucible from the furnace. The temperature within the stir-caster had to be pre- cisely measured and accurately controlled (±1 ◦C), in order to control the fraction solid of the semi-solid alloy. One cal- ibrated K type mineral insulated thermocouple provided the control input to a Eurotherm 2216 PID temperature controller. This thermocouple was placed 80 mm above the surface of the melt. Data logger facilities were used to record the tem- perature at a rate of 1 Hz.

Stainless steel (Ovar Supreme™, Uddeholm) material was chosen for the crucible, stirrer rod and impeller material.

S. Naher et al. / Journal of Materials Processing Technology 166 (2004) 430–439

1 200 16 1 2 300 16 1 3 200 2335 300 4 200 1030 300 5 300 1030 300 6 300 540 300 7 500 540 300 8 500 120 300

Fig. 1. Schematic drawing of the stir-caster: (1) the stirring motor, (2) the stirrer rod, (3) furnace, (4) stirrer impeller, (5) crucible, and (6) actuator.

This material was chosen because of it corrosion resistance.

and to facilitate the stirrer positioning, cleaning and replace- ment. A Servomech screw driven thread linear actuator with a stroke length of 273 mm (ATL101) was used to allow quick extraction of the crucible from the furnace. This required a 70 W, 24 V power supply and was capable of a linear speed of 93 mm/s. A free rotational bearing on top of the actuator plate prevented rotation of the crucible. 2.2. Stir-caster operation

When setting up the stir-caster before an experiment, the crucible was charged with aluminium and SiC particles. Then it was attached to the ceramic spacer on top of the actuator. A locking mechanism was engaged to ensure the height and lateral position of the crucible remained constant throughout the tests. The temperature was then raised to 630 ◦C to melt the charge. A continuous purge of nitrogen gas was used in order to minimize high temperature oxidation problems. When the metal was fully melted, the stirrer was lowered into the crucible and pushed into contact with a bearing pin at the base of the crucible. This ensured that the height of the stirrer off the base (12 mm) was consistent throughout the tests and that it was held concentrically. All the experimental parameters are mentioned in Table 1. The stir caster operation was conducted according to the following sequence of events: (1) Collection and preparation of the raw materials (A356 +

SiC particles). (2) Placing raw materials in the crucible under nitrogen gas

into a furnace. (3) Heating the crucible above the liquidus of A356 and al-

lowing time to become completely liquid. (4) During cooling stirring is started at the semi-solid con-

dition then when the temperature was stabilised at the appropriate level the stirring was recommenced for the specified period and shear rate.

(5) The charge in the crucible was then quenched into water. (6) MMC billets were produced.

The charge temperature was raised to 630 ◦C within 130 min at the start of each experiment. Stirring was then started and continued for the specified periods for the liq- uid state experiments. After the specified shearing periods the crucible was lowered and quenched. In the semi-solid Table 1 The experimental parameters

The crucible material also allowed the melt to be quickly quenched. The diameter of the crucible was 105 mm and the melt height was 65 mm. During experimental work, a four flat bladed 45◦ angled stirrer was chosen. The stirrer impeller diameter was 80 mm, blade height was 20 mm and width was 2 mm. The stirrer was connected to a DC motor which was used to stir the molten matrix material. A lifting mechanism for the rotational drive unit and stirrer assembly was used to extract the stirrer from the melt before quenching of the melt

Experiment number

Stirring speed (rpm)

Stirring time (s) Approximate viscosity ( mPa s)

S. Naher et al. / Journal of Materials Processing Technology 166 (2004) 430–439

Table 2 Composition of A356 aluminium alloy

Si 7.23 Cu 0.127 Mg 0.408 Fe 0.355 Mn 0.135 Ti 0.113 Ni 0.012 Zn 0.059 Pb 0.01 Sn 0.011 Al Balance

experiments, the stirring was also started at 630 ◦C and con- tinued for 5 min to promote the wettability. Stirring was then stopped and the temperature was lowered to 605 ◦C (0.30 fraction of solid), at a rate of 0.0011 ◦C/s. The stirring was then recommenced at the same shear rate. This corresponded to a viscosity of about 300 mPa s in the semi-solid metal [26]. After shearing the alloy at a specified shear rate and for a specified length of time, the stirring was stopped, the stirrer raised, the lock on the actuator released, the crucible con- taining melt was lowered immediately, lifted off the ceramic spacer with steel tongs, and quenched into water. The time be- tween the end of shearing and quenching was approximately 10 s.

