- Email: [email protected]
Al2O3 composite by severe plastic deformation
Materials Science & Engineering A 646 (2015) 234–241
Contents lists available at ScienceDirect
Materials Science & Engineering A journal homepage: www.elsevier.com/locate/msea
Synthesis of an Al/Al2O3 composite by severe plastic deformation Lenka Kunčická a,b, Terry C. Lowe c,n, Casey F. Davis c, Radim Kocich a,b, Martin Pohludka a a b c
Regional Materials Science and Technology Centre, VŠB-TU Ostrava, 17. listopadu 15, 70833 Ostrava-Poruba, Czech Republic Department of Material Forming, Faculty of Metallurgy and Materials Engineering, VŠB-TU Ostrava, 17. listopadu 15, Ostrava-Poruba 70833, Czech Republic The George S. Ansell Department of Metallurgical & Materials Engineering, Colorado School of Mines, Golden, CO 80401, USA
art ic l e i nf o
a b s t r a c t
Article history: Received 7 June 2015 Received in revised form 20 August 2015 Accepted 22 August 2015 Available online 24 August 2015
The influence of severe plastic deformation on fabrication of an aluminum/alumina composite made from pre-sintered Al powders was evaluated. Spherical powder particles with diameters up to 45 μm were pre-compacted using cold isostatic pressing (CIP), subsequently vacuum sintered for 60 min at 500 °C, and subjected to processing using either 4 passes of swaging or 1 or 10 revolutions of high pressure torsion (HPT). For all the samples we calculated average strain and measured microhardness. Results showed an increase in microhardness with an increase of the imposed strain. However, the microhardness values and uniformity were also influenced by possible residual porosity. The smallest and most uniform grain size was achieved for the samples processed by HPT, especially after 10 revolutions (average diameter of 0.22 μm). Residual porosity was completely eliminated only after 10 HPT revolutions. Texture evaluations showed 〈111〉 fiber texture development after swaging, while grain orientations after HPT were more random. & 2015 Elsevier B.V. All rights reserved.
Keywords: Metal matrix composites (MMCs) Powder processing Electron microscopy Hardness
1. Introduction Metal matrix composites (MMC) are among the main trends in recent advanced materials research, especially with the emergence of powder-based additive manufacturing methods. These composites consist of a metal matrix reinforced with load bearing constituents that are designed to exploit the favorable properties of each component. The reinforcing elements can either be dispersed particles [1], fibers [2], or layers [3]. The most widely studied MMC system is Discontinuous Reinforced Aluminum (DRA), which contains particles or short fibers. The popularity of DRA is because aluminum is among the most inexpensive lightweight matrix metals. In addition, because of its low melting point, it is easy to combine with many different reinforcing particles or fibers. The MMC on which this research is focused is aluminum strengthened with alumina (Al2O3) particles. This combination of constituents provides higher strength, stiffness, hardness and chemical and thermal stability compared to aluminum [4]. The already favorable mechanical properties of composites can further be improved by refinement of structural units, which results in a higher volume of grain boundaries (GBs) in the deformed structure [5]. This contributes not only to strengthening via introduction of obstacles for movement of dislocations, but also to improvement of plasticity, since when the grain size is small n
Corresponding author. E-mail address: [email protected] (T.C. Lowe).
http://dx.doi.org/10.1016/j.msea.2015.08.075 0921-5093/& 2015 Elsevier B.V. All rights reserved.
