Sci Eng Compos Mater 2016; 23(1): 1–20

Review

Khushbu Dash*, Suvin Sukumaran and Bankim C. Ray The behaviour of aluminium matrix composites under thermal stresses

Abstract: The present review work elaborates the behav- 1 Introduction iour of aluminium matrix composites (AMCs) under vari- ous kinds of thermal stresses. AMCs find a number of There has been a rapid growth in the development of applications such as automobile brake systems, cryostats, metal matrix composites (MMCs) since the 1960s due to microprocessor lids, space structures, rocket turbine hous- a variety of properties it offered, such as high specific ing, and fan exit guide vanes in gas turbine engines. These strength, high specific stiffness, low density, good ele- applications require operation at varying temperature vated temperature stability, and good wear resistance [1]. conditions ranging from high to cryogenic temperatures. Their commercial availability has also been increasing The main objective of this paper was to understand the steadily, and as a result, information about the thermal behaviour of AMCs during thermal cycling, under induced properties of these materials is essential for the proper thermal stresses and thermal fatigue. It also focuses on designing of MMC applications [2]. Among the MMCs, the various thermal properties of AMCs such as thermal aluminium matrix composites (AMCs) have drawn con- conductivity and coefficient of thermal expansion (CTE). siderable attention in the academic and industrial sectors CTE mismatch between the reinforcement phase and the recently. Improvement in strength and modulus has been aluminium matrix results in the generation of residual reported for AMCs over conventional metal alloys. This, thermal stress by virtue of fabrication. These thermal along with thermal stability, makes them a good choice for stresses increase with increasing volume fraction of the use as materials in aerospace and transport industries [3]. reinforcement and decrease with increasing interparticle The main property that makes MMCs unique is that they spacing. Thermal cycling enhances plasticity at the inter- offer a large variety of properties by means of combin- face, resulting in deformation at stresses much lower than ing possible combinations of matrices and the reinforce- their yield stress. Low and stable CTE can be achieved ments, which in turn enables them to adapt the material by increasing the volume fraction of the reinforcement. properties according to the application requirements [4]. The thermal fatigue resistance of AMC can be increased Using aluminium alloys as the matrix phase provides a by increasing the reinforcement volume fraction and large variety of properties such as low density, better cor- decreasing the particle size. The thermal conductivity of rosion resistance, good thermal and electrical conductiv- AMCs decreases with increase in reinforcement volume ity, high damping capacity, and ability to be strengthened fraction and porosity. by precipitation [5]. AMC finds applications in a number of structural, Keywords: aluminium matrix composite; thermal cycling; nonstructural, and functional applications in different thermal fatigue; thermal stress. fields of engineering due to their economic, environmen- tal, and performance benefits. They are now being used in DOI 10.1515/secm-2013-0185 fabricating aerospace and automotive components such Received August 7, 2013; accepted May 18, 2014; previously as ventral fins, fuel access door covers, rotating blade ­published online September 24, 2014 sleeves, gear parts, crankshafts, and suspension arms. In the electronic industry, they are used in the processing of integrated heat sinks, microprocessor lids, and microwave *Corresponding author: Khushbu Dash, Department of Metallurgical housing. Majority of these applications require operation and Materials Engineering, National Institute of Technology, at varying temperature conditions ranging from high to Rourkela 769008, India, e-mail: [email protected] Suvin Sukumaran and Bankim C. Ray: Department of Metallurgical cryogenic temperatures [6, 7]. Structural components and Materials Engineering, National Institute of Technology used in the aerospace industries such as aircraft wings, Rourkela, Rourkela 769008, India fuselage frames, and landing gears are frequently exposed 2 K. Dash et al.: The behaviour of aluminium matrix composites under thermal shock-a review to severe thermal loads created by aerodynamic heating compared with other reinforcement types. But the [8, 9]. Hence, it is important to study their thermal behav- mechanical properties of particulate-reinforced MMCs iour under these conditions. are less compared with that of whisker- and fibre-rein- The ceramic phase, which is generally used as rein- forced MMCs. These composites are usually isotropic forcement in MMC, has a large difference in coefficient in nature and manufactured mainly by powder metal- of thermal expansion (CTE) with the metal phase. This lurgy processing. Secondary forming operations such thermal mismatch causes large residual thermal stresses as rolling, forging, and extrusion can be performed near the interfaces of the composite when they are cooled easily on particulate reinforced MMCs. from the fabrication temperature [10, 11]. These residual 2. Short-fibre- and whisker-reinforced MMCs: Here, non- stresses have a significant influence on the mechanical continuous fibres with an aspect ratio > 5 are used. and physical properties of the composite. Thermal stresses Whisker reinforcements have better mechanical prop- are important in design because they lead to plastic erties than both continuous and particles reinforced yielding or failure of the material [12]. There are several composites do and are mainly fabricated using pow- mechanisms by which thermal stresses can be relaxed, der metallurgy or squeeze infiltration route. Short which include interface debonding, by micro plasticity fibre reinforcements find major application in use as of the metal matrix and crack initiation and propaga- piston and have intermediate properties in between tion [13]. AMCs are frequently being exposed to environ- that of continuous-fibre-reinforced MMCs and partic- ments where cyclic temperature changes prevail [10]. ulate-reinforced MMCs. Since some health hazards are Thermo-mechanical cycling (TMC) also greatly influences identified with the use of whisker as reinforcement, the thermal expansion behaviour of the composites [14]. the commercial applications of whisker-reinforced Studying the thermal expansion behaviour of composites MMCs have been limited lately. also has significant importance since most of the struc- 3. Continuous-fibre-reinforced MMCs: Continuous fibres tural components are subjected to variable temperature are used here, which can be parallel or prewoven. operation, and hence, the materials used in these compo- They have diameter normally < 20 μm. The high- nents should have lower and stable CTE and outstanding est degree of load transfer is provided by them due mechanical properties [3]. Thermal expansion behaviour to their high aspect ratio. A significant amount of of composites is affected by several material parameters, strengthening takes place along the fibre direction. such as the type of constituents, matrix microstructure, Liquid infiltration techniques are mostly employed reinforcement volume fraction and architecture, the inter- for the fabrication of continuous-fibre-reinforced nal stresses due to their CTE mismatch, thermal history, MMCs. But the main disadvantage of these compos- porosity, and interface bond strength [9]. ites is that they have low strength in the transverse Thermal fatigue study of AMCs is important since they direction compared with the longitudinal direction, are used in engine components that are subjected to thermal which results in anisotropic properties. excursions during service, such as piston crowns and valves 4. Monofilament-reinforced MMCs: Here, fibres hav- [15]. It is also important to understand properties like ing a large diameter, in the range of 100–150 μm, are thermal conductivity of MMC for various thermal manage- used as reinforcements. Chemical vapour deposition ment applications such as heat sinks in electronic devices, (CVD) and diffusion bonding techniques are normally microprocessors, power semiconductors, high-power laser used for their fabrication. Their application is limited diodes, light-emitting diodes, and micro-electro-mechani- to superplastic forming metal matrices. Like contin- cal systems [16, 17]. This paper reviews the current knowl- uous-fibre-reinforced MMCs, these composites also edge on the behaviour of AMCs under thermal stresses. have anisotropic properties.

