Mechanical and functional properties of ultraﬁne grained Al wires reinforced by nano-Al 2 O 3 particles

Abstract A powder metallurgy route based on high-energy ball milling, powder consolidation by hot extrusion and cold rolling was used to produce Al composite wires reinforced with Al 2 O 3 nanoparticles. The process was capable of preparing fully dense nanocomposites characterized by well dispersed nanoparticles in a ultra-fine grained matrix. Ball milling led to the fragmentation of the passivation oxide layer that covers the aluminum particles and of the alumina particle clusters added ex-situ in addition to embedding these nano-sized particles in the Al matrix and thus producing optimal precursors for subsequent consolidation. The nanocomposites showed improved mechanical performances in term of hardness and tensile strength. They also exhibited excellent damping behavior at high temperatures.


Introduction
Metal matrix nanocomposites (MMnCs) are considered interesting materials since they show higher strength than the corresponding base metals while retaining a good toughness [1][2][3].They are generally made up of a ductile metal matrix reinforced with hard nanoparticles (NPs).Different from precipitates formed in precipitation hardening alloys, the reinforcing NPs are thermodynamically stable, making MMnCs ideal for high temperature applications [1][2][3].Such small particles can obstruct the motion of dislocations and are responsible for the formation of geometrically necessary dislocations due to the mismatch in coefficients of thermal expansion and in elastic moduli between the metal matrix and the NPs [4].Low wettability and a high surface to volume ratio of ceramic nano-particles are the main issue to face to prepare MMnCs.NPs tend to agglomerate and form clusters, losing their capability to effectively obstruct the movement of dislocations.For this reason, they cannot be prepared by conventional casting methods.To overcome this problem several non-conventional manufacturing methods have been proposed, and they can be categorized into two major groups: ex-situ and in-situ synthesis routes [1][2][3].The former refers to those processes in which the nano-reinforcement is added to the liquid or powder metal whereas the latter refers to those methods that lead to the formation of nano-sized compounds during the process itself, e.g. through reacting gases.Powder metallurgy routes [5][6][7][8][9][10], ultrasound assisted casting [11,12], disintegrated melt deposition [13] are some of the processes commonly used to produce MMnCs.
The high strength of MMnCs can be further improved by decreasing the grain size of the matrix down to the sub-micrometer level (Hall-Petch strengthening) [14].Indeed, ultrafine grained (UFG) materials, processed by severe plastic deformation methods like equal channel angular pressing or high pressure torsion, have attracted growing interest because of their unique physical and mechanical properties [15].The combination of properties conferred to the aluminum matrix by the combination of an UFG microstructure and hard NPs would be particularly attractive for all those applications requiring low density and high mechanical properties.
In this investigation, a powder metallurgy route to produce Albased MMnCs reinforced by Al 2 O 3 NPs is adopted.An UFG microstructure was conferred to the micro Al particles through highenergy ball milling (BM) and preserved during consolidation thanks to the oxide dispersion.Ball milling has proven to be a suitable technique for breaking the surface oxide layer that covers the aluminum particles into nano-sized fragments (in-situ production of NPs).It also revealed able to embed NPs into the ductile Al matrix.The ball milled powder, after canning, were consolidated via hot extrusion.The extruded rods were cold rolled down into wires to verify the formability.The wires were then characterized for mechanical properties (tensile and Vickers hardness) and microstructures including NPs dispersion and grain sizes by electron microscopy analysis.
The damping capacity, or internal friction (IF), was also investigated.IF is a measure of the energy dissipated by a material during imposed mechanical vibration under cyclic loading.IF of the Al/ Al 2 O 3 MMnC wires was studied to consider these materials as possible candidates for applications in which the combination of good mechanical properties and high internal friction required.Indeed, vibrations generated in response to a dynamic loading are responsible for high noise levels, premature fatigue failure and wear in most of the frequently used structural materials such as steels and Al alloys which exhibit relatively low IF [16,17].