enough, plastic deformation by grain boundary sliding becomes possible. Fine scale substructural features (e.g. cell, microbands) can be introduced by imposing a high amount of strain, which can advantageously be performed using methods of severe plastic deformation. The principle of these technologies lies in imposing a significant amount of shear strain into the material, which consequently results into an increase in mechanical properties, especially strength [6]. These methods include dozens of technologies, some of which are based on conventional methods, such as accumulative roll bonding (ARB) [7], multiaxial forging [8] and swaging [3], but also special severe plastic deformation (SPD) technologies, among which belong methods like high pressure torsion (HPT) [9], equal channel angular pressing (ECAP) [10], ECAP with back pressure (ECAP-BP) [11], ECAP with partial back pressure (ECAP-PBP) [12], ECAP-Conform [13], twist channel angular pressing (TCAP) [14], twist channel multi-angular pressing (TCMAP) [15], twist extrusion (TE) [16], repetitive corrugation and straightening (RCS) [17], friction stir processing (FSP) [18] and many more. However, achievement of finer final structural units is easier when the original material already contains small structural units such as found in some particles, including powders and nanopowders [19]. The particle surfaces also introduce additional interfaces, which can be engineered to impart specific properties when compacted into bulk solids. Shear strain imposed into the material during the above-mentioned processes supports shearing of the powder particles, which results in additional refinement of
L. Kunčická et al. / Materials Science & Engineering A 646 (2015) 234–241
structural units. It is also very beneficial for effective rearrangement of particles within the volume of the deformed material. Although strain imposed into the material during one pass (or revolution) is limited, all of the above mentioned processes can be performed repetitively. Therefore, the overall strain that can be accumulated in the material during multiple passes can be very large. However, most materials reach a saturated steady state condition and do not exhibit any further microstructural changes or increases in the properties with increasing strain [20]. This study focuses on preparation of an aluminum/alumina composite using severe plastic deformation of pre-sintered Al powder. Since the Al particles are already coated with a protective alumina layer, no Al2O3 particles were added. In order to evaluate the influence of strain on microstructural changes, we selected processing with severe plastic deformation by HPT and compared it with a more conventional process, during which the imposed strain is not so high (swaging). Before deformation processing, the pre-sintered Al material was subjected to evaluation of oxygen and alumina content, particle diameter volume fraction analysis and selected mechanical properties measurement. After deformation, we performed electron backscatter diffraction (EBSD) analyses and microstructural observations using scanning electron microscopy (SEM) and measurements of microhardness for the deformed samples. Analyses were also supplemented with calculations of degree of deformation.
2. Materials and methods 2.1. Material preparation We selected an Al powder with a distribution of particle sizes up to 45 μm. An SEM image of the original powder is shown in Fig. 1. We compacted the particles in our own fabricated rubber die using cold isostatic pressing (CIP) device located at Technical University of Ostrava (VŠB-TUO), pressurizing to 200 MPa for 30 s and then immediately depressurizing. The pre-compacted samples were sintered at 500 °C for 60 min in a Zwick vacuum creep testing furnace at ASCR, Brno. Although the working diameter of the dies for both the deformation technologies is 20 mm, we designed the CIP rubber die with a slightly larger diameter (25 mm), since we expected shrinkage of the loose powder during the compaction. After having performed the compression and sintering procedures, we sliced the pre-sintered material into 6 mm thick samples which we subsequently subjected to plastic deformation processing at VŠB-TUO. The deformation by HPT was performed on two
Fig. 1. SEM image of the original powder particles.
235
samples at 5 GPa and 0.5 RPM. One sample was subject to 1 revolution, and the other one to 10 revolutions. Swaging was performed on a cylindrical shaped sample in 4 swaging steps, during which the diameter was reduced from the initial 20 mm to the final 10 mm. All deformation processing was conducted at room temperature without any heat treatment and sintering processes. The deformed samples were subsequently analyzed and the changes in microstructures induced by the plastic deformation technologies were compared. Back-Scattered Electron (BSE) SEM analyses observing the initial powder particles, pre-sintered material and eventually deformed structures were performed at VŠBTUO and Colorado School of Mines (CSM), CO, USA, while electron backscattered diffraction (EBSD) microstructure and texture examinations were carried out at CSM. Prior to the analyses, the samples were ground and polished mechanically and eventually ion milled with a JEOL CP Cross Section Polisher. The locations selected for EBSD microstructural observations were in the midradius on the transverse cross-section for all the samples. The EBSD scan step for the pre-sintered sample was 0.5 μm, while for the swaged sample it was 0.3 μm, 0.1 μm for HPT after 1 revolution and 50 nm for HPT after 10 revolutions, respectively. Porosities, particle size, and grains size in all the states of the material, including the pre-sintered state, were evaluated using images obtained by BSE-SEM observations, and further processed and evaluated using ImageJ software. X-ray analyses of phase composition were performed using the Philips X'Pert system at CSM, CO, USA. Copper with Kα1 wavelength of 1.54054 Å and Kα2 wavelength of 1.54439 Å was used as the X-ray source. A Vickers microhardness measuring device at VŠB-TUO was used to measure microhardness. All the samples were tested under a load of 250 g for the time of 7 s. Hardness measurements of all the samples were performed across their diameters in order to investigate possible non-uniformities within their cross-sections. Average microhardness values from all the measured individual values were calculated as well. Young's and shear moduli of the pre-sintered sample were measured by the IMCE n.v. company in Belgium using a RFDA measuring system (http://www.imce.eu/) in accordance with the ASTM E 1876 standard in order to be able to evaluate the assumption of porosity throughout the entire volume of the pre-sintered material. All the moduli measurements were performed ten times and the average values were eventually calculated.