In addition to the above types, another variant called 2 Classification of MMCs hybrid AMCs having more than one type of reinforcement is also present, which is in use to some extent [6, 18, 19]. There are mainly four types of MMCs based on the type of reinforcement: 1. Particle-reinforced MMCs: These are produced by 3 Synthesis of AMCs incorporating equiaxed reinforcements in the form of particles to the metal matrix. The aspect ratio of par- Primarily, fabrication of MMCs can be classified into two ticles will be < 5, and these MMCs are less expensive processes such as solid-state processes and liquid-state K. Dash et al.: The behaviour of aluminium matrix composites under thermal shock-a review 3 processes. Solid-state processes include powder metal- the matrix and the reinforcement due to the shorter lurgy process, diffusion bonding, and deposition process. processing period involved. Gas pressure infiltration is Powder metallurgy is generally used for the fabrication another variant of liquid infiltration technique where the of particulate-reinforced AMCs, which involves blending, fibre preform is being infiltrated by using a high-pressure cold pressing and sintering, or hot pressing. For a near- inert gas. In the spray forming process, the composites are net-shaped final product obtained, powder forging tech- produced by injecting the ceramic particles into an atom- nique can be used where forging operation is done after ized spray of molten metal matrix. Even though expensive the sintering stage. Diffusion bonding is another solid- compared with other processes, this fabrication route is state process for MMC fabrication where interdiffusion very flexible, and here, no harmful reaction products are of atoms between metallic surfaces takes place at high formed since the time of flight of composite particles is temperatures, leading to bonding. It can be used for pro- much less [6, 18, 19]. duction of monofilament-reinforced AMCs through either foil-fibre-foil route or by evaporating thick aluminium layers onto the fibre surface. This process is not suitable for obtaining high volume fraction of fibres and obtaining 4 Thermal stresses complex shapes. Also, long processing time and high tem- perature requirements limit the commercial applicability Under uniform temperature change, thermal residual of the process. stresses are induced in a composite due to thermal expan- Various deposition techniques are available for MMC sion mismatch between the matrix and the reinforcing production, which includes spray deposition, physical phases. These stresses are important as they exist at room vapour deposition (PVD), and CVD. The PVD process can be temperature and can decrease the initial yield stress rela- used for the fabrication of multilayered MMCs in the nano- tive to its value in the absence of internal stress. Even metre range by passing the fibres continuously through a though a small strain mismatch can be accommodated region of high partial pressure vapour of the metal to be elastically, the entire deformation volume of a MMC may deposited. Here, condensation takes place, which results become plastically deformed for larger strain mismatch in a relatively thick coating of metal on the fibre. The [10–12, 20]. For Al-based MMCs, it has been reported that spray deposition technique involves spraying the molten dislocations are generated at the reinforcement-matrix metal onto winded fibres to form the composite. Rapid interfaces to provide sufficient space for the strains origi- solidification of molten matrix and easier control of fibre nating from the CTE mismatch between the Al matrix and alignment are the main advantages of spray deposition. the ceramic reinforcements, and the internal stresses The decomposition of vaporised metallic component onto resulting from this mismatch are being partially relaxed the substrate fibre at elevated temperature constitutes by these dislocation processes [21–23]. These residual the CVD technique. The ease of controlling the degree of thermal stresses can accelerate creep by orders of mag- interfacial bonding is one of the key aspects of deposition nitude, and when their average value exceeds the exter- processing. The main problem with deposition techniques nally applied stress, they will play a significant role in the is that they are very time-consuming processes. determination of plastic deformation rates. When average Liquid-state processing involves stir casting, infiltra- thermal stresses within a composite become similar to (or tion techniques, and spray deposition process. In the stir greater than) the externally applied stresses, the thermal casting technique, the ceramic particles are first added to stresses will have an important (or dominant) role in the molten metal and then it is allowed to solidify. Proper determining plastic deformation rates [24]. wetting between the reinforcement and the molten metal Even though experimental determination of residual is an important criterion, which determines the mechani- stress in composites is possible through X-ray diffraction cal properties of the final product. One of the main prob- (XRD) [25], dilatometry, and acoustic methods, difficulties lems with this technique is that the reaction between the arise in measurement, which includes lowering the origi- fibre and the matrix may take place, which can deteriorate nal value of residual stress due to relaxation processes the properties of the composite. Compo casting is another taking place. Original residual stresses can be ascertained variant of stir casting, where the particles are incorporated by the comparison of stress/strain curves of the matrix to the matrix alloy in the semisolid state. A fibre or particu- and the fibre. Generally, during the fabrication of compos- late preform is being infiltrated by using molten metal for ites, a tensile residual stress is developed in the matrix, the production of the composite in the case of infiltration which accelerates the change in deformation from elastic technique. Here, minimal reaction takes place between to plastic in the matrix, whereas a compressive residual 4 K. Dash et al.: The behaviour of aluminium matrix composites under thermal shock-a review stress is generated in the fibre, which slows the transi- aluminium system, it predicts the resultant stress dis- tion. This transition occurs when the residual stress in the tribution between the phases. The relaxation of thermal matrix exceeds the yield strength of the matrix. During residual stresses occurs commonly through mechanisms the prestraining of the composite, residual stresses are such as general plastic deformation (plasticity, creep), formed in the fibre and the matrix, which also depend on fracture (fibre cracking, interfacial decohesion, interface the value and sign of original residual stress [26]. Prema- gliding), local plastic deformation (dislocation glide), and ture yielding during axial loading and yielding prior to diffusion (atom movements), which affect stress contribu- loading occur due to the generation of residual stress in tions in many ways. Generally, residual stresses can be the composites. relaxed during cooling at high temperatures, but at low During cooling of a composite from fabrication tem- temperatures, relaxation is very difficult to take place. But perature, there are three different stages of matrix defor- there exists complexity in the Al-SiC system where relaxa- mation. In the first stage, occurring at high temperatures, tion of residual stresses occurs even after it is cooled down there will be a negligible amount of residual stress build- to room temperature [34, 35]. up in the matrix due to stress relaxation. The temperature High density of matrix dislocations was observed near at which the first stage ends depends on the cooling rate the interface in the SiC-Al system created due to residual employed. In the second stage, as temperature lowers stress, and these dislocations are emitted from the inter- further, the residual stress build up begins in the constitu- face. The effect of the aspect ratio of SiC fibres on thermal ents rapidly, and when the value of the residual stress is stresses is very small because at short distances from the large enough to exceed the yield strength of the matrix, fibre corner, the dependence of residual fields on the axial plastic deformation occurs in the matrix, which marks position is negligible. But side-to-side fibre spacing has a the beginning of the third stage. In the case of graphite- profound influence on the magnitude and distribution of fibre-reinforced AMCs, there exists anisotropy in the CTE residual stress in the fibre and on the interfacial stresses at of graphite fibre (0°C-1 in the longitudinal direction and whisker end. As side-to-side fibre spacing was increased, 25 × 10-6°C-1 in the transverse direction), which results in greater plastic strains were observed on the fibre corner larger residual stress in the longitudinal direction com- [36]. Thermal residual stresses have a very small direct pared with the transverse direction. This anisotropy in influence on the predicted ductility of the SiCf-Al system, thermal expansion is beneficial as it, along with slow even during the case when the limiting failure mechanism cooling rate, can eliminate the plastic deformation in the in modelling the system was assumed to be void nuclea- matrix created due to CTE mismatch [25, 27–29]. Volume tion by interfacial debonding at the whisker ends. There is fraction of graphite fibres also plays a significant role in a tendency for redistribution in residual stresses as inter- deciding the magnitude, sign, and distribution of residual face failure begins to occur. Flow strength in compression thermal stresses [30]. was observed to be greater than that in tension when a The effect of modifying the residual thermal stress in closer end-to-end spacing of fibres was maintained [37]. a graphite-fibre-reinforced AMC by tailoring the interfa- The various factors on which the magnitude of cial region was observed to depend on the elastic modulus thermal residual stress depends include the CTE of the of the interfacial region. A larger elastic modulus resulted fibre and matrix, their elastic properties, and matrix-plas- in increasing the contribution of the matrix-interfacial ticity behaviour. In a unidirectional SiC-fibre-reinforced region mismatch terms, which is very harmful to the 6061 aluminium alloy, the amount of age hardening pro- thermal stresses in AMCs as it increases tensile hoop duced is significantly less than that in the bulk alloy due stresses during cooling from processing temperatures to the plastic strain brought about during cooling. Solute [31]. The overall yield surfaces of particulate-reinforced migration tendency to fibres as well as to strain-induced composites, such as SiC-Al, exhibit a combination of kine- dislocations may enhance this effect. Stress relaxation matic and isotropic plastic hardening due to the presence behaviour in the matrix of the composite is more intricate of thermal residual stresses [32]. than that of normal stress relaxation. Due to the elastic In the majority of the earlier investigations, the resid- constraint imposed onto the matrix by the fibres, plastic ual thermal stresses were determined using either the strain due to stress relaxation will create a decrease in Eshelby method or the finite element method (FEM) [22]. elastic strain in both the fibre and the matrix [38, 39]. The Eshelby model was used for the prediction of thermal Analysis of the influence of thermal residual stresses and residual stress in SiC-whisker-reinforced aluminium com- strains on the mechanical behaviour of Al2O3-particulate- posites, which were in good agreement with the experi- reinforced 6061-T0 Al alloy through the FEM pointed out mental results [33]. In the case of SiC-short-fibre-reinforced that the residual stress/strains have a large influence on K. Dash et al.: The behaviour of aluminium matrix composites under thermal shock-a review 5 composite behaviour under compressive loading than the whisker volume fraction. Thermal residual stresses under tensile loading. Also, when the residual thermal were found to delay void nucleation in the matrix and stresses/strains are introduced prior to external loading, interfacial decohesion at the fibre ends during tension at there was a decrease in proportional limit, 0.2% offset lower whisker volume fraction. This was attributed to the yield stress and the apparent stiffness. Regions of con- induction of hydrostatic compression in the interfacial strained plasticity and high stress triaxiality in the matrix region of the composites. But at higher volume fraction, around the particle help in better load transfer in the com- they aid matrix and/or interfacial damage in the fibre posite [40]. ends under tensile loading [48]. In SiC-particle-reinforced Increased matrix dislocation density as a result of aluminium matrix fabricated by casting, residual stress differential thermal contraction is the main strengthen- distribution was observed to increase with the cooling ing mechanism observed in discontinuously reinforced rate. The elastic modulus and yield strength of the com- MMCs, such as SiC-Al. Various undesirable composite posite were increased due to the residual thermal stress properties are caused by thermal stresses such as lower- formation. The value of residual stress was higher at the ing of tensile and yield strengths and elevating Bausch- interface and decreases with distance from the interface. inger effect, affecting the load transferring capability of Interface stresses are strongly influenced by the thermal the matrix to the fibre. Residual stresses can enhance residual stresses when mechanical loading is applied. interface strength in the case of composites with weak Residual stresses were found increase with the volume interfaces during tensile loading, which is more promi- fraction of the fibres in the heat removal direction and nent in regions of plastically deformed matrix near the the direction normal to it for air-cooling and water-cool- interface [41, 42]. ing, but for constant temperature cooling, they remained The average strain in AMC components resulting from approximately the same for different volume fractions thermal stress can be determined by using neutron dif- in the heat removal directions and decreased with the fraction and synchrotron techniques. An equivalent inclu- volume fraction of fibres in other directions [22, 49, 50]. sion method helps in interpreting these thermally induced In a nickel-coated carbon-fibre-reinforced AMC, thermal lattice strains [20, 43–45]. The effect of thermal stress on stresses arising due to temperature gradient resulted in the microstructure of Al-MMCs reinforced with nanosized the failure of nickel coating [51]. Thermal residual stresses dispersoids of mullite and zirconia fabricated using uni- were significantly reduced by using a thermal expan- axial hot-pressing was studied at three different tempera- sion clamp during curing in the case of a carbon-fibre-­ tures in the range of 450°C–610°C. Through the analysis, reinforced aluminium laminates (CARALL) [52]. an estimation of the thermal stress of the AMCs using ana- A two-fold effect of pressure rolling on the residual lytical functions was made possible. The results revealed stresses was observed on the residual stress state of a that the magnitude of the thermal stress increased with SiC-particle-reinforced 2014 aluminium alloy, which was the increase in processing temperature. As a result, the Al investigated by using a modified Eshelby model [53]. matrix was deformed to the extent that the effective par- The residual stresses induced by annealing in silicon- ticle sizes of the matrix decreased, strains increased, and carbide-particle-reinforced aluminium MMC measured the values of dislocation density increased with increasing using the XRD method indicated that hydrostatic tension processing temperatures, which lead to the betterment of was developed for aluminium matrix and hydrostatic the hardness and mechanical properties of the composites compression was developed for the SiC reinforcement, [46]. Analysis using FEM models depicted that thermal which established force equilibrium, making the mac- stresses cause pronounced matrix yielding during axial roscopic thermal stress absent in the composite. But tensile loading and elevates flow stress during transverse when the composite was subjected to grinding treat- tension in unidirectional boron-fibre-reinforced 6061 Al ment, the thermal stresses were reduced and resulted in alloy composite. However, local spacing of the fibres also macroscopic stress such that compressive stresses were influences the local magnitude and sign of the residual measured in both constituents [54]. The tensile stress dis- stress in the matrix [47]. tribution due to the rotation of aluminium-based compos- In the case of discontinuous SiC-whisker-reinforced ite disc containing silicon carbide whisker was changed aluminium-matrix composites, it was observed that resid- significantly due to the presence of residual stress. The ual stresses significantly affect the constrained matrix steady-state tensile creep rate was further enhanced by plastic flow and matrix-fibre load transfer, which were the the tensile residual stress, which may lead to a decrease main strengthening mechanisms. The effect of thermal in the service life of components that find major applica- residual stress on load transfer was found to increase with tions in flywheel, shrink fits, and turbines [55, 56]. When 6 K. Dash et al.: The behaviour of aluminium matrix composites under thermal shock-a review the stiffness of a composite material is increased, it will 5 Thermal cycling also obstruct thermal expansion, which, in turn, will lead to higher residual thermal stresses whose maximum In the case of AMCs, heat treatment has been widely used along the axial direction occurs in the boundary of elastic as an effective technique for acquiring stabilized micro- and plastic zones [12]. structure and best achievable properties. Thermal cycling In the FEM analysis of the distribution and the is one of the effective heat treatment methods used to amount of residual stresses in the SiC-particle-reinforced reduce residual stresses and improve the dimensional aluminium matrix carried out recently by Bouafia et al. stability of the composite, which involves alternatively [57], large amounts of residual thermal stresses were cooling and heating the material to improve its strength generated, mostly at the interface region of the compos- and capacity [8]. Information on stress relaxation mech- ite. The SiC particle volume fraction and interparticle anisms can be obtained from the in-cycle creep data for spacing play a prominent role in the production of resid- MMCs subjected to thermal cycling, which are essential ual thermal stresses. Lowering SiC particle volume frac- in life prediction of structural components under thermal tion and increasing interparticle spacing resulted in the cyclic environments [24, 61]. decrease in internal residual stresses and vice versa. SiC During thermal cycling conditions, the strain rate particles with cylindrical shape developed higher thermal sensitivity exponent (m) of the composite material is unity stresses compared with those with spherical and cubical (Newtonian flow behaviour) since the strain rate will be shape. Increasing the reheating temperature also resulted proportional to the applied stress, and at higher stresses, in the increment of normal internal stresses introduced in the material gradually approaches the isothermal flow the matrix and the particle. behaviour of the matrix. The internal stresses that block or Thermal residual stresses are one of the main causes assist the motion of individual dislocations are the main of interface damage and fatigue crack initiation during cause for thermal cycling behaviour [62]. During thermal high-cycle fatigue in alumina-reinforced AMCs. Surface cycling, shape instabilities and creep at significant strain treatment capable of inducing compressive residual rates are observed in MMCs, even at unusually low stresses stresses can be done to improve the fatigue behaviour [63, 64]. Densification of composites also increases during of the composite by preventing crack propagation in the thermal cycling under stress. Thermal cycling of a mate- initial stage itself [58]. The relaxation of thermal stresses rial can be used to induce internal stress in many ways, in aluminium-based MMCs subjected to thermal cycling such as (1) thermal cycling of a material through a phase was studied using mechanical spectroscopy where the change while applying a small external stress; (2) thermal mechanical losses and shear modulus measurements cycling of polycrystalline pure metals or single-phase were taken as a function of temperature. When the com- alloys with anisotropic CTEs with simultaneous load posite was cooled, maximum mechanical loss appeared application, and (3) thermal cycling of composite materi- because of the dislocation generation and motion near als in which constituents have different CTEs [65]. the interfaces due to differential thermal contraction The behaviour of MMCs will be different from isother- of the matrix and reinforcement. This mechanical loss mal conditions when they are subjected to temperature maximum is absent in the case of monolithic mate- cycling. Strain rate increases substantially even at lower rial. Also, when the reinforcement volume fraction was temperatures during thermal cycling when compared increased, the maximum mechanical loss was found with isothermal conditions. Fracture strain is also high to be increased [59]. The change in residual stress on during thermal cycling. Damage modes occurring during heating and cooling of the magnesium borate whisker- thermal cycling such as fibre and interface cracking and reinforced aluminium composite was investigated using interface void formation may not necessarily occur during CTE curves, where a close relationship was observed isothermal conditions [10, 66]. The plastic flow stress and between the CTE of the composite and the residual stress strain hardening behaviour of the composites are affected changing rate. A linear relationship was observed on by thermal cycling because of the triaxial state of stress tensile and compressive stress changing rates with tem- generated at the particle-matrix interfaces. Residual stress perature while heating the matrix at lower and higher can also have a considerable influence on the composites temperature ranges. The CTE of the composite on cooling subjected to thermal cycling as they can initiate matrix was found to be smaller than that on heating. This is phase yielding, which causes the formation of strain hys- because while heating, the residual stress present in the teresis loops between the heating and cooling segments of Al matrix is elastically relaxed in lower or higher tem- the cycle. Large percentages of plastic deformation, known perature region [60]. as internal stress superplasticity (mismatch induced K. Dash et al.: The behaviour of aluminium matrix composites under thermal shock-a review 7 superplasticity [63]), occurs in Al-SiC particulate com- excess of 150%. The roughness of squeeze-cast AMC rein- posite when they are subjected to concurrent application forced with 15 vol% random discontinuous saffil fibres of small external stress and thermal cycling [11]. Studies subjected to thermal cycling was found to increase with show that enhanced creep deformation is observed during the number of thermal cycles and increased dislocation the thermal cycling of particulate-reinforced MMCs and density was observed at the interfacial region [77]. The that plastic strain per thermal cycle varies linearly with damage mechanisms at the fibre-matrix interface are the the applied stress [67]. chief factor contributing to the enhanced nonlinear strain When the displacement measurements within the accumulation associated with thermal cycling [78]. graphite fibre of a graphite-fibre-reinforced AMC during A comparative study was made on the damage in three thermal cycling was studied using in situ SEM experiments, discontinuously reinforced AMCs produced by powder intrusion/extrusion phenomenon was observed that limits metallurgy route, such as Al-Mg-Cu alloy-SiC (as extruded the matrix to fibre load transfer across the interface. Once and solution treated), Al-SiC (hot rolled), and in situ cast shear begins within the graphite fibre, this phenomenon Al-TiC (as extruded) composites. The matrix and interfaces will further lead to modification of the CTE of the compos- in the Al alloy-SiC composite show cracks as a result of ite [68]. Superplastic behaviour was observed at very low repeated thermal cycling because of the stress generated stresses (∼10% of the yield strength) in SiCp-Al composite from the mismatch in CTE. This leads to a decrease in the when it was subjected to thermal cycling on a conventional Young’s modulus of the composite and matrix microhard- creep machine at relatively low temperatures ( < 200°C) ness near the particle-matrix interfaces. But an increase

