1 Introduction

Ceramic particle reinforced cast lightweight metal matrix composite (MMC) based on aluminum has provided a cheaper lightweight alternative engineering material to steel since late 60s of the last century. Increasing application of these composites in engineering components has helped to achieve better fuel economy in automobile, aero-space, and space industries. The cheapest of these composites was prepared by stirring reinforcing particles into aluminum alloy melt, generally containing surface active element magnesium to facilitate entry of poorly wetting particles into the melt.[1] The resulting slurry was cast to the desired shape of engineering components. Subsequently, composites based on magnesium alloys have also been developed offering further advantage in weight saving. Application of magnesium alloys and composites steadily increased in engineering components, particularly in automobiles.[2]

One of the initial problems encountered in cast MMCs reinforced by hard ceramic particles is inadequate ductility. Difference in deformation characteristics of hard particles and the surrounding metal matrix, results in progressively increasing mismatch in extension with application of increasing load during tensile testing, which, for coarser particles, often leads to early debonding of these particles, limiting both strength and ductility.[3] During early development of cast composites, ceramic particles of relatively coarse size (~80 μm) have been used for the ease of processing and so, ductility used to be very poor. However, the debonding could be delayed by (i) reducing the size of reinforcement to decrease mismatch in extension for a given stress and (ii) increasing the interface shear strength between the particle and the matrix.[4,5]

When the size of reinforcing ceramic particles is lower, the incorporation of particles into molten alloy becomes more difficult due to adverse surface tension of ceramic particles dominating over gravity since most of the ceramic particles are poorly wetted by the molten matrix alloy.[6] Further, the strengthening of these composites depends on the extent of transfer of stresses from the matrix to the stronger reinforcing particles. This in turn is limited by the strength of interfacial bond between the matrix and the reinforcement.[5] Both these requirements of ease of processing and strong interface could be fulfilled by tailoring the surfaces involved in the formation of interface.[4] Continuing efforts to find a processing route to fabricate a composite with lower reinforcement size led to emergence of in situ composites, where particles are not added externally but generated inside the matrix alloy. These particles are generally of sizes less than a micron and the strength of in situ composites improved significantly but ductility did not always improve.[7] There are some instances of improved ductility as well.[8]

The enhanced focus on nanoparticles at the turn of the century offered an opportunity to decrease the particle size further in cast MMC prepared by slurry casting. Since the nanoparticles may have increased surface energy compared to the bulk of the same material,[9] it may be expected that these particles will be easily wetted and have increased interfacial strength. But, high surface energy also gives rise to a tendency of agglomeration of particles creating difficulties in dispersing them to the level of individual particles in molten alloy. Use of intense vibration by ultrasonic method or electromagnetic method helps to reduce the size of the agglomerate[10] but the shear field gradient may not be strong enough on nano-metric scale so as to break the agglomerate of a few particles.

Strengthening by reinforcement of nano-size particles have been attributed to primarily (a) smaller inter-particle distance leading to enhanced Orowan strengthening[11] and (b) prevention of particle cracking.[12] Apart from strengthening, enhanced ductility is more widely observed although there are a few contradicting results.[13] Deng et al.[14] have observed that in 2024 aluminum alloy (∼50 μm) reinforced with 1 wt pct carbon nano-tubes (MWCNTs), Young’s modulus, yield strength, and tensile strength increase over those in the matrix alloy but the elongation of the composite decreases to 3 from 16.5 pct observed in the matrix alloy, which has also been observed by Laha et al.[15] in Al-Si alloy reinforced with multi-walled carbon nano-tubes (MWCNTs). Thus, debonding and the strength of interface are still critical issues. Agglomeration of nanoparticles to clusters could also be another important issue contributing to lower ductility.[16]

The present study involves solidification synthesis of cast composite based on aluminum alloy reinforced by two size classes of particles—(i) nanoparticles of alumina generated by chemical reaction of manganese dioxide with aluminum during high energy milling and also, inside molten alloy and (ii) coarser alumina particles of submicron size generated in the melt by internal oxidation of molten aluminum alloy. The effect of coarser particles to limit ductility in presence of nanoparticles has been examined by comparing the mechanical properties of cast composites containing both the size classes of particles with those containing only coarser particles.