3. Results and discussion 3.1. Properties of pre-sintered material To be able to subsequently evaluate the influence of the deformation technologies, selected properties and parameters were observed for the pre-sintered material. First we performed SEM and X-ray analyses to observe the microstructures and evaluate chemical composition, especially to determine oxygen content. Subsequently, we measured microhardness throughout the crosssection of the pre-sintered material and calculated its average value. Eventually, elastic and shear moduli were evaluated. The results of the measurements are summarized in Table 1, while Fig. 2 shows an SEM image of the pre-sintered microstructure. The results of SEM analyses showed approximately 5 wt% of oxygen in the structure. The source of oxygen is the protective aluminum oxide layer, which initially coated the surface of all the particles. The result of X-ray analysis depicted in Fig. 3 confirmed the presence of oxygen in the form of alumina (Al2O3). The most distinctive peaks in the plot are peaks corresponding to the pure Al phase. Nevertheless, noticeable at low diffraction angles, the detail of which is depicted in the corner of Fig. 3, is diffuse
L. Kunčická et al. / Materials Science & Engineering A 646 (2015) 234–241
236
Table 1 Mechanical characteristics of pre-sintered material. Property
Unit
Value
Elastic modulus Shear modulus Poisson's ratio Microhardness Oxygen content Alumina content
[GPa] [GPa] [–] [HV] [wt%] [wt%]
53.98 19.94 0.353 29.58 5.0 10.6
homogenous isotropic behavior during loading [22]. According to the chemical composition and structure of alumina, the content of the individual elements in the chemical compound is 52.9 wt% Aluminum and 47.1 wt% Oxygen. Since the overall oxygen content in this material was determined to be 5.0 wt%, the supposed content of Al2O3 in the sample is 10.6 wt%. The material can therefore be considered as Al-matrix composite with approximately 10 wt% of Al2O3 strengthening phase. Values of elastic modulus similar to ours were observed by Gudlur et al. for Al/Al2O3 composite with residual porosity of 7% and nominal content of alumina 10% [23]. Similarity of the results also confirms the observed and calculated alumina content. 3.2. Imposed strain The first step during our evaluation procedure of the deformed samples was calculation of the imposed strain. This we performed for swaging according to Eq. (1), as well as for HPT according to Eq. (2) [3,9,24].
S0 Sn
(1)
2πrN t
(2)
φ = ln
φ= Fig. 2. SEM image of pre-sintered structure.
Fig. 3. X-ray plot for pre-sintered sample.