[62]. A drastic strength decrease is observed for the Al-SiCw in the microhardness of the interface by strain hardening system when it was subjected to thermal cycling in which due to dislocation generation was observed, as shown in plastic flow occurs, and this effect limits the use of these Figure 1, in the case of as-rolled Al-SiC and in situ Al-TiC materials in high temperature applications [69]. as their interfaces are very strong so that no decohesion Thermal cycling can result in the generation of inter- takes place. Also, there was only a marginal decrease in nal stresses at Al-SiCw interfaces, which are produced due Young’s modulus in these composites. Bar charts depict- to the CTE mismatch between SiC whiskers and alumin- ing the damages caused by thermal cycling in the Al alloy- ium matrix. The deformation produced due to these inter- SiC, Al-SiC, and Al-TiC composites against the number nal stresses results in superplastic behaviour, leading to of cycles of heating and water quenching are shown in very high ductility of the composite ( > 1000%) and, thus, Figure 2 [79]. high strain rate sensitivity. Also, the rate of realignment The phenomenon where progressive plastic deforma- and redistribution of whiskers are increased as a result tion with increasing number of thermal cycles occurs in of the increase in stress during thermal cycling [70]. The composites even at stress levels far below the yield strength influence of thermal cycling on the precipitation events in is known as thermal ratchetting, which is a function of both a SiC-whisker-reinforced aluminium alloy matrix compos- ite was studied by Hall and Patterson [66] at higher tem- peratures. A dispersion of finer precipitates was observed during thermal cycling in addition to the coarser precipi- tates, and large dimensional changes were noticed. Neutron diffraction techniques and X-ray synchro- tron measurements were found to be useful for measur- ing internal strain changes in SiC-whisker-reinforced AMC undergoing thermal cycling [71–73]. The formation of stable dislocation tangles and cells was observed near the interface regions during thermal cycling of 10 vol% alumina-particulate-reinforced 1060-AMC [74]. Internal stresses resulting from CTE mismatch between the matrix and the particulates are partially relaxed by such disloca- tions [75]. Creep acceleration and reduction in creep stress exponent have been reported [76] in the theoretic model of aluminium reinforced with SiC particles, which was sub- Figure 1 Bright field TEM image showing dislocation tangles in the jected to thermal cycling under load. This further resulted matrix of the as rolled Al-SiC composite. Reprinted from Ref. [79], in superplastic deformation behaviour with extensions in Copyright (2008) with permission from Elsevier. 8 K. Dash et al.: The behaviour of aluminium matrix composites under thermal shock-a review

0.12 50% Al alloy-SiC 50% 45% 6061 Pur Al-SiC 0.10 6061 Composite 40% Pur Al-TiC Re-extruded 6061 composite 35% 0.08 30% 28% 27%28% 25% 0.06 20% Damage

Elongation 20% 15% 0.04 10%12% 10%

0.02 5% 0% Room temperature Isothermal Thermal cyclic 0 Two Five Eight Number of thermal cycles Figure 3 Elongations for various test conditions and material systems. Reprinted from Ref. [3], Copyright (1997) with permission Figure 2 Bar charts depicting the damages caused by thermal from Elsevier. cycling in the Al alloy-SiC, Al-SiC and Al-TiC composites against the number of cycles of heating and water quenching. Reprinted from Ref. [79], Copyright (2008) with permission from Elsevier. not change the dislocation density in the work hardened zone of short-fibre-reinforced AMCs once cyclic satura- tion is achieved. Also, the dislocation loop punching that the applied stress and the amplitude of thermal cycles. occurs during thermal cycling at high applied stresses

Thermal ratchetting behaviour has been observed in SiCf-Al acting for a short period of time is controlled mainly by composites [80] and in C-Al and SiC-Al laminated composites dislocation glide mechanism [85]. [81]. Ratchetting rate is not influenced by the highest tem- Thermal cycling tests carried out on unidirectional perature of the thermal cycle, but it plays a prominent role Mo-fibre-reinforced aluminium composites showed in composite failure at high load levels. Progressive reduc- reduced tensile strength due to the presence of interfacial tion in length was observed during repeated thermal cycling cracks, which are generated to relax the internal stress of saffil-fibre-reinforced AMCs until a steady-state cycle is built up in the composite. When the thermal cycles are reached [82]. Thermally activated recovery by polygoniza- increased, the axial length also increased, but the density tion was the major microstructural damage mechanism was reduced. The interfacial reaction between the Mo- involved with thermal cycling at elevated temperature, as fibre and Al matrix is unimportant after thermal cycling observed in the case of a short alumina-silicate-fibre-rein- as the Mo fibres remained intact [86]. During thermal forced aluminium alloy (A356) composite fabricated by pres- cycling, strain hysteresis was observed in Al-12 wt% Si sure casting. Reduction in residual stresses was observed in alloy matrix composite reinforced with potassium titan-

Al2O3-fibre-reinforced Al matrix and SiC-whisker-reinforced ate whiskers fabricated by squeeze-casting process. Fur- Al matrix during thermal cycling. This effect was due mainly thermore, the composite exhibited a small dimensional to the thermally induced plastic relaxation, and the degree change due to the absence of residual plastic strain [87]. of reduction depends on the difference in CTE between the A theoretical model based on micromechanical approach constituent phases, volume fraction of the reinforcement, was developed by Dutta [88] for examining the thermal and drop in temperature [83, 84]. cycling behaviour of continuous graphite-fibre-reinforced A study on the effect of strain control and thermal AMCs. From the analysis, it was concluded that the influ- cycling on aluminium alloy (AA6061) reinforced with ential deformation mechanism functioning in the matrix alumina particles produced by casting and extrusion changes frequently during thermal cycling as a result pointed out that thermal stresses produced by thermal of continuous stress and temperature revision. Overag- cycling can increase plasticity at the matrix-reinforcement ing, matrix annealing, void coalescence, crack forma- interface whose maximum exists at an optimum strain-rate tion, and reinforcement debonding were observed in condition. A comparison was made on the elongation of SiC-whisker-reinforced 6061 AMC subjected to thermal the composite with unreinforced AA6061 and reextruded cycling at beginning of T6 tempering treatment. Due to 6061 composite under room temperature, isothermal and these defects, the tensile strength, yield strength, Charpy thermal cycling conditions that are represented in the bar impact energy, and fracture toughness of the composite chart in Figure 3 [3]. Loading under thermal cycling does were significantly reduced [89]. K. Dash et al.: The behaviour of aluminium matrix composites under thermal shock-a review 9