2.3. Investigated materials

A356 alloy and silicon carbide were chosen for experi-

mental testing in this work. A356 was used as it provided clear microstructural detail and is of great commercial inter- est. A356 was also highlighted by previous researchers as an alloy which could be easily stir cast [27,28]. More recently, it has been shown that it is possible to successfully thixo- form A356 alloy [29,30]. Certificates of composition were supplied for the A356 ingots purchased. The composition of the A356 aluminium ingots can be seen in Table 2.

The silicon carbide used was of 320 mesh size (30 J.m average diameter). A typical composition of the silicon car- bide used can be seen in Table 3. The weighing of the matrix alloy and silicon carbide were carried out using an analytical balance with an accuracy of 0.01 grams. The particles were preheated to 900 ◦C for 4 h. At volume percentage of SiC particles higher than 10% in the matrix alloy the wettability decreases and the agglomeration and settling tendencies in- crease [31]. Therefore, 10 vol.% of SiC was chosen for the present research.

Table 3 Typical chemical analysis of SiC

% SiC 99.1 % free C 0.1 % free SiO2 0.4 % free Si 0.2 % free Fe2 O3 0.06

2.4. Metallography

Stir-cast MMC ingots were sectioned with a Buehler Abrasiment 2 cut off machine to produce the samples for met- allographic examination. Emulsion coolant was used to avoid overheating of the ingot samples. Three samples were taken from the bottom, middle and top part of the castings which were 58, 30 and 15 mm, from the top surface respectively. A Struers Prestopress-3 and backelite resine backing were used to mount the specimens and a Struers DAP-V grinding machine was used for the grinding and polishing operations. Silicon carbide paper is often used for grinding of metal but this must be avoided when the MMC is reinforced with SiC particles. This is because the soft matrix will quickly be re- moved whereas the SiC particles will in general remain in- tact. Plane grinding was performed with a Struers TEXMET grinding disc with 30 J.m diamond suspended in water lu- bricant. This grinding was done manually and light pressure was applied. The plane grinding took about 5-10 min. This was followed by polishing using the TEXMET grinding disc with 9, 6, 3 and 1 J.m diamond suspended in water as a lubri- cant. An ‘Reichert MeF2’ optical microscope and a digital camera, attached to a PC, were used for image capture and image analysis (IA). 2.5. Image analysis

Liquid and semi-solid stir-cast microstructures were in- vestigated by image analysis (IA) techniques to determine the SiC area percentage, count, perimeter (J.m), aspect ratio and sphericity. Cross section examination areas with a size of 1.26 mm2 were used to perform IA on each cast sample. An edge detect image processing filter was used to capture each particle and approximate the total number. This included all SiC particles within agglomerates as well as unagglomer- ated isolated particles. SiC particles that did not connect with neighbouring particles or porosity were analysed as ‘isolated particles’.

The area percentage of SiC was measurement as the area of a particular microstructure picture divided by the area of the SiC represented in that picture. The sphericity of each SiC particle in a selected microstructure picture was measured by a dimensionless parameter in the range of zero to one, according to [26]. If the shape of the particle is a perfect circle, the sphericity of this particle is one. The more irregular the particle shape is, the lower sphericity value. 3. Results

Fig. 2 shows some of the resulting compocastings. In Fig. 2(a), an impression of the four bladed stirrer was evident at the top surface of the semi-solid compocast produced at ∼0.5 fractional solid. The stirrer was, however, easy to remove from the semi-solid processed system due to the reduced vis- cosity of the sheared semi-solid material. In Fig. 2(b), there is