2 Experimental Procedure

Nanoparticles of alumina have been generated by milling a mixture of aluminum and manganese dioxide powders in ratio of 1:7 by weight, as specified in Table I. A vibratory disk mill (RS200, Retsch, Germany) was used for milling in a medium of toluene for 450 minutes at 700 rpm. The resultant product has been characterized by SEM, TEM, and X-ray for size and chemical nature of particles.

Table I Details of MnO2 and Aluminum Powders Used
Full size table

About 750 g of commercially pure aluminum was melted and superheated to “1033.15 K (760.15 °C)” in a clay-graphite crucible inside a muffle furnace shown in the set-up schematically given in Figure 1. The weighed amounts of milled particles were added and dispersed in the melt by using coated stirrer having four pitched blades (45 deg pitch angle). The rate of addition of oxide particles was controlled at approximately 6 to 8 g/min. Magnesium lump weighing 5 wt pct of the alloy was plunged into the melt-particle slurry to improve the wettability of the melt for alumina. The composition of commercial aluminum and magnesium used is given in Table II. The melt was stirred to make the alloy slurry homogenous before it was cast by bottom pouring into split type permanent steel mold of size 60 × 40 × 130 mm3, coated with graphite and preheated. No degassing of the melt or the slurry was carried out at any stage of processing.

Fig. 1
figure 1

Schematic diagram showing experimental set-up for stir casting used for solidification processing of cast in situ nano-composites and unreinforced base alloys

Full size image
Table II Chemical Composition of the Commercial aluminum and Magnesium Used
Full size table

The composites synthesized following the above procedure, have been designated on the basis of constituents—the first letter A and the following letter M indicating the base metal aluminum and the alloying element of magnesium (5 wt pct.) followed by a number x indicating powder milling time. AMx is followed by letter P, indicating milled powder mix added and the amount in wt pct is given thereafter by y. For example, designation AM0P2 indicates Al-5 wt pct Mg alloy based composite synthesized by addition of 2 wt pct powder mix without any milling while AM4P2 refers to a composite based on the same alloy but synthesized by addition of 2 wt pct of powder mix milled for 450 min. Apart from nanoparticles in the powder mix, the composites also contain coarser particles formed by internal oxidation of alloy melt during processing. The base alloy of Al-5 wt pct has also been made by similar processing route for comparison of mechanical properties and it contains coarse alumina particles formed by internal oxidation.

The chemical composition of the matrix alloy of cast composites was analyzed by atomic absorption spectrometers (AAS, GBC AVANTA-M). X-ray diffraction study was carried out using X-ray diffractometer (D8 Advance, Brukes AXS, Karlsruhe, Germany) in the two theta range of 10 to 110 deg using CuK α radiation and nickel filter. Microscopy was conducted using field emission scanning electron microscope (FEI QUANTA 200 FEG, Czech Republic) and transmission electron microscope (FEI Technai 20) operated at 200 kV. The Brinell hardness of the cast composites was measured following ASTM-E10 standards using a 2.5 mm dia steel ball at a load of 153.3 N by Dia Tester (2RC, Democratic Republic of Germany). The hardness of cast in situ composite was measured along the height of the cast ingot at intervals of 15 mm by taking at least three indentations at a given height. The tensile tests were carried out under uniaxial tension to obtain engineering stress–strain curve following ASTM-E8M specification using a tensile testing machine (Hounsfield, Monsanto, H25KS/05, Surrey, England) at ambient temperature at a strain rate of 6.67 × 10−4 s−1 on at least three defect-free tensile specimens of 5.0 mm gauge diameter and 25 mm gauge length, machined out from the middle of cast composite ingot. The dimension and geometry of tensile specimens is given in Figure 2.

Fig. 2
figure 2

Shape and size of the tensile test specimen

Full size image

3 Results and Observation

3.1 Characterization of Milled Powders and Composites

The mix of aluminum and manganese dioxide powders after milling for 450 minutes has been examined by TEM and the bright field transmission electron micrograph has revealed sizes from 6 to 260 nm, as shown in Figure 3(a). Selected area electron diffraction (SAD) patterns shown in Figure 3(b), show sharp ring-spot pattern indicating the presence of aluminum, γ-Al2O3, and MnO2, spots marked by a, A, and M, respectively in the figure.