The maximum residual stresses were found to be gen- to as near as die-casting condition in moderate range and erated at the interfacial region during the thermal cycling narrow range thermal cycling treatments [92]. of 10 vol% SiC-reinforced aluminium-silicon alloy matrix High dislocation density was observed in the cast composite [90]. Also, crack formation was observed in aluminium alloy matrix composite reinforced discontinu- SEM micrographs under repeated action of thermal cycling ously with fine SiC particulates during thermal cycling, process. The development of strain hysteresis loops was which can be attributed to the difference in CTE of the observed in aluminium-based composites reinforced with constituents [93]. The internal friction of Al-SiC particu- in situ Al3Ti plate and Al2O3 particles produced by reactive late MMC was increased by thermal cycling, which is hot pressing method, which were exposed to fluctuating due mainly to the dislocation motion. Thermal cycling temperature environments [11]. Thermal damage of com- also caused an increase in the dislocation density, which posites subjected to practical thermal cycling conditions enhanced the damping capacity of the composite [94]. can be assessed by thermal strain parameters deduced Reduction in hardness after thermal cycling was observed from the hysteresis loop, and it was observed that in situ in short-fibre-reinforced AlSi12CuMgNi piston alloys, composites displayed better dimensional stability com- which can be attributed to overaging and matrix recovery pared with Al during thermal cycling, which are attrib- [95]. The thermal cycling of extruded aluminium-silicon- uted mainly to the higher thermal stability and improved alloy-based (Al-7% Si-0.7% Mg) composites reinforced interface bonding of in situ reinforcements. The mechani- with particulate silicon carbide displayed higher ductility cal strength of graphite-fibre-reinforced aluminium com- due mainly to internal stresses that are being generated as posites produced by gas pressure infiltration method was a result of CTE mismatch between the matrix and the rein- reduced by 10% due to thermal cycling between 60°C and forcement. Thermal cycling caused the matrix alloy and 300°C up to 1020 cycles. There was no effect of thermal its composites to deform at stresses much lower than the cycling on the thermal conductivity of the composite yield stress. Formation and transversal growth of the cavi-

[91]. The mechanical properties of the AlNp-Al compos- ties were found to be the dominating factor in the failure ite fabricated by squeeze casting were greatly enhanced of thermally cycled composites, which are present along after thermal cycling treatments. The tensile strength the gauge leading cracks in the matrix in the transverse of the composite was improved mainly by wide-range direction to the applied stress, as observed in Figure 4 thermal cycling treatment, whereas the stability of mate- [14, 96]. rial properties was improved by narrow range treatment. A constant and higher strain rate compared with Moderate-range cycling treatment was found effective pure creep conditions was observed in the deformation in improving yield strength and elastic limit. The elastic behaviour of an aluminium alloy matrix composite rein- modulus of the composite was also significantly increased forced with SiC particles during thermal cycling between

A B

Figure 4 Longitudinal cross-sections of (A) AlSi7/SiCp/10 vol% and (B) AlSi7/SiCp/20 vol% thermally cycled samples tested at 5–4 MPa, showing cavities, cracks, and interfacial debonding (ID). Reprinted from Ref. [14], Copyright (2004) with permission from Elsevier. 10 K. Dash et al.: The behaviour of aluminium matrix composites under thermal shock-a review the temperature limits of 305 K and 443 K. The enhanced considerable practical interest. The CTE of MMCs can be deformation taking place in the thermally cycled compos- tailored by controlling the reinforcement type, size, mor- ites can be explained on the basis of emission and punch- phology, and relative quantity [105–108]. The thermal ing out of dislocations from the interfacial region [97]. The expansion behaviour of MMCs, especially AMC, is very plastic strain of the Al-SiCp composite was observed to be complicated since the metal matrix can deform plastically increasing when compared with unreinforced aluminium in reaction to the thermal stresses generated. The super- due to CTE mismatch between the aluminium matrix and plastic flow behaviour exhibited by MMCs during thermal SiC particles during thermal cycling [16]. cycling is due to the generation of internal stresses that

The thermal stresses generated in SiCp-reinforced alu- result from the difference in thermal expansion coefficient minium matrix during the early stages of thermal cycling of the matrix and of the reinforcement [109–111]. Plastic relaxes by creep at temperatures above 363 K. Develop- deformation will increase at elevated temperatures as ment of strain hysteresis loops and residual plastic strains thermal strains are a function of the temperature, which was observed in Al-Mg alloy reinforced with fly ash during increases with temperature [112]. Internal stresses gener- thermal cycling. Further addition of fly ash resulted in the ated due to the CTE mismatch between fibres and matrix reduction of residual plastic strains, thereby increasing strongly affect the thermal expansion behaviour and the dimensional stability of the composite [98, 99]. Co- dimensional stability of AMCs [113]. continuous composites such as NiAl-Al2O3 and Al-Al2O3 In the case of particulate-reinforced AMCs, it was show more resistance to thermal cycling damage when observed that the CTE of the composite can be reduced compared with traditional composites [100]. substantially by the addition of reinforcement particles Investigations on the thermal cycling behaviour of to the aluminium matrix [114, 115]. Also, the CTEs of both ultra-high-modulus carbon-fibre-reinforced aluminium the composite and the unreinforced Al matrix were found alloy matrix composite revealed the occurrence of micro- to increase with increasing temperature. The restriction cracks and debonding at the interface region, which imposed by the particles on the Al matrix during thermal increased with the temperature difference. The bending expansion resulted in a slower increasing rate of the CTE modulus of the composite showed an abrupt increase of the composite compared with that of unreinforced Al initially and then decreased with thermal cycles due to [116]. In the case of SiC-coated boron-fibre-reinforced generation and propagation of microcracks. The amount AMCs fabricated by hot pressing, the thermal expan- and length of pull-out fibre were seen to increase with sion measured perpendicular to the fibres was found to the number of cycles irrespective of temperature differ- be in good agreement with that predicted by theoretical ence. Even though there is a gradual increase in disloca- models, whereas thermal expansion measured in the tion density in the Al matrix with the number of thermal fibre axis direction was lower than that of theoretical cycles, it will eventually decrease after a certain period predictions due to the yielding in the matrix [117]. Ani- due to the movement and annihilation of dislocations sotropy of fibre thermal expansion observed in the case of opposite sign [101]. Thermal cycling tests conducted of graphite fibre can eliminate yielding of the matrix due on a saffil-fibre-reinforced 2014 aluminium alloy matrix to thermal stress [27]. Experimental measurement of the resulted in an interfacial reaction, which strengthened the thermal expansion behaviour of both unreinforced 2080 composite and led to a decrease in its CTE [102]. TMC could Al and Al-SiCp extruded composites revealed that aniso- induce interfacial degradation in the case of boron-fibre- tropic distribution of SiC particles decides the anisotropic reinforced AMC. This effect increases with the number of thermal behaviour of Al-SiCp composites. Also, both the TMC cycles and the applied stress levels [103]. particle volume fraction and the orientation relative to the extrusion direction have a prominent effect on the CTE of the composite. The FEM results used to simulate the thermal behaviour showed that orientation of SiC parti- 6 Thermal expansion cles changes the internal stress in the composite, which results in anisotropic thermal behaviour [118]. Low and stable CTE is a necessity for components that Thermal deformation measurement in AMCs can are subjected to a change in temperature such as struc- be performed using a nondestructive optical technique tural parts in aerospace, automobiles, sensors, measur- known as digital speckle pattern interferometry without ing instruments, and microwave components [104]. Due any restriction on the shape of the specimen [119]. Increas- to suitability of AMCs for elevated temperature applica- ing volume fraction of saffil fibres caused a decrease in tions, studying their thermal expansion property is of the value of CTE of AMC. An elastic response at lower K. Dash et al.: The behaviour of aluminium matrix composites under thermal shock-a review 11 temperatures and a plastic response at higher tempera- [127]. The damping capacity of AMCs increases with the tures were exhibited by the measured CTE of the com- increase in temperature, whereas the dynamic modulus posite [82]. A semiempirical method for determining was found to decrease with the increase in tempera- the average CTE of components in an Al-SiC composite ture. The dislocations induced in the matrix alloy due to system depicted that when the volume fraction of SiC was CTE mismatch between the constituent phases explains increased, the average volumetric CTE of both the fibre the damping capacity of the AMCs at low temperatures, and the matrix were observed to decrease [120–122]. The whereas matrix-reinforcement interface and thermoelas- incorporation of alumina fibres as reinforcement into tic damping explains the damping capacity at higher tem- Zn-Al alloys results in a significant reduction in the creep peratures [128]. rate and the CTE [123]. Out of the AMCs prepared using The CTE curves can be used to qualitatively predict the reinforcements such as carbon, SiC, A12O3, and boron thermal mismatch stress (TMS) and change of its states, fibres, anisotropy of the CTE was found to be very large for as observed in the case of a SiC-whisker-reinforced AMC,

AMCs reinforced with carbon fibres [2]. SiCp-reinforced Al where the CTE of the composite was fluctuating with test alloy composite fabricated by stir casting exhibited lower temperature due to the presence of larger elastic and plastic CTE values than the Al alloy did [124]. TMS in the matrix. Within most temperature ranges, the A monotonic increase in CTE with plastic prestrain natural shrinkage of the matrix was observed to keep a was observed in alumina-particle-reinforced AMCs over balance with the plastic relaxation of the TMS in the matrix. smaller temperature ranges. But the CTE increase was The orientation distribution of SiC whiskers also influences greater for the composite with higher volume fraction the CTE of AMC [108, 129–132]. The CTE of the AMC contain- of alumina particles when they are studied over a wider ing β-eucryptite particles and aluminium borate whisk- temperature change. Also, particle cracking in particle- ers fabricated by squeeze-casting technique was affected reinforced MMCs can be monitored using CTE measure- significantly by the degree of interfacial reaction on the ments [125]. Plastic yielding and flow of matrix void surface of β-eucryptite particle. Stronger interface between nucleation and volume fraction of broken particles have the β-eucryptite particle and the Al matrix resulted in lower a prominent influence on the CTE of the composite [15]. CTE of the composite [103, 133, 134]. The overall CTE of AMC In the case of SiC-particle-reinforced AMCs produced is determined mainly by the total volume fraction of the by vacuum-assisted high-pressure infiltration, the CTE reinforcement than the average particle size and particle of the composite increased with the size of the particles size distribution [135, 136]. The CTEs of SiCp-Al composite [126]. The effect of phase connectivity (i.e., metal-matrix, and unreinforced Al matrix fabricated by squeeze-casting ceramic-matrix, and interpenetrating, where both phases method displayed maxima at 250°C and 350°C, respec- form a continuous network in space) has a strong influ- tively, as shown in Figure 5, which was then reduced at ence on the thermal expansion behaviour of the SiC-Al higher temperatures. The increase in solubility of Si in Al system, with the metal-matrix and ceramic-matrix cases due to the increase in temperature is mainly responsible for showing the highest and lowest CTE values, respectively this behaviour. Hence, the thermal expansion behaviour

AB10 24 SiCp/Al composite SiCp/Al composite )

-3 Matrix Matrix 8 20

6 /°C) 16 -6

4 12 CTE ( × 10 e linear length change ( × 10 2 8 Relativ

0 4 0 100 200 300 400 100 200 300 400 Temperature (°C) Temperature (°C)