S. Naher et al. / Journal of Materials Processing Technology 166 (2004) 430–439

Fig. 2. Top view of the compocasting: (a) semi-solid casting and (b) liquid state casting. Bottom and side view: (c) semi-solid casting and (d) liquid state casting.

no stirrer impression as this was processed in the liquid state, however, a solidification pipe resulted. A smooth external ap- pearance was evident on the bottom of the semi-solid com- pocast billet at ∼0.3 fractional solid, see Fig. 2(c), whereas a non-smooth surface was found on the bottom where much of the SiC collected for the liquid compocast billet (Fig. 2(d)).

3.1. Metallography

The microstructure of composites produced are presented

in Figs. 3–6. These microstructures were obtained from the compocasting performed according to the processing vari- ables outlined in Table 1. Micrographs from the liquid state castings are shown in Fig. 3 and those from the semi-solid state castings are in Figs. 4–6. The main effects of the pro- cessing parameters on the microstructure evolution are seen from a qualitative examination of the presented micrographs. A range of SiC distribution was observed in these microstruc- tures depending upon the chosen compocasting state and pa- rameters. A small dendritic primary aluminium phase was observed in liquid state processed compocasting and a larger primary aluminium phase was observed for semi-solid state

processing. These micrographs represent typical examina- tion areas used for IA of the alloys. From microstructural image analysis of the metallographic samples from experi- ments 1 and two, it was found that there were only small sized and volume fraction of SiC particles present. During metal- lographic examination of the sample from experiment 2, it was noted that there was much gas porosity throughout the specimen. The porosity was more prominent at the top part of the specimen and SiC particle clustering arround these pores was evident. The porosity content was higher in the liquid samples when compared to the semi-solid samples due to the vortex and turbulent nature of the liquid metal flow during the experiments. 3.2. Image analysis

The average area percentage of SiC determined in the in- dicated parts of the compocasting are shown in Table 4. Var- ious average IA parameters measured for the SiC at different position are shown in Table 5. Table 1 lists the processing pa- rameters for each experiment in Tables 4 and 5. From Table 4 it is clear that experiments 5 and 7 resulted in the highest percentage and hence distribution of SiC content. This Ta-

S. Naher et al. / Journal of Materials Processing Technology 166 (2004) 430–439

number Bottom part

Middle part

Top part

Experiment number

Count Perimeter (J.m)

Aspect ratio Sphericity

1 6.1 6.3 5.3 1 87 4.9 2.03 0.35 2 4.1 3.8 3.3 2 51 3.01 2.01 0.35 3 7.9 7.1 7.4 3 103 7.54 1.85 0.56 4 3.6 3.6 2.8 4 49 3.32 1.68 0.54 5 10.6 11.9 11.7 5 116 12.09 1.88 0.57 6 5.9 6.9 4.8 6 100 6.4 1.85 0.51 7 10.6 11.6 9.9 7 115 10.2 1.86 0.50 8 1.6 1.4 1.1 8 14 1.01 1.87 0.53

Fig. 3. (a) Micrograph from experiment 1. Processing parameters: stirring speed, 200 rpm; stirring time, 16 s; viscosity, 1 mPa s. (b) Micrograph from experiment 2. Processing parameters: stirring speed, 300 rpm; stirring time, 16 s; viscosity, 1 mPa s.

ble also indicates that the samples from experiments 1 and 2 contain less volume fraction of SiC than the other samples. It can be seen from this that semi-solid processing of Al–SiC, was better for retaining the SiC particulate than liquid state processing. It is also noted that experiment 2 contains less

Table 4 IA results showing the area percentage of SiC for different experiments and casting position

SiC particles than experiment 1. It is believed that the higher stirring speed of 300 rpm in experiment 2 caused excessive turbulence which resulted in the poorer SiC distribution.