Method
A commercial purity Al powder with an average size of 20 lm (supplied by ECKA Granules) and a colloidal solution of alumina particles with an average particle size of 50 nm in isopropyl alcohol (supplied by Sigma Aldrich) were used as starting materials.The aluminum particles were passivated by exposure to air to possess a thin oxide layer.According to previous works [18][19][20][21][22][23], the thickness of this layer is considered to be in the range of 2-4 nm.The nominal oxygen content in the Al powder was <0.5 wt.% (data supplied by ECKA Granules).A scheme of the starting materials used for MMnCs manufacturing is depicted in Fig. 1.
The Al powder was added to the colloidal solution (with 2 wt.% of ex-situ Al 2 O 3 NPs in the composite) and the mixture was stirred in a beaker and then dried at 50 °C.After this operation, the Al particles were covered by the NPs in clusters, as illustrated in Fig. 2.
High energy ball milling was performed on the Al powder or the Al-2 wt.% Al 2 O 3 powder by using a Vario-Planetary Mill Pulverisette 4 equipped with steel vials and balls (10 mm in diameter).1.5 vol.% of ethanol was used as process control agent to avoid excessive cold welding and agglomeration of the particles.To minimize oxidation during attrition, the two vials were back-filled with argon.The milling was performed for 16 h with a ball-topowder weight ratio of 10:1.An excessive temperature rise was avoided by interrupting the procedure for 10 min after each 30 min of milling.The speed of the main disk was set at 250 rpm clockwise whereas the speed of the two planets at 200 rpm counter-clockwise.After milling, the powders were packed in cylindrical Cu alloy cans (external diameter = 10 mm and thickness = 1 mm).They were closed by means of press-fit plugs and then subjected to hot extrusion at 400 °C, as described in [24].The extrusion speed was 5 mm/min.The die was heated by an induction coil and the temperature was monitored by a type-K thermocouple.The starting billet diameter was reduced to 4 mm after extrusion.After hot deformation, the can material was stripped off and the rods were cold rolled down to a square section of 1 mm 2 by a caliber rolling mill, with intermediate annealing at 400 °C for 5 min after each area reduction of about 20%.
The density of the materials was estimated based on the Archimedes' principle using polished samples.Five measurement were performed for each type of sample.
Microstructural analysis of the transversal section of the rolled wires was carried out by scanning electron microscopy (SEM) -Zeiss Supra 40 equipped with a high efficiency in-lens SE detector.To obtain better contrast between the NPs and matrix, samples were etched with the Keller's solution.TEM foils with a final thickness of $100 nm were prepared using a Nova Nanolab 200 focused ion beam (FIB).High angle annular dark field (HAADF) imaging by scanning transmission electron microscopy (STEM) was performed using a FEI Tecnai F20 operating at 200 kV.X-ray diffraction (XRD) was performed using PANalitical X-Pert PRO equipped with a RTMS X'Celerator sensor.Cu Ka (k = 0.15418 nm) radiation was used.The crystallite size and micro-strain were calculated using the Williamson-Hall analysis [25].
Vickers microhardness (HV) was measured using Future Tech Corp. FM-700, applying a 2 N load for 15 s.The gauge length of the samples was 30 mm and the cross-section was 1 mm 2 .Tensile tests were performed with a crosshead speed of 0.5 mm/min (de/ dt = 2.7 Â 10 À4 s À1 ).Since the wire samples were too short to adopt special clamps for wires, premature fractures occasionally occurred close to the clamping position at strain levels exceeding about 4.5%.For this reason, the stress-strain curves presented have been are terminated at a strain of 4.5%.
Mechanical spectroscopy [26] is a technique that consists of applying a sinusoidal stress to a material and measuring the strain response.IF is related to the time-dependent elasticity of a material.Metals and alloys respond to an applied load not only by time-independent elastic strain, but also by time-dependent strain that lags behind the applied load.Because of the lag induced by the relaxation, the stress r and strain e can be expressed as:  e ¼ e 0 expðixt À dÞ ð 2Þ where r 0 and e 0 are the stress and strain amplitudes, respectively, x is the angular vibration frequency and d is the loss angle by which the strain lags behind the stress.By combining these two equations, the resultant complex modulus, E is defined as: where E 0 is the storage modulus and E 00 the loss modulus.The storage modulus represents the stiffness whereas the loss modulus is a measure of the oscillation energy transformed into heat.The ratio of the loss modulus to the storage modulus is usually defined as tan d and commonly used to indicate the damping capacity of a material: Accordingly, the IF results in the present investigation are expressed in terms of tan d as a function of temperature.The tests were carried out using a DMA Q800 TA Instrument equipped with a liquid nitrogen cooling system.The samples were tested in the single cantilever configuration at 0.1, 1 and 10 Hz in the temperature range of À130 to 400 °C with a heating rate of 2 °C/min.The width (1 mm) and the thickness (1 mm) of the wire were measured by micrometer (±0.01 mm).The length of the specimens was 17.5 mm.