Fig. 3
figure 3

(a) TEM micrograph of powder mix after 450 min of milling and (b) selected area electron diffraction patterns showing the presence aluminum by rings 1, 2, and 3 and spots marked by a, A, and M for aluminum, γ-alumina, and manganese dioxide, respectively

Full size image

From a number of micrographs, a rough number distribution of sizes has been obtained by classifying the particles in three broad size classes—class A in the size range less than 10 nm, class B in the size range between 10 and 200 nm, and class C in the size range larger than 200 nm. Table III compares the size of powder mix before milling with those after milling. The initial powder mix had particles in the size class C but during milling, the powders, particularly the brittle oxide, got fragmented to smaller particles belonging to size classes B and C. The alumina particles formed by reaction at the point of high energy impact between aluminum and manganese oxide particles during milling are likely to be small and belong possibly to size class A. The particle in size class C has a large average size and these particles could be particles of metallic aluminum, which are more ductile.

Table III Size Distribution in Powder Mix Before and After Milling—Average Size and Number of Particles in Different Size Ranges
Full size table

During solidification synthesis of the composites, the remaining aluminum particles in the milled powder mix, when added to the molten alloy, melt and mix in the alloy while manganese dioxide particles remaining including the nano-sized ones, will react with molten alloy to form alumina-based particles and release manganese for alloying with the melt. The γ-Al2O3 particles formed during milling at separate locations surrounded by aluminum particles are expected to remain separated spatially when the powder mix is added to the melt. Although nanoparticles of alumina are expected to have better wettability due to their relatively higher surface energy compared to larger size particles, magnesium is also added to the melt since the coarser particles of alumina formed by internal oxidation of the melt is also to be wetted and retained. The designation of the cast composites synthesized for this study are given in Table IV.

Table IV Designation and Nominal Composition of the Composites
Full size table

The chemical composition of the matrix alloy as determined by AAS is given in Table V, and it shows that the magnesium content is around 5 wt pct as targeted. There is some iron pick-up from the coated stirrer to result in a small variation in the iron content of the alloy, increasing with the increasing amount powder mix added, because of consequent increase in stirring time. The manganese content in the matrix alloy increases from about 0.3 to 1.25 wt pct, almost in the same ratio of increasing addition of powder mix, from 1 to 4 wt pct.

Table V Chemical Composition of Matrix of Composites
Full size table

The FESEM micrograph of the composite shown in Figure 4(a) reveals the presence of the coarser submicron-sized oxide as bright particles, which have formed during solidification synthesis of composites and the TEM micrograph in Figure 4(b) clearly shows the nanoparticles, mostly occurring individually without agglomeration to a bigger cluster of particles indicating the success of innovative strategy of using powder mix in the synthesis of composites.

Fig. 4
figure 4

(a) FESEM micrographs (b) TEM micrograph of cast composites designated as AM4P2

Full size image

During melt processing, the ceramic particles added to the melt are often retained partially due to flotation along with bubbles.[1] The amount of MnO2 particles retained and reacted has been estimated by the amount of manganese released to the matrix of the composite as shown in Table VI. It is observed that manganese content of the matrix varies almost linearly with increasing MnO2 due to larger addition of milled powder, as shown in Figure 5. But for AM4P4, larger addition of milled powder may have led to relatively lower retention within the melt as indicated by manganese in the matrix, lower than that anticipated from linear trend. Similar amounts of manganese released in the matrix of composites AM4P2 and AM0P2, when the latter has been prepared by addition of coarse MnO2 powder, is an indication of almost complete reduction of MnO2 by aluminum melt. Linear trend in Figure 5 may be the result of retention of added MnO2 particles in the same proportion and similar extent of reaction of these particles. A little higher manganese in the matrix of AM0P2 is possibly due to a little higher retention of coarse particles, favoured by gravity. The amount of alumina generated by reaction of MnO2 particles has also been estimated and reported in Table VI. For those composites where nanoparticles of MnO2 have been added, these alumina particles are of nano-size. The total alumina content of the composite has been determined by the amount of particles extracted by dissolving the composite to leach out the alloy matrix. The amount of alumina generated by internal oxidation has been estimated by subtracting the amount of alumina generated by reaction of MnO2 particles from the total alumina content in a composite as extracted. The effect of compounds of magnesium and manganese are ignored as these elements are mostly in the matrix alloy of the composite as shown in Table V. The variation in the amount of extracted particles with increasing addition of powder mix in the composites is shown in Figure 6(a) in terms of wt pct of MnO2 in the mix. The amount of extracted particle is more than that added due to particles generated by internal oxidation but the extent of internal oxidation could be restricted by using a cover of argon gas during processing. Figure 6(b) shows the variation of porosity in cast composite, which increases from 2.61 to 6.79 vol pct with increasing addition of milled powder mix from 1 to 4 wt pct.

