Read Zhou 1997 - Casting of MMCs text version

W. Zhou, Z.M. Xu / Journal of Materials Processing Technology 63 (1997) 358-363

Casting of SiC Reinforced Metal Matrix Composites

W. Zhou*, Z. M. Xu** * School of Mechanical and Production Engineering Nanyang Technological University, Nanyang Avenue, Singapore 639798 ** Singapore Aerospace Manufacturing Pte. Ltd. 503 Airport Road, Paya Lebar, Singapore 539932

Contact Prof. Wei Zhou for further information through e-mail [email protected]

Abstract Composites based on two aluminium alloys (A536 and 6061) reinforced with 10% or 20% volume fraction of SiC particles were produced by gravity casting and a novel two-step mixing method was applied successfully to improve the wettability and distribution of the particles. The SiC particles were observed to be located predominantly in the interdendrit ic regions, and a thermal lag model is proposed to explain the concentration of particles. It was found that the SiC particles acted as substrates for heterogeneous nucleation of Si crystals in one of the cast composites. This observation can also be explained by the thermal lag model proposed. Keywords: casting, metal matrix composites, wettability, particle distribution, and thermal lag.

1. Introduction Composite materials are continuously displacing traditional engineering materials because of their advantages of high stiffness and strength over homogeneous materials formulations. Composites may have metal or polymer matrices and may be reinforced with continuous fibres, discontinuous fibres, or particles. The present work considered only metal matrix composites (MMCs) reinforced with discrete particles. The development of MMCs has been one of the major innovations in materials in the past two decades [1]. Conventional stir-casting technology has been employed for producing particulate reinforced metal matrix composites (PMMCs) for decades. The casting methods and associated techniques used to fabricate composites based on aluminium alloys have been amply studied [1-7]. The major problem in this technology is to obtain sufficient wetting of disper soid by the liquid metal and to get a homogeneous dispersion of the ceramic particles. Several structural defects such as porosity, particle clusters, oxide inclusions and interfacial reactions were found to arise from the unsatisfactory casting technology [5, 8]. So far, only a few researches have been reported on the successful casting of PMMCs [7, 9], but details of the casting techniques are always considered proprietary and rarely reported by the manufacturers. Among all the liquid-state processes, stircasting technology is considered to be the most potential method for engineering application in terms of production capacity and

cost efficiency. In the present work, a two-step mixing method was tried in hope of producing cast composites with better particle distribution. Microstructures of the cast composites were carefully studied, with special attention to nucleation of Si crystals and solidification process.

2. Experimental Procedure Two SiC particulate reinforced composites were produced by gravity casting. The matrix alloys of the composites were A356 and 6061 respectively. They are both aluminium alloys but differ greatly in amount of Si, as shown in Table 1. The matrix alloys were reinforced by nominally the same SiC particles, which were 25 µ in size and had an aspect ratio of 1.8. Main features of m the particles used are summarized in Table 2. It can be seen from the table that the two composites differed in particle volume fraction by a factor of 2. For convenience in description, the composite with A356 as matrix and reinforced by 10.8 Vol% SiC particles is hereafter referred to as A356-10% SiC, and the other composite as 6061-20%SiC. All the melting was carried out in a clay-graphite crucible in a resistance furnace. Scraps of alloy A356 or 6061 were preheated at 450 °C for 3 to 4 hours before melting, and before mixing the SiC particles were preheated at 1100 make their surfaces oxidized.


for 1 to 3 hours to


W. Zhou, Z.M. Xu / Journal of Materials Processing Technology 63 (1997) 358-363

Table 1 Nominal chemical composition of matrix alloys studied (in wt.% ) Matrix Alloy A356 6061 Si 6.5-7.5 0.4-0.8 Fe 0.20 0.70 Cu 0.20 0.15-0.4 Mn 0.10 0.15 Mg 0.25-0.45 0.8-1.2 Zn 0.10 0.25 Ti 0.20 0.15 Balance Al Al

Table 2 The two composites produced and the main features of the SiC particles used in casting Matrix Alloy A356 6061 Reinforcement SiC SiC Volume Fraction (%) 10.8 20.6 Average Particle Size (µm) 25 25 Aspect Ratio 1.8:1 1.8:1 Interparticle Distance (µm) 76 55