scattering, indicative of the semi-crystalline or non-crystalline nature of the alumina phase. Since the above described observations of distinctive peaks are valid only for crystalline structures, it is probable that Al2O3 is also present in the structure in an amorphous state. In a reference plot of X-ray data for pure Al2O3, a set of distinctive peaks starts to rise at the diffraction angle of 28° (not shown here). Although the low intensities in our plot are probably burdened with noise, a slight increase forming a small local peak can be distinguished between the angles of 28° and 29°. This is associated with the presence of Al2O3. As can be seen in Table 1, both the measured moduli are lower than the theoretical values for aluminum. Considering the elastic modulus, this value is approximately 77% of the theoretical one (70 GPa). This is caused mainly by residual porosity, since porosity has the ability to decrease the moduli [21]. The average porosity for the compressed sample before sintering was 20.76%, while after sintering, the porosity decreased to the average value of 7.74%. Fig. 2 clearly shows the presence of voids and pores in the pre-sintered material. The measured Poisson's ratio indicates
where S0 and Sn are the cross-sectional areas of the swaged sample before and after a swaging step [mm2], respectively, r is radius of the sample deformed using HPT in the measured location [mm], N is number of HPT revolutions and t is thickness of the sample [mm]. According to these calculations, the average strain imposed into the material during swaging was 0.27 after the first pass, 0.58 after the second pass, 0.94 after the third pass and eventually 1.39 after the fourth pass. For the HPT processed samples, the imposed strain was significantly different at the observed mid-radius location of both the samples. Whereas the calculated strain was 5.24 for the sample after one revolution, it was 52.36 for the sample after ten revolutions. The differences in the imposed strains are due to the different designs of the deformation technologies. Considering pressure conditions for the two processes, conditions for HPT are more favorable for particle compaction. During HPT the flow of the material is theoretically constrained by the closed dies surrounding the sample. However, the real conditions are quasi-constrained, since there is a slight gap between the two dies and the volume of a sample is slightly larger than the volume between the closed dies [25]. During HPT processing the gap gets filled with flash, which subsequently limits outward material flow. The high axial compressive force during HPT restricts the sample from sliding and rotating and also supports imposing of a high amount of shear strain into the material. On the other hand during swaging the material flow is limited in the radial direction due to the presence of the swaging dies, while the axial direction is not constrained. Therefore, material can flow out from the contact region of the swaging dies in the axial direction, which consequently reduces the amount of strain that can be accumulated in the sample [26]. 3.3. Porosity Microstructures of the samples deformed by swaging, 1 HPT revolution and 10 HPT revolutions are shown in Fig. 4a to 4d, respectively. The best consolidation of the pre-sintered powder material was found after 10 HPT revolutions (Fig. 4d). The microstructure of this sample did not exhibit any residual porosity.
L. Kunčická et al. / Materials Science & Engineering A 646 (2015) 234–241
237
Fig. 4. Microstructures of the deformed samples; swaged (a) (b), after 1 HPT revolution (c), after 10 HPT revolutions (d).
Sporadic isolated pores were still evident in the sample after 1 HPT revolution (Fig. 4c). The volume fraction of pores remaining in the structure of this sample was 0.78%. The swaged sample exhibited the largest residual porosity among the deformed samples, namely 5.67% (Fig. 4b). As can be seen in the microstructure of this sample in Fig. 4a, the pores were homogenously distributed in the swaged structure. According to these results residual porosity decreased with an increase in imposed strain. A similar conclusion was also reported e.g. by Senkov et al. within their study of an Al-based fine powder compound processed by ECAP [27]. Shear strain is especially efficient for reduction/elimination of porosity via rearrangement of powder particles within a sample, since by its influence smaller particles can fill free volumes between larger particles [28]. When comparing the HPT process to the swaging process from the point of view of the amount of the imposed shear strain, this is more significant for HPT. During swaging the axis of the sample is perpendicular to the axis of the imposed strain, while during HPT the axis of the deformation process is identical to the axis of the sample and also to the axis of the compacting force [29]. This combination enables one to impose the highest possible amount of shear strain into the deformed material. Therefore, pores in the structure of the sample deformed with 10 revolutions were eliminated since the particles experienced the most significant rearrangement during processing. 3.4. Grain size The microstructure of the undeformed pre-sintered material has already been shown in Fig. 2. The Figure clearly shows that the initial Al powder particles are all covered with a protective oxide layer. Due to its high melting temperature, alumina cannot melt or evaporate during the sintering process and remains in the
structure. Deformation processing results in straining of the aluminum matrix structure and subsequent development and movement of dislocations. At relatively low strains, brittle structural components (alumina) are largely fractured or experience no deformation, while the more ductile structural matrix (aluminum) is readily deformed [27]. The smallest alumina particles and also their cracked fragments act as obstacles for dislocations movement [30]. At higher strains, dislocations developed inside the grains/particles accumulate at these obstacles. The accumulated energy consequently results into a formation of subgrains, which eventually results into a formation of new grains and therefore grain refinement [31]. Larger alumina particles that remain in the matrix after deformation processing are able to bear load, while the smaller particles are more effective barriers to dislocation motion. Fig. 5a depicts an example of structure processed using ImageJ software which we used to evaluate structural units sizes (presintered sample already shown in Fig. 2), while the histogram of particles size volume fractions is shown in Fig. 5b. The histogram shows that most of the structural units in the pre-sintered material had diameters smaller than 10 μm. As can be seen in all the deformed structures in Fig. 4b–d, some grains have boundaries surrounded with Al2O3 (white particles), whereas some grains have boundaries without any presence of the oxide. These are grains formed during deformation processing. Different orientations of these newly nucleated grains can be distinguished according to their different shades of gray in the SEM images. The microstructure of the swaged sample experienced the lowest imposed strain. As can be seen in Fig. 4b, this strain was sufficient for nucleation of new grains within the ductile Al phase. However, the brittle and hard alumina particles did not experience full consolidation and voids surrounded with oxides still remained evident on some grain boundaries. ImageJ analysis results,
238
L. Kunčická et al. / Materials Science & Engineering A 646 (2015) 234–241
Fig. 5. Structure of pre-sintered material processed with ImageJ software (a); volume fractions of particles according to diameter (b).