Table VI The Details of Particle Contents in Composites—Particles of Alumina Generated by Chemical Reaction of MnO2 and By Internal Oxidation
Full size table
Fig. 5
figure 5

Variation of weight percent of Mn released to the matrix alloy of composites synthesized by increasing addition of weight percent MnO2 particle in powder milled for 450 min (* indicates the result for AM0P2 prepared using MnO2 powder without milling)

Full size image
Fig. 6
figure 6

Variation of (a) total particle content of nanoparticles as well as larger particles and (b) porosity, in composites synthesized by increasing addition of milled powder mix

Full size image

The XRD pattern of particles extracted from cast composites AM4P2 is indexed on the basis of JCPDS-Data Cards as shown in Table VII. It is evident that except for MgMn2O4 which has shown a feeble independent peak at 2θ = 26.99 deg, all the other peaks are overlapping and could originate from γ-Al2O3, MnAl2O4, MgAl2O4, and MgMn2O4, which could form during processing. Therefore, XRD alone does not identify the extracted particles unambiguously. It is likely that some of the spinels compounds may also have formed but their amount will be relatively much less than alumina due to presence of magnesium and manganese mostly as alloying elements in the alloy.

Table VII Position of Peaks in XRD Pattern of Particles Extracted from AM4P2
Full size table

The particles extracted from cast composite AM4P2 have been transferred to an adhesive tape and the morphology has been examined under FESEM at lower and at higher magnification as shown in Figure 7. It is observed that there are particles belonging to different size classes and nanoparticles are adhering on larger particles, as shown in Figure 7(b).

Fig. 7
figure 7

SEM micrographs showing the morphology of extracted particle from AM4P2, at (a) lower and (b) higher magnifications

Full size image

The elemental composition of the extracted particle has been determined by using EDX attached with FESEM and the results are shown in Figure 8. The composition of the particles marked in micrograph is indicated in the boxes marked (a), (b), (c), and (d). The particle whose analysis is given in (a) is alumina primarily along with small amounts of magnesium and manganese. Magnesium indicated could be a surface layer of MgAl2O4 or MgO on alumina as indicated by XRD of extracted particles in Table VII and manganese indicated could be from small remaining MnO2 in some particles. The signals for these elements could as well be due to exposure of the matrix below, which has these alloying elements in it. In (b), (c), and (d), manganese increases progressively. There could be different proportions of γ-Al2O3, MnAl2O4, MgAl2O4, and MgMn2O4 in these particles. There is some iron which may have been picked up from the stirrer when coating falls off during processing, as mentioned earlier.

Fig. 8
figure 8

EDX of the extracted particles from the cast composite AM4P2

Full size image

EPMA analysis of the extracted powders has been carried out in terms of oxide constituents of MgO, Al2O3, MnO, and FeO and the results of some particles other than pure alumina from different locations are shown in Table VIII. In location 1/1 and 5/1, the particles are MnO2 partially reacted to alumina while at 2/1 in the table, the particle is primarily MgO.

Table VIII EPMA Result Shows the Presence of Different Oxide Particle in Weight Percent
Full size table

In 3/1 it is MgO mixed with some alumina. In location marked 2, the particles appear to be primarily MgAl2O4. In location 3, the particles are MgMn2O4. These findings are in line with the observations of XRD in Table VII. However, the relative amounts of these oxides involving magnesium, manganese, or iron are smaller compared to alumina.