These results also gives one an idea about the effect of stirring time at a particular stirring speed. For example, cast-

Table 5 IA results showing the SiC count, perimeter, aspect ratio, and sphericity

Experiment Area percentage of SiC

S. Naher et al. / Journal of Materials Processing Technology 166 (2004) 430–439

Fig. 4. (a) Micrograph from experiment 3. Processing parameters: stirring speed, 200 rpm; stirring time, 2335 s; viscosity, 300 mPa s. (b) Micrograph from experiment 4. Processing parameters: stirring speed, 200 rpm; stirring time, 1030 s; viscosity, 300 mPa s.

ing number three was processed at 200 rpm, 300 mPa s and 2335 s before quenching whereas casting four was processed for approximately half the period at 200 rpm and 300 mPa s. The higher distribution percentage for casting three can, therefore, be seen to be caused by the extended process- ing period. From experiment 3, it can also be noted that the 200 rpm stirring velocity was not enough at 300 mPa s vis- cosity to fully disperse the SiC particle from the bottom of the crucible.

The SiC count was found to be at a minimum for ex- periment 8 and highest for experiment 5. This was due to the shorter shearing period for experiment 8. The maximum perimeter was found for experiment 5 and minimum perime- ter was for experiment 8. This is due to smaller particles rising more quickly in the early stages of shearing when compared with the larger particles. These latter two results correspond with the % of SiC present.

For liquid state processing, the aspect ratio was found to be approximately two whereas for SSP the aspect ratio of the SiC particles ranges from 1.68 to 1.88. The SiC particles en- trapped in the liquid state processing, were less spherical than in the semi-solid state. This is also shown by the sphericity results. The more spherical nature of the SiC particles from the semi-solid compocastings is possibly due to the additional drag produced by irregularly shaped particles with relatively large surface area, compared with the quicker settling of the spherical particles in the liquid state. 4. Discussion

In the liquid state experiments all the grain sizes were very small, see Fig. 3, and it is clear that these were quick quenched castings. The aluminium crystals were larger for the semi-

S. Naher et al. / Journal of Materials Processing Technology 166 (2004) 430–439

Fig. 5. (a) Micrograph from experiment 5. Processing parameters: stirring speed, 300 rpm; stirring time, 1030 s; viscosity, 300 mPa s. (b) Micrograph from experiment 6. Processing parameters: stirring speed, 300 rpm; stirring time, 540 s; viscosity 300 mPa s.

solid casting, see Fig. 4. As a quick quenched structure, so- lidification time even for the liquid compocastings was very short. It would, therefore, seem that the large primary alu- minium grains were formed during the stirring period. These grains were globular rather than dendritic due to the stirring action and holding period. Shear rate and period would be ex- pected to affect the grain size. A higher shear rate and shorter period during processing should produce smaller grains [32]. In this experimental work, at higher shear rates (500 rpm), smaller grains were observed, see Fig. 6(b) (∼40J.m) versus Fig. 4(a), 5(a) and 6(a) (∼110J.m).

From the metallography, it was clear that liquid state samples contained much more porosity than the semi-solid state, compare Fig. 3 with Figs. 4–6. It is also noticeable that the produced samples with 300 rpm stirring speed in experiment 2 (Fig. 3(b)) generated more porosity than the sample produced at 200 rpm stirring speed, experiment 1 (Fig. 3(a)).

From experiment 3, it was found that even at 200 rpm and after 2335 s, see Table 4, not all the SiC was dispersed from the bottom of the tank. Only an average of 7.5% SiC was recorded through the casting instead of 10%. At approxi- mately half this stirring period but for the same other process- ing parameters, only 3.7% SiC was dispersed, see Table 4. The viscosity of the melt generally increases with fractional solid. The incorporation of the reinforcement particles will immediately increase the viscosity of the matrix melt due to the increased solid content [32]. The viscosity reduces as the shear rate (stirring speed) increases. At higher shear rates, the clusters of SiC particles are broken, reducing the resis- tance to flow. Higher viscosity helps to enhance the stability of the slurry by reducing the settling velocity, but also cre- ates resistance to flow in mould channels during casting [33]. After the incorporation of the particles in the melt is com- pleted (in the case in which the stirring action was preformed in semi-solid condition) the slurry is usually remelted to a

S. Naher et al. / Journal of Materials Processing Technology 166 (2004) 430–439

Fig. 6. (a) Micrograph from experiment 7. Processing parameters: stirring speed, 500 rpm; stirring time, 540 s; viscosity 300 mPa s. (b) Micrograph from experiment 8. Processing parameters: stirring speed, 500 rpm; stirring time, 120 s; viscosity, 300 mPa s.

temperature above the liquidus before being poured into the mould. This might cause settling of the ceramics particles.