Figure 5 Temperature dependence of relative linear length change (A) and CTE (B) for the composite and unreinforced matrix. Reprinted from Ref. [137], Copyright (2003), with permission from Elsevier. 12 K. Dash et al.: The behaviour of aluminium matrix composites under thermal shock-a review is greatly influenced by the heterogeneities of Al-based cooling rates. Analysis pointed out that the dislocation materials [137–139]. The temperature dependence of the density and TMS were lower in slowly cooled composites solubility of Si in Al also influences the CTE of Al-Si matrix compared with fast cooled (water quenched) composites. alloys significantly. In the case of SiC-particulate-reinforced Hence, cooling rate greatly influences the thermal expan- aluminium-matrix composites, it was observed that visco- sion behaviour of AMCs [9]. plastic matrix deformations and changing void volume The thermal expansion response of the fibre-reinforced fractions were responsible for the anomalies in the CTE of composites differs from that of the particulate compos- the composite [140]. TMC significantly affects the thermal ites as they show only very small residual strains when expansion behaviour of boron-fibre-reinforced AMC by they are cooled down from the peak temperature to room inducing interfacial degradation, which increases with temperature [111]. When the fibre content of continuous the number of cycles. Thermal expansion behaviour of the Mo-fibre-reinforced AMCs produced by diffusion bonding composite would depend on the thermal expansion of the was increased, the CTE of the composites decreased as fibres, when there is no interfacial sliding or fibre fracture close to the values of Mo fibres. Moreover, the CTE of the observed during CTE measurement [102]. unidirectional composite was found to increase with the Hybrid Al composites reinforced with SiC foam and increase in temperature, whereas in the case of dual direc- SiC particles developed using squeeze-casting technol- tional composites, with increasing temperature, the CTE ogy displayed lower CTE values than did those of SiC-par- first decreased due to the expansion constraints created ticle-reinforced aluminium composite with the same SiC by large accumulated thermal stress, then after reaching a content. This was attributed to the characteristic interpen- minimum at about 250°C, it starts increasing, which may be etrating structure of hybrid composite, which makes them attributed to matrix yielding and interfacial decohesion [1]. suitable to be used as an electronic packaging substrate Multiwalled carbon-nanotube-reinforced AMCs are material [141]. The strengthening due to CTE mismatch promising material for space structures and electronic was not observed in the case of a bulk particulate- packaging due to their low CTE value [147, 148]. Recently, reinforced nanostructured AMC that was fabricated by Lei et al. [149] studied the thermal expansion behaviour cryomilling and subsequent consolidation, since the for- of AMCs reinforced with hybrid (nanometre and micro- mation of geometrically necessary dislocations was con- metre) alumina particles and observed that nanoparticle strained to a nanostructured interfacial region containing concentration had an important effect on the CTE of the grains 30–50 nm in size [142]. The CTE of aluminium- composite. Stress relaxation process determines the CTE graphite composites fabricated by vacuum hot pressing of composites with lower nanoparticle concentration, was observed to decrease with increasing flake graphite. whereas percolation process determines the CTE of com- A control in CTE of the composite was achieved by control- posites with higher nanoparticle concentration. ling the graphite percentage [143]. Phase transformation of reinforcement particle can also affect the thermal expansion behaviour of AMC. In the case of β-eucryptic particle-reinforced AMCs prepared 7 Thermal fatigue by spark plasma sintering (SPS), it was observed that the metastable phase of eucryptic particle got transferred Thermal fatigue is caused mainly due to the cyclic change to normal eucryptic phase due to stress relaxation by of temperature and complete or partial restriction of annealing process. This transformation of the metastable thermal deformation due to external or internal limita- phase has a prominent effect on the thermal expansion tions, which can lead to fatigue cracking at locations with behaviour of the composite [144, 145]. stress concentration. The study of thermal fatigue behav- The thermal expansion behaviour of a carbon-fibre- iour of particulate-reinforced AMCs is of great importance reinforced Al composite is determined mainly by the since they are frequently used in components of engineer- thermal expansion of aluminium matrix and the restric- ing applications such as connecting rods and cylinder tion of carbon fibre through interfaces. This signifies the liners that may undergo thermal fatigue during service importance of interface on the thermal expansion behav- due to internal and external restriction against thermal iour of composites [146]. The CTE has close relationship expansion and contraction [150, 151]. It is one of the fore- with the changing rate of TMS and the dislocation density most design considerations in diesel engines [152]. of matrix. This was observed in the case of SiC-whisker- Generally, the fatigue behaviour of discontinuously reinforced aluminium composite where the composites reinforced MMCs is characterized by the competition were cooled down from 580°C with lower and higher between the cyclic hardening process and the cyclic K. Dash et al.: The behaviour of aluminium matrix composites under thermal shock-a review 13 softening process, which occurs at elevated tempera- damage was initiated at the intersection region of laser ture. Formation of new dislocations in the matrix and irradiated brim region and the perpendicular direction of interaction among them constitute the cyclic hardening tensile load [163]. behaviour, whereas interface failure, matrix cavitation, The crack initiation of SiC-particulate-reinforced alu- precipitate coarsening, and the rearrangement and anni- minium alloy composite subjected to fully reversed axial hilation of dislocations constitutes the cyclic softening fatigue tests was found to be dependent on temperature process [151]. The dependence of fatigue behaviour on and particle size [153]. In the numerical investigation temperature and particle size at elevated temperatures made on alumina-reinforced AMC under low cycle fatigue, is attributed mainly to the cyclic softening taking place, it was found that stress and strain concentrations in the which results in reduction of strength in the matrix [153]. microstructure play a major role in the formation of extru- Combinations of cyclic hardening and cyclic softening sions and intrusions, which subsequently leads to fatigue to failure were observed in aluminium alloy matrix dis- crack initiation [164]. Particle fracture, debonding, and continuously reinforced with SiC particles during fully matrix cracking were the chief damage mechanisms that reversed strain amplitude control [154–156]. cause fatigue damage in alumina-particulate-reinforced Cyclic softening was observed in an alumina-rein- 6061 Al matrix composites. Short fatigue crack initia- forced 6061-T6 aluminium alloy composite at lower strain tion and extension can occur due to damage localization amplitudes, independent of test temperature. At high caused by lump of alumina reinforcement particles. The strain amplitude, the composite displayed cyclic harden- hysteresis loop variation of strain-controlled cylindri- ing. Also, the low cycle fatigue behaviour of the composite cal specimens subjected to low cycle fatigue test pointed was inferior to that of the unreinforced 6061-T6 Al alloy, out cyclic plastic strain as the suitable low cycle fatigue especially in the short-life, high-amplitude region. The damage parameter [165]. cyclic yield stress decreased with the test temperature. In The fatigue crack propagation in AMCs takes place the high-cycle-fatigue region, fatigue lives are generally preferentially through the matrix region. A decrease in the controlled by strength [8]. Submicron-scale alumina-par- microhardness value of composites promotes increased ticle-reinforced aluminium alloy fabricated using powder crack propagation speed [166]. Higher fatigue life and metallurgy displayed cyclic hardening behaviour under tensile strength were observed for 2124 aluminium alloy tension and cyclic softening behaviour under compres- reinforced with SiC particles when compared with the sion at small strain amplitudes during low-cycle fatigue unreinforced Al alloy. By decreasing SiC particle size and testing. Rapid stabilization of cyclic deformation was increasing SiC volume fraction, the strength and fatigue life observed in the composite, which produced symmetric of the composite were found to increase. Size and volume stress-strain hysteresis loops at higher strain amplitudes. fraction of SiC particles, stress intensity maximum at crack Bauschinger effect was observed at small strain ampli- tip, and matrix strength govern the frequency of particle tudes due to constrained cyclic deformation created by fracture during fatigue crack propagation [167]. The cyclic the distribution of particle clusters [157]. plastic deformation taking place in the fatigue-damaged The main thermal fatigue fracture mechanisms oper- zone is considered to be the main mechanical driving ating in AMCs are interfacial decoherence between the force for the fatigue crack growth [168, 169]. Retardation of particle and the matrix, particle breakage, and crack prop- fatigue crack growth was observed when crack propagated agation. Fatigue crack initiation was always generated from low-high volume fraction of SiC, and the deflection of in the inclusion or particle cluster area [8, 102, 158, 159]. cracks also decreased the crack growth rate [170, 171]. Particle breakage was rare at higher temperatures, and Thermal fatigue crack initiation and propagation in a most of the cracks were nucleated in the matrix [151, 160, SiCp-Al composite fabricated by infiltration technique take 161]. In the fatigue behaviour study of AMCs reinforced place mainly by fracturing of the large reinforcing parti- with SiC particles, thin matrix ligaments that separate cles in the composite by thermal shock. SEM micrographs the uncracked subsurface particles from free specimen representing the initiation and propagation of thermal surface were found to be the fatigue crack initiation sites fatigue crack in an Al matrix containing 65 vol% SiC and for specimens with smaller size SiC particles. However, 35 vol% SiC are shown in Figure 6. Also, it was observed for specimens with larger size SiC particles, crack initia- that by reducing the size of the reinforcing particles, the tion occurred within the particle [162]. Thermal fatigue thermal fatigue resistance of the composite was found to damage induced in SiC-particulate-reinforced 6061 Al be increased [150]. composite as a result of repeated pulsed laser heating and Debonding and cracking of the interface were the mechanical load was studied, and it was found that the main degradation mechanisms associated with the zero 14 K. Dash et al.: The behaviour of aluminium matrix composites under thermal shock-a review