Microstructures
Morphological analysis of powders was carried out before and after milling which was responsible for severe alteration of the shape of metal particles.The as-received Al particles were round, smooth and separated but, after ball milling, they appeared agglomerated and became flat with sharp corners (Fig. 3).The Al-2 wt.% Al 2 O 3 powder had a similar appearance after milling.In Fig. 4a, a close-up of the milled particles.It is reasonable to assume that high-energy ball milling was able to break the alumina particle clusters and fragmentize the surface oxide into small pieces, forming in-situ alumina particles (Fig. 4b).Thus, highenergy ball milling may lead to two types of nanocomposites: (a) the Al powder milled alone to produce MMnCs reinforced with oxide particles formed by fragmentation of the surface oxide layers, i.e. produced in-situ; and (b) the Al-2 wt.% Al 2 O 3 powder milled to produce MMnCs reinforced by a combination of alumina particles produced insitu and alumina particles added ex-situ.
The milling process led to severe plastic deformation of the particles.The effect was evaluated by XRD technique.Peak broadening and a reduction in peak intensity were observed after milling for both the Al and Al-2 wt.% Al 2 O 3 composite powders.Peak broadening is due to micro-strain caused by lattice defects and to reduction in the crystallite size.The crystallite size and micro-strain was estimated by Williamson-Hall method and the results are summarized in Table 1.The average crystallite size of the as-received Al powder was estimated to be 662 nm and the micro-strain 0.03%.The effect of ball milling (BM) was similar on the Al and composite powders.It led to a significant reduction in the crystallite size to about 120 and 114 nm, respectively, and to an increase in the micro-strain to 0.17%.
After ball milling, the powders were successfully consolidated through hot extrusion and then cold rolled without failures down to square wires (1 mm Â 1 mm).The square section was constant along the length of the wires, which was about 30 cm.The whole process aimed at obtaining fully dense composite materials with in-situ and ex-situ Al 2 O 3 particles embedded into Al particles during milling and well dispersed in the metal matrix after extrusion.The average density of the wires was (2.708 ± 0.008) g/cm 3 and (2.716 ± 0.008) g/cm 3 , respectively.
The particle dispersion in the MMnCs wires was evaluated by SEM and TEM.In Fig. 5, the microstructure of the Al sample after etching is presented.As expected, the BM process led to the breaking into small fragments of the surface oxide layers.It is worth noting that these particles formed in-situ, revealed to be nano-sized and homogeneously distributed in the Al matrix.The oxide particles were generally smaller than 50 nm.The fragments aligned to the rolling direction (highlighted with red1 circles in Fig. 5) seem to confirm the thickness of the passivation layer measured in previous research work (<4 nm) [18][19][20][21][22][23].The particle size and distribution were also investigated by TEM-EDX chemical mapping (Fig. 6).The particles are outlined by the oxygen-rich areas and their sizes were consistent with those from SEM. Owing to their small sizes, these NPs are expected to act as obstacles for dislocation slip (Orowan strengthening) when they are not placed at grain boundaries [27].Fig. 7 represents the microstructure of the composite wire reinforced with in-situ Al 2 O 3 as well as 2 wt.% ex-situ Al 2 O 3 NPs.The fraction of oxide is of course higher and some small clusters are also present.It is worth noting that the micrographs shown in Figs. 5 and 7 were from etched specimens.Therefore, the actual amount of the reinforcement oxide should be lower than it seems.The chemical etching, indeed, removed part of the Al matrix from the surface of the samples, leaving a higher fraction of nanoparticles.For the sake of clearness, the schematic depicted Fig. 8 describes in a simplified form the whole consolidation process.In Fig. 9, the HAADF images highlight the grain structures in the two types of wires, both showing ultrafine grains.The average grain size of the Al wire reinforced with 2 wt.% of ex-situ Al 2 O 3 NPs was slightly smaller than that of the Al wire reinforced only with the in-situ NPs (180 nm and 220 nm respectively, as measured by the linear intercept method).
Although the samples were extruded and annealed at a high temperature (400 °C), it can be assumed that grain growth was restricted by NPs which possibly acted as pinning points on grain boundaries.The crystallite size of the wires calculated showed that the growth of the coherently scattering domains was limited (Table 1).For the Al wire reinforced with the in-situ particles the grain size was estimated to be 228 nm (120 nm before extrusion) and for the wire reinforced also with the ex-situ particles, the grain size was 206 nm (114 nm before extrusion), in fairly good agreement with the results from TEM.The results summarized in Table 1 also show that the micro-strain decreased only slightly, even though the consolidation and annealing were performed at 400 °C.It means that a high amount of lattice defects is retained in the matrix.