Some of the extracted fine particles have been transferred to acetone by sonication and some of these particles were transferred to copper grid and dried for examination under TEM for examining the particle sizes. The extracted particles have been similarly classified based on size classes used earlier, as given in Table IX, showing that the particles extracted belong broadly to the same size classes as those added. However, the statistics of these particles are indicative only as the transfers involved could have their biases.

Table IX The Average Size and Number of Extracted Nano-Sized Particles from Cast Composite in Different Size Classes
Full size table

A typical micrograph of the particles extracted from the composite AM4P2 and the SAD pattern are given in Figures 9(a) and (b), respectively. The SAD pattern shows both array of spots marked “abcd” and “efgh,” apart from spots arranged on rings marked by R with subscripts 1 to 8. The indexing of the array of spots is shown, respectively, in Figures 9(c) and (d) identified to have originated from MgAl2O4 and MnAl2O4 respectively. The spots arranged on rings have also been indexed as shown in Table X and these spots are identified to have originated from γ-Al2O3.

Fig. 9
figure 9

(a) TEM micrograph of extracted particle from composite AM4P2. (b) Selected area electron diffraction patterns showing the compounds of MgAl2O4 and MnAl2O4 as indexed in (c) and (d), respectively

Full size image
Table X The Compounds as Identified from the Planes Giving Rise to Spots Arranged on a Ring in the SAD Pattern of Extracted Particles from the ZAS AM4P2
Full size table

3.2 Mechanical Properties of Cast Composites

The Brinell hardness has been measured along the height of the composite ingots, containing different amounts of milled powder mix. There are inhomogeneities in particle and porosity distribution along the height of ingot and it is reflected by variation of hardness along the height. Figure 10 shows the variation of averaged hardness with height, h, from the bottom of cast ingot, normalized as h/H ratio, where H is the total height of the ingot. The results for the composites have been compared with those for base alloy without addition of milled powder mix. It is observed that in an ingot, the hardness is lower toward the top due to higher porosity as observed commonly in cast composites.[1,3] There is high mean hardness up to a significant height from the bottom (h/H ~ 0.6 or 0.65), which is relatively sound for making tensile specimens. The mean hardness in this segment shows an interesting pattern—within the higher and lower limits for 2 (AM4P2) and 0 wt pct addition of milled powder, the mean hardness decreases in the following order in composites—1 wt pct addition (AM4P1), 3 wt pct addition (AM4P3), and 4 wt pct addition (AM4P4) as shown in Figure 10. This result is surprising since the composite synthesized with higher addition powder mix has more extracted particles but higher particle content also implies higher porosity in the composite as revealed by Figure 6(b). For composites fabricated by addition of up to 2 wt pct powder mix, the particle content appears to dominate hardness but beyond this addition, porosity starts to assert itself by lowering hardness.

Fig. 10
figure 10

Variation of hardness along the normalized height h/H of the ingot of cast composites where h is the height from the bottom and H is the total height of the ingot, for composites developed by addition of 0, 1, 2, 3, and 4 wt pct milled powder mix

Full size image

The stress–strain behaviors displayed by the best of the specimens of different cast composites tested are shown in Figure 11. The composites show ductile deformation behavior with a maximum in engineering stress–strain diagram indicating the onset of necking. There is significant improvement in strain hardening in composites fabricated by addition of milled powder mix, leading to significant increase in ductility as well as strength over those observed in the base alloy without any addition of powder mix. But in composite prepared by addition of 4 wt pct powder mix, there is decrease in both strength and ductility compared to those in other composites prepared by addition of 1, 2, and 3 wt pct powder mix, which have fairly close stress–strain diagrams.