A B

C

Figure 6 SEM micrographs showing (A) crack initiation. (B) Propagation of thermal fatigue crack from notch tip in 65 vol% SiC-reinforced composite. (C) Propagation of thermal fatigue crack in 35 vol% SiC-reinforced composite. Reprinted from Ref. [150], Copyright (2006) with permission from Elsevier. load thermal fatigue testing of AMCs with weak inter- compressive mean stress and tensile mean stress were face, whereas matrix plastic deformation and interfacial generated during IP and OP TMF, respectively [174]. In the product formation were the degradation mechanisms case of thermal fatigue tests conducted on the aluminium associated with that of strong interface. All these degra- alloy EN AW-6061-T6 reinforced with Al2O3 particles fabri- dation mechanisms are attributed to the CTE mismatch cated using stir casting, more damage was observed near between the fibre and the Al matrix [45, 81, 172]. Powder the particles due to thermal and mechanical mismatch metallurgy processing was used to fabricate AMC with of the phases. Four mechanisms causing damage and nickel-titanium (Ni-Ti) shape memory alloy as the rein- residual stress development were reported, which include forcement phase. On comparison with the unreinforced thermally induced global deformation due to inhomo- Al, the fatigue life, yield strength, and ultimate strength geneous distribution of temperature, thermally induced of the composite were found to be enhanced due to the local deformation due to CTE mismatch, mechanically strengthening effect produced as a result of shape memory induced local deformation due to different deformation effect exhibited by the Ni-Ti alloy. Due to shape memory behaviour of both phases, and overaging. Global deforma- effect, residual internal stresses were created around each tion was the dominant mechanism in nonreinforced alloy, particle, which resulted in the strengthening of the com- whereas in composites, local mechanisms were signifi- posite [173]. cant [175]. Thermal fatigue of short-fibre-reinforced alu- Cyclic softening was observed in SiC-whisker-rein- minium AE42 alloys was investigated with an emphasis forced 6061 Al matrix composites during out-of-phase on the changes in the strain and hardness before and after (OP) and in-phase (IP) thermomechanical fatigue (TMF) thermal cycling. It was observed that the thermal strain tests. Thermal and mechanical cyclic strains are applied was affected by the experimental condition of the thermal simultaneously to the material during these tests. Also, cycling and the strength of matrix. After thermal cycling, a K. Dash et al.: The behaviour of aluminium matrix composites under thermal shock-a review 15 decrease in hardness was observed due to the occurrence and the size of the inclusions due to a finite metal-ceramic of matrix overaging and recovery [176, 177]. interface thermal resistance [178, 180–182]. By examining the literature, it has been observed that the most common ceramic particle used for reinforcing Al matrix is silicon carbide (SiC) because of its low cost and 8 Thermal conductivity low CTE [45, 178]. A low value of thermal conductivity was observed in the SiC-particulate-reinforced AMC produced One of the major advantages offered by MMCs is that by by plasma spraying of the ball-milled pure Al powders suitably selecting the reinforcement phase and matrix with 55 and 75 vol% SiC particles. Thermal boundary variables, the thermal properties of the composite can resistance offered by the interfacial gaps formed at the be tailored in such a way as to make it appropriate for Al-SiC interface was found to be the dominant factor electronic packaging applications. Particularly, there responsible for lowering the value of thermal conductiv- should be proper matching between the CTE of materi- ity. Porosity, iron impurity, longer ball-milling time, and als used for heat dissipation and that of semiconductors finer SiC size also lower the thermal conductivity of the and ceramic insulators in electronic packaging to prevent composite. Hot rolling and hot isostatic pressing treat- solder-joint failure during thermal cycling [178, 179]. For ment can be performed to improve thermal conductivity example, the heat generated due to high power densities as they improve the interfacial bonding and porosity of of power electronic devices such as insulated gate bipolar the composites. The decline in thermal conductivity value transistor should be transported from the ceramic chips due to the finer SiC particle size is due to the fact that as to the ceramic substrate and then through the solder and particle size becomes small, there will be a large contribu- the base plate into a heat sink. Here, if there is no proper tion of interface to thermal resistance [180, 183]. matching of CTE between the components, then it may The effect of SiC particle size distribution on the lead to delamination of the solder when there is a change thermal conductivity of the Al matrix composite was in temperature [45]. The CTE of metals can be lowered by studied by Molina et al. [184], where monomodal and reinforcing them with low CTE fillers. Ceramics such as bimodal size distributions of SiC particles were compared. SiC and AlN are suitable for this purpose as they have both In the case of composites based on powders with mono- high thermal conductivity and low CTE. modal size distribution, a steady increase in thermal con- The study of thermal conductivity of AMCs is impor- ductivity was observed. For bimodal particle mixtures, tant since they are increasingly considered as a promis- thermal conductivity started increasing with increasing ing material for thermal management applications. Al volume fraction of coarse particles and then reached a has high thermal conductivity and also high value of CTE. roughly constant value. A theoretical model based on For lowering the CTE of Al, it has to be reinforced with equivalent diameter approach was successful in predict- a higher volume fraction of filler material. Liquid metal ing whether the composite system such as Al-SiCP with infiltration is thus a suitable fabrication process for AMCs multimodal particle mixtures can acquire an appropriate since it allows high reinforcement volume fractions (up distribution design for attaining better thermal conduc- to 70%) in addition to its capability to produce a near-net tion properties [185]. shape. AMCs with high volume fractions of SiC, AlN, etc., In the case of SiCp-Al composites with high particle are used in a number of electronic applications. Increas- volume fraction fabricated by SPS [186], a small portion ing volume fraction of these ceramic reinforcements in the of pores can always appear as an unavoidable phase Al matrix also increases its brittleness, which is accept- because of the nonwetting nature of SiC and aluminium, able for electronic applications. In addition to liquid metal resulting in weak interfaces and incomplete sintering. infiltration, other liquid phase techniques such as stir The effect of porosity on the thermal conductivity of SiC- casting and compo casting can also be used for the fab- particulate-reinforced AMC was analysed numerically rication of AMCs with high volume fraction of reinforce- using a model based on the effective medium approxima- ments. Using the powder metallurgy route, fabrication of tion (EMA) scheme. Porosity can intensely deteriorate the AMCs with high reinforcement volume fractions is diffi- thermal conductivity of the composite by scattering the cult since the formation of protective alumina layer on the flow of heat. The EMA approach was found to be in good surface of each Al particle hinders the sintering process. agreement with the experimental data, especially with The thermal conductivity of composites relies on a variety porosity below 10%, but for higher levels, predictions of parameters, which includes the thermal conductivity of were a bit higher compared with the experimental values. the constituent phases, their volume fraction and shape, Furthermore, microstructural features such as voids and 16 K. Dash et al.: The behaviour of aluminium matrix composites under thermal shock-a review matrix-particle interface status also contribute to decrease fabricated in a solid-liquid coexistent state from the in thermal conductivity as they scatter the heat flow powder mixture of diamond, pure Al, and 5056-type around them [187]. Al-Mg alloy by the SPS process. The thermal conductivity Bad wettability and formation of undesirable interfa- of the Al matrix composite containing 45.5 vol% diamond cial brittle reaction product (Al4C3) are the main problems reached 403 W/mK, approximately 76% the theoretical faced with the processing of SiC-particulate-reinforced thermal conductivity estimated by Maxwell-Eucken’s aluminium composites. Aluminium carbide (Al4C3) equation. No interfacial reaction was observed between weakens the interface and by reacting with atmosphere the diamond particle and the Al matrix. moisture degrades thermal properties. A steady increase in thermal conductivity was observed in AMCs reinforced with mixtures of diamond 3SiC +→4Al3Si+ Al C 43 and SiC particles of equal size, which were fabricated by gas pressure-assisted liquid metal infiltration. Replac- The Si, which is the other reaction product form, also ing SiC gradually by diamond particles results in this creates nonuniformity in phase distribution and mechani- increase, and thermal conductivity of the nominally pure cal property distribution by dissolving in the aluminium aluminium matrix is influenced by take-up of some silicon matrix and substantially lowering its melting point. The from the silicon carbide [206]. Vacuum hot pressing is wettability of SiCp-Al composites can be enhanced by the better than SPS to fabricate diamond-Al composites with co-addition of Si and Mg to it, but care must be taken to high thermal conductivity for heat management appli- maintain a balance between the wettability improve- cations [207]. Also, the contact time between melt and ment and prevention of interfacial reaction for obtaining diamond reinforcement during liquid infiltration process- optimum properties. Using Al-Si alloy as the matrix is ing and the addition of silicon to the Al matrix material one of the effective ways to reduce the interfacial reaction were found to significantly affect the thermal conductivity as the Si present in the alloy promotes an opposite reac- of the composites by modification of the interface [179]. tion to that of interfacial reaction and suppresses it. But the disadvantage is that since Al-Si alloy is less ductile compared with aluminium matrix, it will decrease the mechanical properties of the composite [178, 181]. The 9 Conclusions thermal conductivity of SiC-particle-reinforced Al matrix composite decreased with increasing the amount of SiC A detailed review was carried out on the behaviour of and increased with increasing SiC particle size [188]. AMCs under different kinds of thermal stresses, and the Aluminium-nitride-reinforced AMCs offer better following conclusions were made: thermal conductivity than SiC-Al composites do due to 1. Residual thermal stresses generated in AMCs increase nonreactivity of AlN with aluminium. The thermal con- with increasing volume fraction of the reinforcement ductivity of aluminium-nitride-particle-reinforced com- and decrease with increasing interparticle spac- posites is very sensitive to the relative packing density of ing. They improve the hardness and strength when the composite [138, 178, 189]. the AMC is cooled from a low fabrication tempera- Alumina also does not react with aluminium, but its ture. But when the composite is cooled from a high agglomeration tendency and low value of thermal con- fabrication temperature, residual stress relaxation ductivity limit its usage [178]. Thermal conductivity was mechanisms become prominent and plastic deforma- found to vary linearly with temperature in the case of Al- tion takes place, which results in premature yield-

Al2O3-MMC prepared by powder metallurgy method. Also, ing prior to loading. The magnitude of the thermal the thermal conductivity of composites decreases with stress also increases with the increase in processing increasing Al2O3 volume fraction and decreasing Al2O3 temperature. particle size [190]. The thermal conductivity of the Al2O3- 2. The internal stresses generated, which either block reinforced Al-AlN matrix composite showed a marginal or assist the motion of individual dislocations, deter- decrease with an increase in temperature up to 150°C. mine the thermal cycling behaviour of AMCs. Strength, Carbon fibres, carbon nanotubes, and diamond parti- damping capacity, and dimensional stability of AMCs cles have also been used as filler material for AMCs in a are improved by thermal cycling treatment when the number of studies [13, 143, 179, 191–204]. internal stresses generated due to them blocks the Mizuuchi et al. [205] measured the thermal con- dislocation movement. On the other hand, thermal ductivity in diamond particle dispersed AMCs that were stresses produced by thermal cycling can increase the K. Dash et al.: The behaviour of aluminium matrix composites under thermal shock-a review 17

plasticity at the interface, resulting in plastic defor- [3] Tan MJ, Chew MC, Hung NP, Sano T. J. Mater. Process. Technol. mation at stresses lower than their yield stress. In 1997, 67, 62–66. [4] Degischer HP, Prader P, Marchi CS. Compos. Part A 2001, 32, combination with relatively small applied loads, ther- 1161–1166. mal cycling can result in superplastic deformation of [5] Surappa MK. Sadhana Acad. Proc. Eng. Sci. 2003, 28, the composite to very large strains ( > 300%). 319–334. 3. The CTE of AMCs can be lowered by increasing the [6] Cayron C, http://cimewww.epfl.ch/people/cayron/Fichiers/ volume fraction of the reinforcement and decreasing thesebook-chap1.pdf. Accessed July 23, 2013. the particle size. The CTE of composite is determined [7] http://libback.uqu.edu.sa/hipres/Indu/indu10542.pdf. Accessed July 23, 2013. mainly by the thermal expansion and heterogeneities [8] Perng CC, Hwang JR, Doong JL. J. Compos. Sci. Technol. 1993, of the aluminium matrix. 49, 225–236. 4. The thermal fatigue resistance of AMC can be [9] Fei WD, Hu M, Yao CK. Mater. Chem. Phys. 2002, 77, 882–888. increased by increasing the reinforcement volume [10] Chen YC, Daehn GS, Wagoner RH. Scripta Metal. Mater. 1990, fraction and decreasing the particle size. During ele- 24, 2157–2162. [11] Tjong SC, Tam KF, Wu SQ. Compos. Sci. Technol. 2003, 63, vated temperatures, cyclic softening was observed 89–97. at low strain amplitudes and cyclic hardening was [12] Sayman O, Ozer MR. Compos. Sci. Technol. 2001, 61, prominent at high strain amplitudes. Fatigue crack 2129–2137. initiation always occurred in the inclusion or particle [13] Mizuuchi K, Inoue K, Agari Y, Morisada Y, Sugioka M, Tanaka M, cluster area and the fatigue fracture mechanisms were Takeuchi T, Kawahara M, Makino Y. Compos. Part B 2011, 42, crack propagation, particle breakage, and interfacial 1029–1032. [14] Ozdemir I, Onel K. Compos. Part B 2004, 35, 379–384. decoherence between the particle and the matrix. [15] Nam TH, Requena G, Degischer P. Compos. Part A 2008, 39, 5. The thermal conductivity of AMCs decreased with 856–865. an increase in reinforcement volume fraction. The [16] Chen N, Zhang H, Gu M, Jin Y. J. Mater. Process. Technol. 2009, reduction in particle size also leads to a decrease in 209, 1471–1476. thermal conductivity due to the large contribution of [17] Rao VV, Murthy MVK, Nagaraju J. J. Compos. Sci. Technol. 2004, 64, 2459–2462. interfacial thermal resistance. The presence of pores [18] Rosso M. J. Mater. Process. Technol. 2006, 175, 364–375. and voids scatter heat flow and severely lowered the [19] Chawla KK, Chawla N. http://enpub.fulton.asu.edu/chawla/ thermal conductivity of the AMCs. papers/Wileymmcfinal.pdf. Accessed December 3, 2013. [20] Agrawal P, Conlon K, Bowman KJ, Sun CT, Cichocki FR, Trumble KP. Acta Mater. 2003, 51, 1143–1156. [21] Shieu FS, Raj R, Sass SL. Acta Metall. Mater. 1990, 38, 10 Future research scope 2215–2224. [22] Ho S, Saigal A. Acta Metall. Mater. 1994, 42, 3253–3262. Tailoring the interfacial thermal mismatch and multiply- [23] Douina J, Donnadieub P, Finela A, Dirrasc GF, Silvaind JF. Com- pos. Part A 2002, 33, 1397–1401. ing or reducing the thermal stresses of AMCs for achieving [24] Daehn GS, Hebbar K, Suh YS, Zhang H. Compos. Eng. 1993, 3, desired properties are areas facing dearth in the litera- 699–713. ture. Proper understanding of the mechanism of thermal [25] Tsai SD, Mahulikar D, Marcus HL. Mater. Sci. Eng. 1981, 47, conductance is required for meeting the ever-growing 145–149. demands of the electronics industry. Thermal history of [26] Shorshorov MKH, Ustinov LM, Kuznetsov YUG, Vinogradov LV, advanced MMCs with hybrid reinforcement distribution Jamnova VI. Composites 1976, 7, 17–20. [27] Vedula M, Pangborn RN, Queeney RA. Composites 1988, 19, 1. also lacks detailed study. [28] Tang T, Horstemeyer MF, Wang P. Int. J. Eng. Sci. 2012, 51, 161–167. Acknowledgments: The authors acknowledge the National [29] Sayman O. Compos. Part B 2005, 36, 61–72. Institute of Technology, Rourkela, for providing the neces- [30] Zywicz E, Parks DM. J. Compos. Sci.Technol. 1988, 33, 295–315. sary financial and infrastructural support. [31] Vedula M, Pangborn RN, Queeney RA. Composites 1988, 19, 2. [32] Liu HT, Sun LZ. Int. J. Solids Struct. 2004, 41, 2189–2203. [33] Arsenault RJ, Taya M. Acta Metall. 1987, 35, 651–659. [34] Withers PJ, Stobbs WM, Pedersen OB. Acta Metall. 1989, 37, 3061–3084. References [35] Lilholt H. Mater. Sci. Eng. A 1991, 135, 161–171. [36] Povirk GL, Needleman A, Nutt SR. Mater. Sci. Eng. A 1990, 125, [1] Chen LG, Lin SJ, Chang SY. Compos. Sci.Technol. 2006, 66, 129–140. 1793–1802. [37] Povirk GL, Needleman A, Nutt SR. Mater. Sci. Eng. A 1991, 132, [2] Mykura H, Mykura N. Compos. Sci.Technol. 1992, 45, 307–312. 31–38. 18 K. Dash et al.: The behaviour of aluminium matrix composites under thermal shock-a review