Mechanical properties
Vickers hardness tests and tensile tests were carried out in order to estimate the mechanical properties of the wires.The Al wire reinforced with the ex-situ particles showed higher hardness than the wire only reinforced with the in-situ particles (106 ± 1 HV vs. 96 ± 1 HV).These values are considerably higher than those typically found in commercial purity (CP) Al ($20 HV) and in UFG CP Al ($40 HV) [28].The tensile results are summarized in Table 2 and typical tensile stress-strain curves are shown in Fig. 10.The wire reinforced ex-situ also showed both higher yield strength (YS) and ultimate tensile strength (UTS).Therefore, MMnCs showed YS and UTS values much higher than those generally related to pure Al [29].In particular, the material reinforced with 2 wt.% of ex-situ Al The improvement is possibly due to the additive strengthening effects of dispersed nanoparticles, ultrafine grain sizes and coefficient of thermal expansion (CTE) mismatch that generates increased dislocation density [5,14,28,30].Moreover, as for conventional metal matrix micro-composites, the load transfer from the soft matrix to the hard particles under an applied load contributes to the strengthening of the base material.The volume fraction and aspect ratio of nanoparticles in MMnCs are generally quite low,  and therefore the contribution from load transfer is considered relatively small [5].The grain size has a strong influence on strength since the grain boundaries can obstruct dislocation movement.This is due to the different orientations of adjacent grains and to the high lattice disorder characteristic of the boundary regions, which prevent the dislocations from moving in a continuous slip plane [14].The particles play an important role in the final grain size of the composites since they can hinder grain growth, acting as pinning points for grain boundary migration.According to the Zener equation [5], an increase in the volume fraction and a decrease in particle size lead to a finer grain structure.The Orowan mechanism relies on the interaction of nano-particles with dislocations.The nano-particles act as pinning points hampering the movement of the linear lattice defects and leading to dislocations bowing around the particles [27].Also, the mismatch in CTE between the reinforcement and the metal matrix is believed to contribute to the final strength of nanocomposites.The aforementioned mismatch leads to the formation of dislocations, which are geometrically necessary to accommodate the different shrinking behaviors of the metal and the particles [30].

Mechanical spectroscopy
In Fig. 11, the results of damping tests are depicted in terms of tan d as a function of temperature.At 0.1, 1 or 10 Hz, a peak tan d is observed for both types of MMnCs.Higher amount of Al 2 O 3 NPs led to better damping performance (i.e. higher tan d) at temperatures higher than 50 °C for 0.1 Hz, 70 °C for 1 Hz and 100 °C for 10 Hz.The tan d (1 Hz) value for the monolithic CP aluminum is $0.001 at 25 °C and $0.007 at 275 °C [31].Thus, a significant increase in internal friction was achieved in both MMnCs investigated.The damping behavior is strongly sensitive to lattice defects (point defects, dislocations, grain boundaries and interfaces) [26].From Fig. 11, the shift of peaks towards higher temperatures can be clearly identified for both materials when the loading frequency  is increased.This phenomenon is generally related to relaxation processes (relaxation type peak) [16,26,32].The higher density of defects (higher micro-strain) and higher amount of interfaces in the wire reinforced with ex-situ Al 2 O 3 are thought to cause the increased IF at high temperatures.
In fact, although the external applied stress is small, the internal stress concentration in regions of high dislocation density may be large enough to cause relative atomic sliding.At room temperature, the displacements are typically fractions of an atomic diameter whereas, at high temperatures, sliding can be much more  $20 $30 Pure UFG bulk Al [21] $110 $120 extensive and lead to viscoelastic strain.For this reason, the damping effects are more pronounced at elevated temperatures.However, with continued increase in temperature, dislocation mobility keeps increasing, resulting in annihilation of dislocations and eventually a decrease in tan d [16,26,32].The results show that the storage modulus of the MMnC reinforced with ex-situ NPs is higher than that of the MMnC reinforced only with in-situ NPs in the whole temperature range and for all the frequencies investigated (Fig. 11d), although as temperature increases the difference in the storage modulus decreases.This behavior can be related to the relaxation phenomena as well.At room temperature the storage modulus values reached by the in-situ and the ex-situ MMnCs are about 64 GPa and 68 GPa, respectively.The higher modulus of the composites is believed to be due to the combination of the elastic moduli of alumina and aluminum.