Fig. 11
figure 11

Tensile stress–strain behavior of cast composites developed by addition of 0, 1, 2, 3, and 4 wt pct milled powder mix

Full size image

The variation of yield strength, tensile strength, and percent elongation in composites with increasing amount of fine particles as estimated from manganese content of the matrix alloy in Table VI, is shown in Figure 12(a). Both the yield strength and tensile strength increase with increasing amount of fine particles till 3 wt pct addition of milled powder and the ductility is maintained at a high level although it decreases slightly with increasing amount of fine particles. But, there is decrease in all these three tensile properties when the composite is prepared with 4 wt pct addition of powder mix. Similar trends of these results are observed when plotted with increasing amount of extracted particles as shown in Figure 12(b). This is expected since the amount extracted particle increases with increasing addition of powder mix, which increases the amount of nanoparticles as given in Table VI.

Fig. 12
figure 12

Variation of yield strength, tensile strength, and percent elongation of cast composites with (a) increasing weight percent of fine nano-sized alumina and (b) increasing amount of particles extracted

Full size image

The SEM micrographs of the fracture surfaces of the composite shown in Figure 13(a) reveals dimpled ductile fracture which is consistent with high ductility observed. At higher magnification, many smaller nanoparticles could be observed in individual dimples, as shown in Figure 13(b), possibly indicating that these nanoparticles have not initiated void nucleation by debonding, which led finally to ductile fracture.

Fig. 13
figure 13

SEM micrograph (a) of the fractured surfaces of tensile specimen of AM4P1 at relatively lower magnifications and (b) highlighting position of small particles in reference to dimples

Full size image

4 Discussion

The milling procedure[17] adopted to generate nanoparticles uses substantial amount of aluminum powders to disperse nanoparticles of either manganese dioxide or alumina generated so as to keep the nanoparticles physically separate minimizing their encounter and possible agglomeration during solidification synthesis of composite. The success of this strategy could be clearly seen in composites prepared with powder mix up to 3 wt pct as indicated by the mechanical properties and revealed by TEM micrograph given in Figure 4(b).

The larger size particles formed by internal oxidation during solidification synthesis, is mostly alumina and the size of these particles could be clearly seen in the FESEM microstructure in Figure 4(a). The particles extracted from the composites by leaching out the matrix show the presence of both nanoparticles and larger particles of nearly micron (submicron) size as observed in Figure 7. The number distribution of milled particles in different size classes after milling as given in Table III clearly shows that most of the particles in the milled powder mix have been reduced to nanoparticles of size below 200 nm as revealed in Table III. The reduction of MnO2 particles on addition of powder mix to molten alloy is confirmed by the increasing manganese content in the matrix alloys with increasing addition of powder mix in different composites, as given in Table V and Figure 5. However, the amount of MnO2 nanoparticles retained within the melt is lower than that added through powder mix as estimated on the basis of the amount of manganese released to the matrix alloy of the composite and given in Table VI. It is presumed that MnO2 reacts with aluminum primarily resulting in alumina and magnesium mostly remains as alloying element in the melt without participating significantly in reduction reaction. Although the results of XRD and EPMA in Tables VII and VIII, respectively, and EDX in Figure 7 show presence of some compounds involving magnesium and manganese oxides but Table V reveals their presence mostly in the matrix alloy. The amount of total reinforcement including both nanoparticles and larger particles is indicated in Figure 6(a) and the relative amounts of nanoparticles and larger particles as estimated are given in Table VI. Since increasing addition of milled powder mix during solidification synthesis of composite increases stirring time, there is increased formation of porosity (Figure 6(b)).

The oxide-based reinforcing particles embedded in alloy matrix have significantly higher elastic modulus compared to surrounding ductile matrix, which will undergo plastic deformation at a much lower stress compared to the breaking stress of brittle oxides. At a given tensile stress below breaking stress, the particle will undergo smaller elastic extension but at the same stress, the matrix undergoes relatively much larger elastic/plastic extension due to lower elastic modulus/higher ductility. When the particle is embedded in the matrix, it will therefore restrain the immediate surrounding matrix from undergoing as much extension and the nature will devise a strategy to accommodate this mismatch in extension across the interface.