[38] Li DS, Wisnom MR, Dingley DJ. Compos. Sci.Technol. 1991, 42, [75] Girard C, Fougeres R, Vincent A. J. Alloy Compd. 1994, 413–427. 211/212, 169–172. [39] Jiang Z, Li G, Lian J, Ding X, Sun J. Compos. Sci.Technol. 2004, [76] Pickard SM, Derby B. Mater. Sci. Eng. A 1991, 135, 213–216. 64, 1661–1670. [77] Aria FR, Lichti T, Gagnon G. Scripta Metall. Mater. 1993, 28, [40] Meijer G, Ellyin F, Xia Z. Compos. Part B 2000, 31, 29–37. 587. [41] Dutta I. Compos. Sci.Technol. 1991, 41, 193–213. [78] Dunn ML, Taya M. Mater. Sci. Eng. A 1994, 176, 349–355. [42] Sajjadi SA, Ezatpour HR, Parizi MT. Mater. Des. 2012, 34, 106–111. [79] Pal S, Bhanuprasad VV, Mitra R. Mater. Sci. Eng. A 2008, 489, [43] Allen AJ, Bourkei AM, Dawesi S, Hutchings T, Withers PJ. Acta 11–20. Metall. Mater. 1991, 40, 2361–2372. [80] Colclough A, Dempster B, Favry Y, Valentin D. Mater. Sci. Eng. [44] Fitzpatrick ME, Withers PJ, Baczmanski A, Hutchings MT, A 1991, 135, 203–207. Levy R, Ceretti M, Lodini A. Acta Mater. 2002, 50, 1031–1040. [81] Ghorbel E. Compos. Sci. Technol. 1997, 57, 1045–1056. [45] Schobel M, Altendorfer W, Degischer HP, Vaucher S, Buslaps T, [82] Kehoe FP, Chadwick GA. Mater. Sci. Eng. A 1991, 135, 209–212. Michiel MD, Hofmann M. Compos. Sci. Technol. 2011, 71, [83] Herr AE, Canumalla S, Pangborn RN. Mater. Sci. Eng. A 1995, 724–733. 200, 181–191. [46] Shee SK, Pradhan SK, De M. Mater. Chem. Phys. 1998, 52, [84] Wetland A, Johannesson T. Mater. Sci. Eng. A 1995, 190, 228–234. 131–142. [47] Nakamura T, Suresh S. Acta Metall. Mater. 1993, 41, 1665–1681. [85] Eggeler GF, Earthman JC. Scripta Mater. 1998, 38, 341–348. [48] Dutta I, Sims JD, Seigenthaler DM. Acta Metall. Mater. 1993, 41, [86] Kuo SL, Kao PW. Mater. Chem. Phys. 1999, 58, 83–86. 885–908. [87] Wu SQ, Wei ZS, Tjong SC. Compos. Sci.Technol. 2000, 60, [49] Ho S, Saigal A. Mater. Sci. Eng. A 1994, 183, 39–47. 2873–2880. [50] Jeong GS, Allen DH, Lagoudas DC. J. Solids Struct. 1994, 31, [88] Dutta I. Acta Mater. 2000, 48, 1055–1074. 2653–2677. [89] Badini C, Vecchia MLA, Giurcanu A, Wenhui J. J. Mater. Sci. [51] Gupta N, Nguyenb NQ, Rohatgi PK. Compos. Part B 2011, 42, 1997, 32, 921–930. 916–925. [90] Ozdemir I, Toparli M. J. Compos. Mater. 2003, 37, 1839–1850. [52] Xuea J, Wangb WX, Takao Y, Matsubara T. Compos. Part A 2011, [91] Etter T, Papakyriacou M, Schulz P, Uggowitzer PJ. Carbon 42, 986–992. 2003, 41, 1017–1024. [53] Hanus E, Ericsson T. Mater. Sci. Eng. A 1995, 90, 155–163. [92] Zhao M, Wu G, Zhu D, Jiang L, Dou Z. Mater. Lett. 2004, 58, [54] Lee RS, Chen GA, Hwang BH. Composites 1995, 26, 425–429. 1899–1902. [55] Singh SB, Ray S. J. Mater. Process Technol. 2003, 143–144, [93] Chen AN, Arai Y, Tsuchida E. Compos. Part B 2005, 36, 623–628. 319–330. [56] Altan G, Topcu M, Bektas NB, Altan BD. J. Mech. Eng. Sci. 2008, [94] Zhang H, Gu M. J. Alloy Comp. 2006, 426, 247–252. 22, 2318–2327. [95] Huang YD, Hort N, Kainer KU. Compos. Part A 2004, 35, [57] Bouafia F, Serier B, Abbes B, Bouiadjra B. Compos. Mater. Sci. 249–263. 2012, 54, 195–203. [96] Ozdemir I, Toparli M, Onel K, Tsunekawa Y. J. Compos. Mater. [58] Guagliano M. J. Mater. Eng. Perform. 1998, 7, 183–189. 2005, 39, 601–615. [59] Morelli1 EC, Urreta SE, Schaller R. Acta Mater. 2000, 48, [97] Durieux S, Buffiere JY, Lormand G, Rapoport A, Vincent A. 4725–4733. Mater. Sci. Eng. A 1997, 234, 953–957. [60] Wu PL, Tian Z, Wang LD, Fei WD. Thermochim. Acta 2007, 455, [98] Requena G, Yubero DC, Corrochano J, Repper J, GArces G. 7–10. Compos. Part A 2012, 43, 1981–1988. [61] Furness JAG, Clyne TW. Mater. Sci. Eng. A 1991, 141, 199–207. [99] Uju WA, Oguocha INA. Mater. Sci. Eng. A 2009, 526, 100–105. [62] Flour JCL, Locicero R. Scripta Metall. Mater. 1987, 21, 1071– [100] Rioa ED, Nasha JM, Williams JC, Breslin MC, Daehn GS. Mater. 1076. Sci. Eng. A 2007, 463, 115–121. [63] Zhang H, Daehn GS, Wagoner RH. Scripta Metall. Mater. 1990, [101] Daguanga L, Guoqin C, Longtao J, Ziyang X, Yunheb Z, 24, 2151–2155. ­Gaohui W. Mater. Sci. Eng. A 2013, 586, 330–337. [64] Chen YC, Daehn GS. Scripta Metall. Mater. 1991, 25, 1543–1548. [102] Badinia C, Finoa P, Mussob M, Dinardo P. Mater. Chem. Phys. [65] Sherby OD, Wadsworth J. Prog. Mater. Sci. 1989, 33, 169–221. 2000, 64, 247–255. [66] Hall IW, Patterson WG. Scripta Metall. Mater. 1991, 25, [103] Qin YC, He SY, Yang DZ. Mater. Chem. Phys. 2004, 86, 805–810. 204–209. [67] Pickardi SM, Derby B. Acta Metall. mater. 1991, 38, 2537–2552. [104] Fei WD, Wang LD. Mater. Chem. Phys. 2004, 85, 450–457. [68] Cheong YM, Marcus HL. Scripta Metall. Mater. 1987, 21, [105] Smagorinski ME, Tsantrizos PG, Grenier S, Cavasin A, 1529–1534. ­Brzezinski T, Kim G. Mater. Sci. Eng. A 1998, 244, 86–90. [69] Daehn GS, Oyama T. Scripta Metall. Mater. 1988, 22, 1097–1102. [106] Qing CG, Shu YW, Kang MA, Hussain M, Tao JL, Hui WUG. [70] Doncel GG, Karmarkar SD, Divecha AP, Sherby OD. Compos. Trans. Nonferrous Met. Soc. China 2011, 21, 262–273. Sci. Technol. 1989, 35, 105–120. [107] Zhiwu X, Jiuchun Y, Weiwei Z, Huibin X, Shiqin Y. Compos. [71] Daymond MR, Withers PJ. Scripta Mater. 1996, 35, 1229–1234. Sci. Technol. 2005, 65, 1461–1467. [72] Daymond MR, Withers PJ. Scripta Mater. 1996, 35, 717–720. [108] Wang T, Fan T, Zhang D, Zhang G, Xiong D. Mater. Des. 2008, [73] Daymond MR, Withers PJ. Appl. Compos. Mater. 1997, 4, 29, 1275–1279. 375–393. [109] Wu My, Sherby OD. Scripta Metall. Mater. 1984, 18, 773–776. [74] Hong SI, Gray GT, Vecchio KS. Mater. Sci. Eng. A 1993, 171, [110] Bohm HJ, Degischer HP, Lacom W, Qu J. J. Compos. Eng. 1995, 181–189. 5, 37–49. K. Dash et al.: The behaviour of aluminium matrix composites under thermal shock-a review 19