Fig. 1. Plot of SiC area fraction in each measured filed versus serial number of field. The furnace temperature was first raised above the liquidus to melt the alloy scraps completely and was then cooled down just below the liquidus to keep the slurry in a semi-solid state. At this stage the preheated SiC particles were added and mixed manually. Manual mixing was used because it was very difficult to mix using automatic device when the alloy was in a semi-solid state. After sufficient manual mixing was done, the composite slurry was re-heated to a fully liquid state, and then automatic mechanical mixing was carried out for about 20 minutes at an average stirring rate of 150-200 rpm . In the final mixing pro cesses, the furnace temperature was controlled to be within 730 ± 10 °C. The pouring temperature was controlled to be around 720 ° C. A preheated permanent steel m ould with diameters in the range of 10 mm to 18 mm was used to prepare cast bars. The preheating temperature for the mould was either 50 cooling or 350 °C for slower cooling. Metallographic samples were sectioned from the cast bars and was prepared using a technique specially developed for such composites. A 0.5% HF solution was used to etch the samples wherever required. Microstructures were examined on the samples either under the op tical microscope or under a Cambridge 360 scanning electron microscope (SEM) equipped with EDX and WDX. EDX and WDX analyses were performed to identify elements. X-ray diffraction was carried out to identify some phases in the composites using Philips MPD 1880 system. Quantitative measurement of SiC particles and porosity was carried out at a magnification of 200 times in a field area of 1.51141×105 square microns using a Leica Q520 automatic image analyser. For each sample, a total of 20 fields were randomly chosen for the measurement.

3. Experimental Results 3.1. Distribution of SiC particles in cast composites Effort was made to observe distribution of SiC particles in the two cast composites produced. It was found that the par ticles showed a strong tendency to accumulate in the colonies which froze in the last stage of solidification and usually contained eutectic phases. The SiC particles were also observed to be accommodated on the grain boundaries. To characterize the particle distribution quantitatively, values of area fraction of SiC particles were measured in 20 randomly selected fields for each composite and were plotted versus the field serial number in Fig. 1. It can be seen from the figure that scatters in the measured

°C for fast


W. Zhou, Z.M. Xu / Journal of Materials Processing Technology 63 (1997) 358-363

values of area fraction are generally small, indicating a reasonably homogeneous distribution of SiC particles in the cast composites. This quantitative evaluation is consistent with visual assessment under the optical microscope (Figs. 2(a) and 2(b)).

The scatters in SiC area fraction appears to be larger for composite A356-10% SiC than for composite 6061-20%SiC (Fig. 1), but this may be due to the smaller SiC contents in composite A356 -10%SiC. When SiC particles are smaller in number, segregation of a small fraction of the particles may cause apparently larger scatter in SiC area fraction of each measured filed. Microstructu res of the two cast composites were examined carefully and were observed to contain many phases. The main phases are primary -Al, eutectic structures and SiC particles, as shown in Figs. 2(a) and 2(b). In the microstructure o f composite A356 -10%SiC, Si phase was seen apparently in the eutectic regions, and intermetallic compounds such as Mg Si, 2 iron-containing phases or spinels may also exist due to the secondary alloying elements and impurities present in the A356 alloy scraps. Though such phases cannot be seen clearly in Fig. 2(a), they were observed at higher magnifications. Interdendritic microshrinkage and porosity are the main defects observed. These kinds of defects were usually found to be associated with clusters of SiC particles and they were seldom observed in the matrix. The porosity levels of the as -cast composites were measured to be in the range 2% to 4%. These porosity levels are quite high but can be considered to be reasonable since the composite slurries were not degassed and the casting was carried out in the open atmosphere. 3.2. Interfaces between SiC and matrix Since reinforcement-matrix interfaces play an important role in determining mechanical properties of particulate reinforced composites [1, 9, 10], attention was paid to the interfaces between the SiC particles and alloy matrix in the present study. Careful SEM observation of polished and etched A356-10%SiC samples revealed that SiC particles appeared to act as substrates for nucleation of Si phase, as shown in Fig. 3. To study the phenomenon further, some of the samples were deeply etched and then observed under the SEM. The etchant was found to attack the eutectic -Al preferentially, making it possible to observe Si crystals clearly. Many Si crystals were observed to be attached to the surface of SiC particles, further supporting the observation that Si phase nucleated heterogeneously from the SiC particles. Reaction products were occasionally observed at particlematrix interfaces. Some of them were found by EDX analyses to be rich in both O and Mg. This might suggest the existence of spinels (either MgAl2O 4 or MgO). From thermodynamic considerations the formation of MgAl2O4 is more likely. Indeed, existence of MgAl2O4 at interfaces was confirmed in a detailed TEM study [11]. The spinel compound may result from the following two possible reactions: 2SiO2+2Al+MgMgAl2O 4 +2Si 3Mg+4Al2O32Al+3MgAl2O4 (1) (2)


(b) Fig. 2. Microstructures of cast composites A356-10%SiC (a) and 6061-20%SiC (b) showing homogeneous SiC distribution.