Conclusions
Al based nanocomposite wires were successfully produced via a powder metallurgy route based on high-energy ball milling, powder consolidation by hot extrusion and cold rolling.Ball milling led to fragmentation of the surface oxide layers on Al particles and the breakup of alumina clusters added ex-situ into nano-sized alumina particles.It was also able to homogeneously embed these nanoparticles in the Al matrix, producing the optimal precursors for subsequent consolidation.The milled powders were successfully consolidated by hot extrusion and then formed into wires by cold rolling without fracture.The process led to fully dense MMnCs characterized by well dispersed NPs and ultra-fine grains.The MMnCs so prepared showed excellent mechanical properties, especially compared to those of CP Al produced by conventional processes.The Al wire reinforced with ex-situ particles showed higher hardness, YS and UTS than the Al only reinforced with insitu particles.Internal friction in the nanocomposites was determined at 0.1, 1 and 10 Hz, with well defined tan d peaks.Higher amount of Al 2 O 3 NPs leads to higher damping.The relaxation peak shifts towards higher temperatures with increasing loading frequency.The storage modulus of the MMnC reinforced with ex-situ NPs is higher than that of the MMnC reinforced with only in-situ NPs under the testing conditions.

r ¼ r 0 expðixtÞ ð 1Þ Fig. 1 .
Fig.1.Schematics of a passivated aluminum particle and the colloidal solution of alumina particles in isopropyl alcohol.

Fig. 2 .
Fig. 2. Schematic of Al particles covered by alumina NPs after mixing and drying.

Fig. 3 .
Fig. 3. (a) As-received Al powder and (b) the Al powder after 16 h of high-energy ball milling.

Fig. 4 .
Fig. 4. (a) Clusters of the milled Al particles and (b) schematic of what might have happened during milling.
2 O 3 reached YS and UTS of 282 MPa and 373 MPa while the composite reinforced with in-situ Al 2 O 3 reached YS and UTS of 225 MPa and 302 MPa, respectively.By comparison, CP Al and UFG CP Al show much lower YS of 20 MPa and 110 MPa and UTS of 30 MPa and 120 MPa, respectively [24].

Fig. 5 .
Fig. 5. SEM micrographs of the Al sample reinforced with the in-situ Al 2 O 3 NPs at (a) low and (b) high magnifications.

Fig. 6 .
Fig. 6.(a) HAADF image of Al with in-situ Al 2 O 3 NPs and (b) oxygen map of the area selected by the frame in (a).

Fig. 7 .
Fig. 7. SEM micrographs of Al with in-situ Al 2 O 3 NPs as well as 2 wt.% ex-situ Al 2 O 3 NPs at (a) low and (b) high magnifications.

Fig. 8 .Fig. 9 .
Fig. 8. Schematics of the hot extrusion consolidation and the cold rolling processes.

Fig. 10 .
Fig. 10.Tensile stress-strain curves for the Al wire reinforced with Al 2 O 3 NPs formed in-situ and the Al wire reinforced with both Al 2 O 3 NPs formed in-situ and Al 2 O 3 NPs added ex-situ.

Table 1
Crystallite size and micro-strain of the powders and wires obtained from XRD results by the Williamson-Hall analysis.Al powder as-received + Al 2 O 3 passivation layer 662 0.03 Al powder + in-situ Al 2 O 3 particles after BM 120 0.17 Al powder + in-situ Al 2 O 3 particles + 2% wt.exsitu Al 2 O 3 particles after BM