One may characterize the extent of mismatch in extension when stress in the composite increases by a small amount, from σ to σ + dσ, a free particle of original length l p, undergoes an increase in extension of dl p and immediately surrounding matrix of similar original length l m (~l p) undergoes an increase in extension of dl m. The mismatch in increased extension, δl, developing due to increase in stress dσ is given by,

$$ \delta l = dl_{\text{m}} - dl_{\text{p}} = \frac{{l_{\text{m}} d\sigma }}{m} - \frac{{l_{\text{p}} d\sigma }}{{E_{\text{p}} }}, $$
(1)

where E p is the elastic modulus of the oxide based particle and m is the elastic modulus of the matrix when the matrix deformation is in the elastic range or m is the slope of the stress strain diagram at stress σ when matrix deformation is plastic. For l m = l p, Eq. [1] becomes

$$ \delta l = l_{\text{p}} \left( {\frac{1}{m} - \frac{1}{{E_{\text{p}} }}} \right)d\sigma . $$
(2)

One may get the mismatch in extension at a given stress σ by integrating Eq. [2] from 0 to σ. If stress exceeds yield stress of the matrix, the slope, m, of stress–strain diagram at stress σ decreases in the plastic region from the elastic modulus at the yield point to the value of true stress corresponding to σ at the onset of necking. So, during tensile test, the mismatch is going to be more severe in the plastic region for increasing stress as well as for larger size of the particle, l p. Now, for the successful development of particle reinforced composite, there should be accommodation of this mismatch in extension with increasing stress ensuring preservation of particle matrix interface, which fails by debonding eventually initiating ductile fracture and limiting the mechanical properties that can be achieved.

During tensile test, the matrix undergoing elastic deformation initially, the mismatch in extension between the particle and the matrix in which it is embedded, is very small. It could be accommodated by lattice bending across the interface giving rise to shear stress, which helps in load transfer from the matrix to the embedded particle.[18] The load transfer to the particle depends also on its geometry and the stress in the particle could be lower than that of the matrix, thereby aggravating the mismatch compared to that estimated by Eq. [2]. For poorly bonded interface, the shear stress developing even with a small mismatch could exceed the shear strength of the interface leading to interface failure and consequently, lower yield strength, tensile strength, and ductility of the composite compared to those of the matrix. While increasing the load on the composite, the shear stress increases beyond the shear strength of the matrix and the mismatch can no longer be accommodated by lattice bending. There will be generation of geometrically necessary dislocation on the matrix side and the number of these dislocations increases with increasing mismatch of extension with increasing stress. Using a simple expression of dislocation theory,[18] the number of dislocation, dN m, with burgers vector b required to be generated in length l p, to accommodate the extension mismatch, δl, is given by the expression,

$$ \delta l = \frac{{dN_{\text{m}} b\overline{x} }}{{l_{\text{p}} }}, $$
(3)

where the average position of dislocation within l p may be taken as \( \overline{x} = l_{\text{p}} /2 \). Thus, using the approximate estimate of δl given by Eq. [2], the number of such geometrically necessary dislocation, dN m, is given by,

$$ dN_{\text{m}} = \frac{{2l_{\text{p}} }}{b}\left( {\frac{1}{m} - \frac{1}{{E_{\text{p}} }}} \right)d\sigma . $$
(4)

For a given σ, the number of geometrically necessary dislocation, N m, may be calculated by integrating Eq. [4]. The energy of these N m dislocations may be roughly estimated as

$$ E_{\text{d}} = \alpha G(N_{\text{m}} b)^{2} , $$
(5)

when α is a nearly constant term involving also the slowly varying logarithmic term in dislocation energy. N m will increase with increasing particle size, l p and for a given particle size in a composite, the stress will increase with loading under tensile test and reach a point when E d just exceeds the interface energy, γ to cause failure of the interface transferring the entire load to the matrix.