W. Zhou, Z.M. Xu / Journal of Materials Processing Technology 63 (1997) 358-363

Fig. 3. Optical micrograph showing nucleation of Si crystals (some of them are pointed) on SiC particles (the dark ones).

studied the problem of poor wetting could not be solved by mechanical stirring in a completely liquid state. Mechanical stirring could indeed mix the particles into the melt, but when stirring stopped, the particles tended to return to the surface. Most of these particles still stuck to one another to remain in clusters. It is not surprising for these clusters to resurface because it might be argued that pores could exist in them to make them float. However, the fact that single particles also tended to return to the surface strongly indicates that the particles floated mainly because of the surface gas layers surrounding them. The gas layers might be the main factor for the poor wettability. First, gas layers can cause the buoyant migration of particles, making it difficult to incorporate the particles into melts. Secondly, even the particles can be suspended in the melts by vigorous agitation, it is still difficult for the particles to be wetted by the molten metals because of the gas layers. The above analysis leads to the conclusion that it is necessary to break the gas layers in order to achieve good wettability. Single particles and particle clusters can flow easily in a completely liquid melt, therefore, no large mechanical forces are actually applied to the particles during agitation, making it very difficult to break the gas layers simply by stirring in the conventional way. A two-step mixing method (as described in Section 2) was thus tried and was found to be effective. In a semi-solid state, primary -Al phase exists, so agitation can apply large forces on the SiC particles through abrasion and collision bet ween the primary -Al nuclei and particles. This process can help to break the gas layers and perhaps oxide layers as well and to spread the liquid metal onto surfaces of the particles, thus helping to achieve good wettability. It was found that cast composites with up to 25 Vol% particles could be obtained using this method. Mechanisms of pseudoplastic behaviour of composite slurries in both semi-solid and liquid states were studied [12-15] but have not been clearly understood. The advantages of using semisolid slurries are usually considered to be the increase in the apparent viscosity and the prevention of the buoyant migration of particles [12-15]. In the present study, the breaking of particle-surface gas layers is emphasiz ed. When the gas layers are broken and the particles are wetted, the particles will tend to sink to the bottom (due to higher specific weight) rather than float to the surface. However, this does not ensure a uniform particle distribution. To improve the particle distribution, the second mixing step is needed, i.e., to heat the slurry to a temperature above the liquidus and then to stir the melts using an automatic device. It was found that the two-step method resulted in a homogeneous distribution of part icles. The two-step mixing method is not the only effective one. It can be seen from the above analysis that any processes can be effective if they can break the gas layers surrounding the particles. Other methods such as bottom-mixing process and particle-injection method were tried by other researchers..

SiO 2 was formed when SiC particles were pre-oxidized, and Al 2O 3 was easily available from the oxide films on the melt surface. At present it can not concluded which reaction played a dominant role in resulting the MgAl 2O 4 spinel. With some difficulty Al4C3 was detected by X-ray diffraction in both cast composites. It has been widely reported [1, 5, 8] that Al4C3 can form due to the following reaction: 3SiC+4Al Al4C3 +3Si (3)

This reaction is always considered to be undesirable and detrimental to SiC reinforced composites based on aluminium alloys. Fortunately, results of X-ray diffraction suggested that the reaction product Al4C3 was very small in amount and could only be discerned occasionally.

4. Discussion 4.1. Advantage of two-step mixing In the present study, when the SiC particles were added into the molten alloys, they were observed to be floating on the surface, though they have a larger specific density than the molten alloys. This was due to high surface tension and poor wetting between the particles and the melt. In fact, wettability between most ceramic particles and liquid metals is poor. A mechanical force can usually be applied to overcome surface tension to improve wettability. However, for the composites


W. Zhou, Z.M. Xu / Journal of Materials Processing Technology 63 (1997) 358-363

4.2. Interactions between particles and solidification front The main microstructural features in both cast composites are in the form of dendrites. During growth of the dendrites, the freely suspended SiC particles in the melts could be either entrapped by the dendritic front or pushed ahead by the front, depending on the velocity of the growing front and geometrical compatibility between the dendrite arm spacing (DAS) and particle sizes. The entrapment models are generally applicable to planar front solidification of single phase systems [16], whereas alloys used in practice always have dendritic solidification fronts and multiple phases. For dendritic solidification, solidification rate can be related to DAS by the following expression [1]:

= KR -n


where is average secondary dendrite arm spacing in µm, R is solidification rate in °C/s , K is a constant (which may depend on the temperature gradient), and n is an exponent ranging from 0.3 to 0.4. The relationship is generally considered to be independent of alloy composition, so it was used in the present study to estimate solidification rates from values of DAS. The values of DAS for the two cast composites were measured and are listed in Table 3, together with the estimated solidification rates. Two different preheating temperatures of the mould resulted in two different cooling rates and thus led to different values of DAS (Table 3). Table 3 Variation in average dendrite arm spacing (DAS) with mould temperature (MT) and solidification rate (SR) Composite A356-10%SiC 6061-20%SiC M T50°C SR 40-50°C/s DAS=12.2 µm NA M T 350 °C SR20-25 °C/s DAS=24.6 µm DAS=18.4 µm

during solidification regardless of the dendritic arm spacing. Pushing of particles by dendrite fronts could almost certainly occur if they were not entrapped; however, the trapping model alone cannot account for all the phenomena observed in the present study. The particles in the cast composites are generally large (they differ in size and some of them are larger than 30 µm) and they were observed to be accumulated in interdendritic regions in large numbers. Though pushing of the particles by dendrite fronts was unavoidable, it is not plausible to assume that the concentration of so many large particles was solely due to pushing by dendrites, especially for the steel mould casting. The steel mold used was preheated at either 50 °C or 350 °C, but the cooling rates were still high (see Table 3). Since the solidification rates were high, there would not be sufficient time for the relatively small dendritic arms (Table 3) to move the large particles for a long distance. The above analysis suggests that there might be other reasons for the interdendritic accumulation of particles. A possible reason may be the thermal lag in the particles during solidification. The thermal conductivity ( SiC and Al) and heat diffusivity (K SiC and K Al) of SiC and aluminium are compared as follows [18]:

SiC = 0.2 Jcm-1 S-1 K-1 < Al = 0.96 Jcm-1 S-1 K-1 -2 -1 -1/2 -2 -1 -1/2 KSiC = 0.84 Jcm K S < K Al = 2.42 Jcm K S

(5) (6)

Obviously, SiC particles have a lower thermal conductivity and heat diffusivity than those of aluminium melt. Therefore, SiC particles are unable to cool down as fast as the melt. As a result, the temperature of the particles is somewhat higher than the liquid alloy. The hotter particles may heat up the liquid in their immediate surroundings, and thus delay solidification of the surrounding liquid alloy. According to this analysis, nucleation of

-Al phase starts in the liquid alloy at a distance away from the particles, where the temperature is lower. The growth of the Al nuclei will lead to enrichment of Si (in composite A35610%SiC) and other solutes in the remaining melt. In composite A356-10%SiC, Si crystals were observed to nucleate from SiC particles (Fig. 3). This heterogeneous nucleation can be explained as the result of enrichment of Si in the melt around the particles. Another effect of thermal lag is that the melt around the particles will solidify in the last stage. This will make the particles located between dendrites. In other words, the interdendritic clusters of SiC particles are partly inherited from inhomogeneous distribution of particles in original slurries. In cast A356 alloy ingot (without SiC particles), dendrites were observed to be distinctively columnar and almost randomly distrib uted; however, in cast composite A356-10%SiC, dendrites were found to be equiaxed and in the regions with clusters of SiC particles the primary that introduction of SiC particles caused refinement of -Al grains can be explained based upon the thermal lag model

Rohatgi et al. [17] and Lloyd [5] related movement of SiC particles to values of DAS. According to their researches, the particles will be incorporated into the solids by geometrical trapping between converging dendrites only when values of DAS are approximately the same as the particle sizes; and the particles will be pushed rather than trapped if values of DAS are either less or much larger than the particle sizes. For composite A356-10%SiC cooled at a rate of 20-25 °C/s, the value of DAS is 24.6 µm (Table 3), which is very close to the average SiC particle size of 25 µm (Table 2). According to the above model, the SiC particles in this composite would be expected to be incorporated into the solid by geometrical trapping between converging dendrites. However, this was found not to be the case. In the microstructures produced in the study, SiC particles were generally observed to be accumulated in the interdendritic regions (Fig. 2), and geometrical trapping by dendrites was rarely observed. This observation seems to suggest that the SiC particles were always pushed by dendrite fronts

-Al grains seemed to be finer. The fact


W. Zhou, Z.M. Xu / Journal of Materials Processing Technology 63 (1997) 358-363

proposed. When a dendrite front grows near a particle, the dendrite would grow larger if it can push the particle forward. According to the above analy sis, particle pushing could not be the dominant growing mechanism, so the particle can be assumed to act as a barrier to the growth. Therefore, the dendrite front will be split to avoid the barrier, as illustrated schematically in Fig. 4.