Fig. 6. Orientation imaging map; pre-sintered state (a), after swaging (b), after 1 HPT revolution (c), after 10 HPT revolutions (d).
L. Kunčická et al. / Materials Science & Engineering A 646 (2015) 234–241
calculated based on four longitudinal and four transversal observations, showed the average grain diameter of 2.75 μm. This value was the largest among the deformed structures. With progressing plastic deformation, higher imposed strains result in more significant grain refinement and also in a redistribution of the microstructural components [32]. Both can be observed in the samples deformed by HPT. The structure of the sample after 1 revolution still exhibited clusters of coarse oxides. Nevertheless, redistribution of oxide layers into smaller individual particles is already evident in certain locations. Besides formation of new grains, this structure also exhibited light gray areas with higher amount of Al2O3. These are most probably remnants of particles and grain boundaries located slightly below the observed surface, which were also scanned by the electron probe. The average grain diameter after 1 HPT revolution was 0.47 μm. The influence of high shear can be seen the most in the sample deformed by 10 revolutions. The final grain refinement was the most significant in this sample. The structure was most highly homogenized, with alumina redistributed into a more homogenous pattern. The high strain also influenced the arrangement of grains and oxide particles into bands. It is evident that the spacing of oxide particle bands was usually 1–2 μm, although the average grain diameter, calculated from four different locations was only 0.22 μm, somewhat smaller compared to as the sample after 1 HPT revolution. Several bands of newly formed sub-micron grains can therefore be observed between the bands of oxides originating from the powder particles surfaces. 3.5. Grains orientations and texture All the conditions of the individual processes also influence the final grain shape. Orientation imaging maps (OIMs) for the deformed samples, as well as of the undeformed pre-sintered sample are depicted in Fig. 6a–d. As can be seen in Fig. 6a, the orientation of the structural units in the undeformed sample were random,
239
however this structure also contained a high amount of pores as could be seen in Fig. 2. Grain orientations for the samples processed with HPT were more or less random (Fig. 6c and d), while the situation for the swaged sample was entirely different (Fig. 6b). The strain imposed into the HPT samples was remarkably higher than for the swaged sample. The accumulated strain energy caused significant refinement of the original structural units, as could be seen in Fig. 4. The newly formed grains were small and more or less equiaxed with random orientations. On the other hand, the structural units after swaging were mostly oriented in the 〈111〉 direction with some grains possessing an approximately 45° inclination to the 〈001〉 direction. Similar grain orientations were observed for swaged Cu by Otto et al. [33]. This fiber structure was imparted by the above mentioned geometrical conditions during swaging, when axial material flow is encouraged by the constraints provided by the surrounding swaging dies. The influence of the deformation processes on grains orientation can be understood by additional texture analyses, as described next. Inverse pole figure (IPF) analyses were performed to evaluate the differences between the influences of the two processes on orientations of grains. [001] IPFs for all the states are shown in Fig. 7a–d. In the unprocessed pre-sintered sample, a slight notion of 〈111〉 fiber formation can be seen (Fig. 7a). However, the texture is more or less random – considering the texture intensity scale, the minimum value was slightly less than 1x random, while the maximum value was slightly higher than 1.3x random. This makes the difference between the minimum and maximum value smaller than 0.5. The most significant texture formation was found in the swaged sample, in which a strong 〈111〉 fiber texture developed (Fig. 7b). The maximum intensity value was as high as almost 9x random. During the swaging process, our pre-sintered material was consolidated and at the same time the newly formed grains oriented in their preferred orientations along the unconstrained
Fig. 7. [001] inverse pole figures; pre-sintered state (a), after swaging (b), after 1 HPT revolution (c), after 10 HPT revolutions (d).