5. Conclusions

• A novel quick quenched stir-caster has been designed and built for processing Al–SiC composites in liquid and semi- solid state. Temperature controlled compocasting experi- ments were performed on Al–10% SiC.

• Stirring the MMC slurry in semi-solid state, during the so- lidification process, helps to incorporate ceramic particles into the alloy matrix.

• The quick quenched compocasting method was found to be successful to fabricate Al–SiC metal matrix composite.

• One of the successes of the present study lies in the fact that without any addition of a wetting agent, a full incor- poration of SiC particles into the aluminium matrix was obtained in the semi-solid state.

References

[1] P.K. Ghosh, S. Ray, Fabrication and properties of compocast aluminium-aluminiuma particulate composites, Indian J. Technol. 26 (1988) 83–94.

[2] Y.H. Seo, C.G. Kang, Effects of hot extrusion through a curved die on the mechanical properties of SiCp /Al composites fabri- cated by melt-stirring, Composites Sci. Technol. 59 (1999) 643– 654.

[3] M. Yilmaz, S. Altintas, Properties of Al–Mg–SiC composites produced by a modified compocasting technique, in: Proceedings of the 2nd Bienniel European joint conference on Engineering Systems, London, UK, 1994, pp. 119–124.

[4] A. Vogel, R.D. Doherty, B. Cantor, Stir-cast microstructures and slow crack growth, in: Proceedings of the Conference on Solidification and Casting of Metals, Sheffield, July 18–21, 1977.

[5] M. Skibo, P.L. Morris, D.J. Lloyd, Structure and properties of liq- uid metal processed SiC reinforced aluminium, in: Proceedings of the World Materials Congress, Chicago, 1988, pp. 257–262.

[6] B. Ralph, H.C. Yuen, W.B. Lee, The processing of metal matrix composites—an overview, J. Mater. Process. Technol. 63 (1997) 339– 353.

S. Naher et al. / Journal of Materials Processing Technology 166 (2004) 430–439

[7] M.K. Surappa, Microstructure evolution during solidification of DR- MMCs (discontinuously reinforced metal matrix composites): state- of-art, J. Mater. Process. Technol. 63 (1997) 325–333.

[8] F. Bonollo, R. Guerriero, E. Sentimanti, I. Tangerini, V.L. Yang, The effect of quenching on the mechanical properties of powder metallur- gically produced Al–SiC (particles) metal matrix composites, Mater. Sci. Eng. A144 (1991) 303–309.

[9] M.C. Flemings, Behaviour of metal alloys in the semi-solid state, Met- all. Trans. 22 (1991) 957–981.

[10] J.W. McCoy, C. Jones, F.E. Wawner, Preparation and properties of cast ceramic/aluminium composites, SAMPE Q. 19 (1988) 37–50.

[11] S. Caron, J. Masounave, A literature review on fabrication techniques of particulate reinforced metal composites, in: J. Masounave, F.G. Hamel (Eds.), Fabrication of Particulates Reinforced Metal Composites, ASM International, Montreal, Que., Canada, September 1990, pp. 79–86.

[12] J.T. Lin, D. Bhattachaaryya, C. Lane, Machinability of a silicon carbide reinforced aluminum metal matrix composite, Wear 181–183 (1995) 883–888.

[13] S.Y. Oh, J.A. Cornie, K.C. Russell, Wetting of ceramic particle with liquid aluminium alloys. Part II. Study of wettability, Metall. Trans. A20 (1989) 533–541.