SiC particles were observed to be located predominantly in interdendritic regions. However, the concentration of particles cannot be explained solely by particle pushing model and "thermal lag" of SiC particles is considered to play an important role in the solidification process. SiC particles were observed to act as substrates for heterogeneous nucleation of Si crystals in one of the composites produced (A356-10%SiC). This observation can be explained satisfactorily by the thermal lag model proposed. The interfaces between SiC particles and matrix were studied and interface reaction products such as Al4C3 were detected experimentally.

6. References [1] [2] D.J. Lloyd, Inter. Mater. Rev., 39 (1994) 1-23. M.A. Bayoumi and M. Suery, in Proc. of Inter. Symposium on Advances in Cast Reinforced Metal Composites (S.G. Fishman and A.K. Dhingra, eds.), Materials Park, OH: ASM International Publication, (1988) 167-172. T.Z. Kattamis and J.A. Cornie, ibid, 47-51. P.K. Rohatgi, R. Asthana and F. Yarandi, in Solidification of Metal Matrix Composites , (P.K. Rohatgi ed.), TMSASM Committee, TMS Publication, (1989) 51 -76. D.J. Lloyd, Compos. Sci. Technol., 35 (1989) 159-179. A. Kolsgaard and S. Brusethaug, Mater. Sci. & Technol., 10 (1994) 545-551. A Luo, Metall. & Mater. Trans. A 26A (1995) 2445, 2455. A.M. Samuel, H. Liu and F.H. Samuel, Compo. Sci. Technol., 49 (1993) 1-12 M.D. Skibo and D.M. Schuster, US Patent No 4 786 467, (1988). W. Zhou, N.L. Loh and N.P. Hung, Proc. of 2nd Inter. Conf. on Compo. Eng. (D. Hui ed.), New Orleans, (1995) 875-876. N. Wang, Z. Wang and G.C. Weatherly, Metall. Trans. A, 23A (1992) 1423-1430. R. Mehrabian, A. Sato and M.C. Flemings, Light Metal, 38 (1975) 177-193. P.R. Gibson, A.J. Clegg and A.A. Das, Foundry Trade J., Feb. 1982, 253-263. H.K. Moon, J.A. Cornie and M.C. Flemings, Mater. Sci. & Eng. A, A144 (1991) 253-265. M.C. Flemings, Metall. Trans., 22A (1991) 957-981. R. Asthana and S.N. Tewari, Proc. Adv. Mater., 3 (1993) 163-180. P.K. Rohatgi, F.M. Yarandi and Y. Liu, in Proc. of Inter. Symposium on Advances in Cast Reinforced Metal Composites (S.G. Fishman and A.K. Dhingra, eds.), Materials Park, OH: ASM International Publication, (1988) 249. J.W. McCoy and F.E. Wawner, ibid, 237-242.

[3] [4]

[5] [6] [7] [8] [9] [10] Fig. 4. Schematic illustration of dendrite splitting (or branching) induced by foreign particles. The occurrence of splitting is mainly owing to the role of particle as physical barrier to solute diffusion field ahead of the growth front. The dendrite front may also be broken upon touching the barrier because of mechanical force. Both splitting and breaking of dendrites can lead to refinement of grains. The existence of SiC particles can result in instability in the growth front. 5. Summary and Conclusions Composites of two different matrix alloys (A356 and 6061) reinforced with 10% or 20% volume fraction of SiC particles were produced by gravity casting. A two-step mixing method was tried and found to improve the wettability of the SiC particles and ensure a good particle distribution. The mechanisms for the improvement in wettability and particle distribution are discussed in the paper.

[11] [12] [13] [14] [15] [16] [17]




Zhou 1997 - Casting of MMCs

6 pages

Report File (DMCA)

Our content is added by our users. We aim to remove reported files within 1 working day. Please use this link to notify us:

Report this file as copyright or inappropriate


You might also be interested in