One may improve the estimates given by Eq. [1] to [5] but these are adequate to provide a scheme for understanding the results on the tensile properties of the composites containing both fine and coarse particles in comparison to those in composites containing coarse particles alone. For the same total volume of particles, if the size of each particle is smaller the amount of mismatch of extension to be accommodated at each particle becomes smaller and there is more likelihood that E d < γ up to a large stress. Thus, lower particle size will delay the interface breakdown to higher stress and will generate smaller number of geometrically necessary dislocations at each particle to result in more uniform dislocation density if the particles are well distributed in the matrix. Kang and Chan[19] observed that dislocation density is higher near the nanoparticles in a nanocomposite, which could be indicative of the dislocations necessary to accommodate mismatch in extension. These geometrically necessary dislocations are expected to enhance strain hardening, thereby contributing to increased uniform elongation. Thus, ductility should increase as observed by a number of investigators in nanocomposites.[16]

In the present investigation, it is clearly observed that the presence of coarser particles alone in the base alloy containing 1.89 pct alumina formed by internal oxidation has resulted in limited ductility of around 10 pct as shown in Figure 11. Although the composites fabricated by addition of milled powder mix have sometimes more coarser particles as in AM4P1 but it could be counteracted by the presence of nanoparticles, resulting in increased ductility to above 22 pct. The strain hardening coefficient of these composites with high ductility is also higher. If one considers matrix along with nanoparticles, the geometrically necessary dislocations will harden the matrix increasing the effective value of m and so, mismatch in extension of the larger particles given by Eq. [2] decreases delaying the debonding of larger particles to higher extension. Thus, the ductility is higher in presence of nanoparticles. Mula et al.[20] have observed that in cast Al-2 wt pct Al2O3 (average size ~10 nm) composite prepared by ultrasonic stirring of melt-particle slurry, yield strength increases by ~57 pct and UTS by ~48 pct, compared to those in cast commercial aluminum processed similarly by ultrasonic stirring but the ductility in composite decreases. In the present study, for addition of 2 wt pct addition of powder mix (corresponding to about 1.3 pct nano-sized alumina) apart from 1.68 pct of nearly micron sized alumina formed by internal oxidation, there is improvement of yield strength by about 25 pct and tensile strength by about 35 pct over those in the composite without nanoparticles and the ductility increases by more than 120 pct. The decrease in ductility in the study by Mula et al.[20] could possibly indicate widespread clustering which affects all the tensile properties adversely as observed for higher addition of powder mix in the present study, as shown in Figures 11 and 12. Mazahery et al.[21] have reported that there is significant improvement in mechanical properties in stir cast A356 aluminum alloy when reinforced with 1.5 vol pct nano-Al2O3 (average size ~50 nm), showing tensile strength of 182 ± 2 MPa and percent elongation of 2.2 pct, but in the present investigation it has been possible to achieve a much higher strength using even lower amounts of nanoparticles in combination with larger particles as in AM4P1, while retaining much higher ductility. The composite synthesized here shows a potentially cheaper route to achieve better properties by reinforcing with larger particles along with nanoparticles in order to achieve higher amount of reinforcements while retaining reasonable amount of ductility.

5 Conclusions

  1. 1.

    Nano-sized γ-alumina particles have been generated possibly by high energy impact at points where aluminum particles collided with manganese dioxide particles at high energy triggering chemical reaction.

  2. 2.

    The microstructure of the cast composites synthesized by addition of milled powders shows two distinct size classes of alumina particles—the larger particles of nearly micron size formed by internal oxidation of the melt during processing and nanoparticles generated during high energy milling or by reaction between aluminum melt and milled nanoparticles of manganese dioxide.

  3. 3.

    The larger particles are fairly well distributed in the microstructure and the nanoparticles are mostly distributed individually although there are some clusters of two or three particles observed sometimes under TEM.

  4. 4.

    The engineering stress–strain behavior shows that a composite containing 1.89 pct of coarser (nearly micron size) particles of alumina has limited ductility of around 10 pct while the presence of nanoparticles could achieve a higher ductility even when composites contain higher amounts of larger particles, presumably due to well distributed geometrically necessary dislocations at the interface of nanoparticles contributing to significant strain hardening of the matrix to cause delay in debonding of larger particles to higher stress.

  5. 5.

    The yield strength and tensile strength also increases with increasing addition of powder mix, which also results in higher particle content in the composites.

  6. 6.

    For higher addition of powder mix (beyond 3 pct) in the composite, ductility as well as both yield and tensile strengths are adversely affected and it has been attributed to clustering of nanoparticles.

  7. 7.

    The composites investigated fail by ductile fracture under tensile loading and debonding of larger particles results in dimples which often contains several nanoparticles inside.