[111] Kustov S, Golyandin S, Sapozhnikov K, Vincent A, Maire E, [148] Liu ZY, Xiao BL, Wang WG, Ma ZY. Compos. Sci. Technol. 2012, Lormand G. Mater. Sci. Eng. A 2001, 313, 218–226. 72, 1826–1833. [112] Vaidya RU, Chawla KK. Compos. Sci. Technol. 1994, 50, 13–22. [149] Lei Z, Zhao K, Wang Y, An L. J. Mater. Sci. Technol. 2013. [113] Neite G, Mielke S. Mater. Sci. Eng. A 1991, 148, 85–92. In Press, Corrected Proof, DOI: 10.1016/j.jmst.2014.04.011 [114] Gudlur P, Forness A, Lentz A, Radovic M, Muliana A. Mater. Available online 6 May 2014. Sci. Eng. A 2012, 531, 18–27. [150] Wu CML, Han GW. Compos. Part A 2006, 37, 1858–1862. [115] Uju WA, Oguocha INA. Mater. Des. 2012, 33, 503–509. [151] Llorca J. Int J. Fatigue 2002, 24, 233–240. [116] Kima BG, Donga SL, Park S. Mater. Chem. Phys. 2001, 72, [152] Song HW, Yua G, Tan JS, Zhou L, Yu XL. Int. J. Heat Mass Trans- 42–47. fer 2008, 51, 757–767. [117] Kreider KG, Patarini VM. Metall. Trans. 1970, 1, 3431–3435. [153] Uematsu Y, Tokaji K, Kawamura M. Compos. Sci. Technol. [118] Chawla N, Deng X, Schnell DRM. Mater. Sci. Eng. A 2006, 426, 2008, 68, 2785–2791. 314–322. [154] Srivatsan TS, Hajri MA. Compos. Part B 2002, 33, 391–404. [119] Aswendt P, Hofling R. Composites 1993, 24, 611. [155] Srivatsan TS, Hajri MA, Hannona W, Vasudevan VK. Mater. Sci. [120] Lee YS, Gungor MN, Batt TJ, Liaw PK. Mater. Sci. Eng. A 1991, Eng. A 2004, 379, 181–196. 145, 37–46. [156] Zhang G, Li B, Zhang J, Cai W. Prog. Nat. Sci. 2012, 22, 153–159. [121] Chan KC, Liang J. Compos. Sci. Technol. 2001, 61, 1265–1270. [157] Gasemn ZM, Ali SS. Mater. Sci. Eng. A 2013, 562, 109–117. [122] Zhang Q, Wu G, Chen G, Jiang L, Luan B. Compos. Part A 2003, [158] Akiniwa Y, Machiya S, Tanaka K. Int. J. Fatigue 2006, 28, 34, 1023–1027. 1406–1412. [123] Dellis MA, Keustermans JP, Delannay F. Mater. Sci. Eng. A [159] Zhua SJ, Iizuka T. Compos. Sci. Technol. 2003, 63, 265–271. 1991, 135, 253–257. [160] Tjong SC, Wang GS. Mater. Sci. Eng. A 2004, 386, 48–53. [124] Karthikeyan B, Ramanathan S, Ramakrishnan V. Mater. Des. [161] Tjong SC, Wang GS, Mai YW. Compos. Sci. Technol. 2005, 65, 2010, 31, 82–86. 1537–1546. [125] Elomari S, Boukhili R, Lloyd DJ. Acta Mater. 1996, 44, [162] Lukasak DA, Koss DA. Composites 1993, 24, 3. 1873–1882. [163] Long SG, Zhou YC. Compos. Sci. Technol. 2005, 65, 1391–1400. [126] Elomari S, Skibo MD, Sundarrajan A, Richards H. Compos. Sci. [164] Schneidera Y, Soppa E, Kohler C, Mokso R, Roos E. Proc. Eng. Technol. 1998, 58, 369–376. 2011, 10, 1515–1520. [127] Shen YL. Mater. Sci. Eng. A 1998, 252, 269–275. [165] Li C, Ellyin F. Mater. Sci. Eng. A 1996, 214, 115–121. [128] Sastry S, Krishna M, Uchil J. J. Alloy Compd. 2001, 314, [166] Wei Li, Zhenhua C, Ding C, Cang F, Canrang W. J. Alloy Compd. 268–274. 2010, 504, 522–526. [129] Hu M, Fei WD, Yao CK. Scripta Mater. 2002, 46, 563–567. [167] Hall JD, Jones JW, Sachdev AK. Mater. Sci. Eng. A 1994, 183, [130] Ohnuki T, Tomota Y. Scripta Mater. 1996, 34, 713–720. 69–80. [131] Christman T, Suresh S. Acta Metall. 1988, 36, 1691–1704. [168] Ding HZ, Biermann H, Hartmann O. Compos. Sci. Technol. [132] Qin CG, Yang XZ, Shu YW, Tao JL, Hui WUG. Trans. Nonferrous 2002, 62, 2189–2199. Met. Soc. China 2010, 20, 2143–2147. [169] Ding HZ, Biermann H, Hartmann O. Int. J. Fatigue 2003, 25, [133] Wang LD, Fei WD, Yao CK. Mater. Sci. Eng. A 2002, 336, 209–220. 110–116. [170] Xua FM, Zhua SJ, Zhao J, Qi M, Wang FG, Li SX, Wang ZG. [134] Wang LD, Fei WD. Mater. Sci. Eng. A 2006, 433, 291–297. Mater. Sci. Eng. A 2003, 360, 191–196. [135] Arpon R, Molina JM, Saravanan RA, Cordovill CG, Louis E, [171] Sharma MM, Ziemian CW, Eden TJ. Mater. Des. 2011, 32, Narciso J. Acta Mater. 2003, 51, 3145–3156. 4304–4309. [136] Tjong SC, Tam KF. Mater. Chem. Phys. 2006, 97, 91–97. [172] Scholel M, Degischer HP, Vaucher S, Hofmann M, Cloetens P. [137] Zhang Q, Wu G, Jiang L, Chen G. Mater. Chem. Phys. 2003, 82, Acta Mater. 2010, 58, 6421–6430. 780–785. [173] Porter GA, Liaw PK, Tiegs TN, Wu KH. Mater. Sci. Eng. A 2001, [138] Couturier R, Ducret D, Merle P, Disson JP, Joubert P. J. Eur. Cer. 314, 186–193. Soc. 1997, 17, 1861–1866. [174] Qian LH, Wang ZG, Toda H, Kobayashi T. Mater. Sci. Eng. A [139] Ejiofor JU, Reddy RG. Mater. Sci. Eng. A 1999, 259, 314–323. 2003, 357, 240–247. [140] Huber T, Degischer HP, Lefranc G, Schmitt T. Compos. Sci. [175] Klaska AM, Beck T, Wannera A, Lohe D. Mater. Sci. Eng. A Technol. 2006, 66, 2206–2217. 2009, 501, 6–15. [141] Zhao LZ, Zhao MJ, Cao XM, Tian C, Hu WP, Zhang JS. Compos. [176] Huang YD, Hort N, Dieringa H, Maier P, Kainer KU. Int. J. Sci. Technol. 2007, 67, 3404–3408. Fatigue 2006, 28, 1399–1405. [142] Vogt R, Zhang Z, Li Y, Bonds M, Browning ND, Lavernia EJ, [177] Minaka M, Ceschini L, Boromei I, Ponte M. Int. J. Fatigue 2010, Schoenung JM. Scripta Mater. 2009, 61, 1052–1055. 32, 218–226. [143] Chen JK, Huang IS. Compos. Part B 2013, 44, 698–703. [178] Chung DDL. Appl. Therm. Eng. 2001, 21, 1593–1605. [144] Dong WL, Ye CUI, Tao YC, Peng WK, Dong FW. Trans. [179] Ruch PW, Beffort O, Kleiner S, Weber L, Uggowitzer PJ. ­Nonferrous Met. Soc. China 2011, 21, 280–284. ­Compos. Sci. Technol. 2006, 66, 2677–2685. [145] Wang LD, Xue ZW, Cui Y, Wang KP, Qiao YJ, Fei WD. Compos. [180] Gui M, Kang SK, Euh K. Scripta Mater. 2005, 52, 51–56. Sci. Technol. 2012, 72, 1613–1617. [181] Qu XH, Zhang L, Wu M, Ren SB. Prog. Nat. Sci. 2011, 21, 189–197. [146] He ZY, Hui WUG. Trans. Nonferrous Met. Soc. China 2010, 20, [182] Chu K, Jia C, Liang X, Chen H, Gao W, Guo H. Mater. Des. 2009, 2148–2151. 30, 4311–4316. [147] Deng CF, Ma YX, Zhang P, Zhang XX, Wang DZ. Mater. Lett. [183] Raj R, Saha A, An L, Hasselman DPH, Ernst P. Acta Mater. 2008, 62, 2301–2303. 2002, 50, 1165–1176. 20 K. Dash et al.: The behaviour of aluminium matrix composites under thermal shock-a review

[184] Molina JM, Narciso J, Weber L, Mortensen A, Louis E. Mater. [196] Prieto R, Molina JM, Narciso J, Louis E. Compos. Part A 2011, Sci. Eng. A 2008, 480, 483–488. 42, 1970–1977. [185] Chu K, Jia C, Tian W, Liang X, Chena H, Guo H. Mater. Des. [197] Zhu C, Ma N, Jin Y, Bai H, Ma Y, Lang J. Mater. Des. 2012, 41, 2009, 30, 3497–3503. 208–213. [186] Chu K, Jia C, Tian W, Liang X, Chena H, Guo H. Compos. Part A [198] Wu J, Zhang H, Zhang Y, Wang X. Mater. Des. 2012, 41, 344–348. 2010, 41, 161–167. [199] Wu J, Zhang H, Zhang Y, Li J, Wang X. Mater. Des. 2012, 39, [187] Lee JM, Lee SK, Hong SJ, Kwon YN. Mater. Des. 2012, 37, 87–92. 313–316. [200] Monje IE, Louis E, Molina JM. Compos. Part A 2013, 48, 9–14. [188] Kady OE, Fathy A. Mater. Des. 2014, 54, 348–353. [201] Tan Z, Li Z, Fan G, Kai X, Ji G, Zhang L, Zhang D. Diam. Relat. [189] Mizuuchi K, Inoue K, Agari Y, Nagaokaa T, Sugioka M, Mater. 2013, 31, 1–5. Tanaka M, Takeuchi T, Tani J, Kawahara M, Makinoc Y, Ito M. [202] Xue X, Yu JK. Surf. Coat. Technol. 2013, 217, 46–50. Compos. Part B 2012, 43, 1557–1563. [203] Shin Z, Yang W, Wang J, Liu G, Qiaon G, Jin Z. Ceram. Int. 2013, [190] Tatar T, Ozdemir N. Physica B 2010, 405, 896–899. 39, 3365–3370. [191] Hochengap H, Yenb SB, Ishihara T, Yen BK. Compos. Part A [204] Tan Z, Li Z, Xiong DB, Fan G, Ji G, Zhang D. Mater. Des. 2014, 1997, 28, 883–890. 55, 257–262. [192] Wang TC, Fan TX, Zhang D, Zhang GD, Xiong DS. Mater. Lett. [205] Mizuuchi M, Inoue K, Agari Y, Morisada Y Sugioka M, 2007, 61, 1849–1854. Tanaka M, Takeuchi T, Tani J, Kawahara M, Makino Y. Compos. [193] Bakshi SR, Patel RR, Agarwal A. Comp. Mater. Sci. 2010, 50, Part B 2011, 42, 825–831. 419–428. [206] Molina JM, Rheome M, Carron J, Weber L. Scripta Mater. 2008, [194] Xue C, Yu JK, Zhu XM. Mater. Des. 2011, 32, 4225–4229. 58, 393–396. [195] Fukuchi K, Sasaki K, Katagiri K, Imanishi T, Kakitsuji A. Proc. [207] Tan Z, Li Z, Fan G, Guoa Q, Kai X, Ji G, Zhang L, Zhang D. Mater. Eng. 2011, 10, 912–917. Des. 2013, 47, 160–166.