[14] M. Taya, R.J. Arsenault, Metal Matrix Composites—Thermo Mechan- ical Behaviour, Pergamon, Oxford, 1989.

[15] P.K. Ghosh, S. Ray, Effect of porosity and alumina content on the high temperature mechanical properties of compocast aluminium alloy- alumina particle composite, J. Mater. Sci. 22 (1987) 4077–4086.

[16] K.A. Lucas, H. Clarke, Corrosion of Aluminium Based Metal Matrix Composites, Research Studies Press Ltd., Taunton, Somerset, England, 1993.

[17] T. Christman, A. Needleman, S. Suresh, An experimental and numeri- cal study of deformation in metal ceramic composites, Acta Metall. 37 (1989) 3029–3050.

[18] S. Ray (Ed.), Casting of composite components, in: Proceedings of the Conference on Inorganic Matrix Composites, Banglore, India, 1996.

[19] R. Warren, C.H. Anderson, Silicon carbide fibres and their potential for use in composite materials. Part II, Composites 15 2 (1984) 101–111.

[20] W. Zhou, Z.M. Xu, Casting of SiC reinforced Metal Matrix Composites, J. Mater. Process. Technol. 63 (1997) 358–363.

[21] W.M. Zhong, G. L’Esperance, M. Suery, Interfacial reaction in Al–Mg (5083)/SiC composite-fabrication and remelting, Metall. Mater. Trans. A26 (1995) 2637–2649.

[22] V.M. Kevorkijan, B. Sustarsic, A new production technology for dis- continuously reinforced Al–SiC composites, Key Eng. Mater. 127–131 (1997) 471–478.

[23] M. Gupta, M.K. Surappa, Processing-microstructure-mechanical prop- erties of Al based Metal Matrix Composites synethesized by casting route, Key Eng. Mater. 104–107 (1995) 259–274.

[24] L. Salvo, G. L’Esperance, M. Suery, J.G. Legoux, Interfacial reac- tion and age hardening in Al–Mg–Si metal matrix composites re- inforced with SiC particles, Mater. Sci. Eng. A177 (1994) 173– 183.

[25] J. Hashim, Production of metal matrix composites using the stir casting technique, Ph.D thesis, Dublin City University, 1999.

[26] D. Brabazon, Processing and properties of rheocast alloys. PhD thesis, UCD, Dublin, Ireland, October 2001.

[27] S. Chu, R. We, The structure and bending properties of squeeze-cast composites of A356 aluminium alloy reinforced with alumina particles, Composites Sci. Technol. 59 (1999) 157–162.

[28] M. Mada, F. Ajersch, Rheological model of semi-solid A356–SiC com- posite alloys. Part 1. Dissociation of agglomerate structures during shear, Mater. Sci. Eng. A212 (1996) 157–170.

[29] T. Witulski, A. Winkelmann, G. Hirt, Thixoforming of amu- minium components for lightweight structures, in: Proceedings of the 4th International Conference on Semi-Solid Processing of Al- loys and Composites, University of Sheffield, UK, 1996, pp. 242– 247.

[30] N.S. Kim, C.G. Kang, An investigation of flow characteristics consid- ering the effect of viscosity variation in the thixoforming process, J. Mater. Process. Technol. 103 (2000) 237–246.

[31] J. Hashim, L. Looney, M.S.J. Hashmi, The enhancement of wettability of SiC particles in cast aluminium matrix composites, J. Mater. Process. Technol. 119 (2001) 329–335.

[32] H.K. Moon Rheological behavior and microstructure of ceramic particulate-aluminium alloy composites, PhD thesis, MIT, Cambridge, MA, September 1990.

[33] H.K. Moon, J.A. Cornie, M.C. Flemings, Rheological behavior and mi- crostructure of ceramic particulate/aluminium alloy composites, in: C. Shi, H. Li, A. Scott (Eds.), The First Pacific Rim International Confer- ence on Advanced Materials and Processing (PRICM-1), The Minerals, Metals and Materials Society, Hangzhou, China, June 23–27, 1992, pp. 669–674.