L. Kunčická et al. / Materials Science & Engineering A 646 (2015) 234–241
240
axial direction. We previously observed such texture development for another type of swaged Al sample [3], although the material investigated in the mentioned study was cast. The IPFs for the samples processed with HPT confirmed the conclusions made on the basis of OIMs. The grains did not exhibit any strong texture formation. Although both the samples were inclined to form a 〈011〉 C fiber texture component, the maximum intensities were only 1.51x random for the sample after 1 revolution and 1.85x random for the sample after 10 revolutions. Similar texture formation was also observed by Orlov et al. [34] for a HPT processed CP Al sample, although the intensity observed in the mentioned study after a similar amount of the imposed strain was slightly higher. This is most probably caused by consumption of some portion of the imposed strain on consolidation of the individual particles and densification of the pre-sintered structure of our material. Also notable is the only slight difference between the textures for both the HPT samples. This is most probably caused by a saturation of texture which occurs after a certain amount of imposed shear strain [9]. After reaching this state, the grain orientations do not exhibit any further rearrangement with additional straining. 3.6. Microhardness Microhardness for all the deformed samples was measured at individual points along a line passing through the axis of a sample on its transverse cross-sectional cut. The measurements were made at 1 mm spacings starting at a distance of 0.5 mm from the edge of the samples. For the samples deformed with HPT the analyses were performed along bottom, top and central diametral lines. Results of all the HPT measurements are depicted in Fig. 8, while an example of an indentation for the swaged sample is shownin Fig. 9. The values measured along the diameter of the swaged sample did not exhibit any decreasing/increasing trend towards the central axis of the sample. As was already shown in Section 3.3, the pores in the structure of this sample were more or less homogenously distributed. Therefore, the measured microhardness values did not exhibit any remarkable trends and only the average value was calculated for this sample. The results of measurements for the samples processed with HPT exhibited trends typical for this type of processing [35]. Along the horizontal direction, the lowest microhardness values were generally observed in the central area of a sample in the vicinity of its rotational axis. This is caused by the character of HPT. During the HPT process the die rotates along the central axis, which is identical to the axis of the deformed sample. Therefore, the distribution of the imposed strain from the axis towards the surfaces
Fig. 8. Microhardness values for HPT processed samples.
Fig. 9. Example of a microhardness measurement indentation (swaged sample).
of the dies is not uniform. The smallest amount of shear strain is imposed into the central area of the sample, while towards the vertical walls of the dies the strain increases within the volume of the sample and decreases again in the areas close to the die walls. Numerical simulations performed by Lee et al. [36] showed that even a dead zone with unprocessed material can occur in the corners of the dies. This is most probably the reason why all the microhardness curves for top and bottom areas of the deformed samples in Fig. 8 exhibited a drop of the values at the corners of the samples. However, with an increasing number of revolutions the imposed strain increases also in the corner regions. Towards the surfaces of the sample that are in contact with the top and bottom horizontal dies surfaces, the deformation along a horizontal line in the vicinity of the surfaces is inhomogeneous as well. As mentioned above, the imposed strain in the area close to the sample axis as well as in the corners of the dies is low. However, between these two areas the imposed strain increases due to the influence of friction of the sample with the dies. Increasing friction supports shearing of the grains, which subsequently leads to their refinement and eventual increase in hardness. This conclusion is consistent with the work of Jahedi et al. [35]. When the overall imposed strain is lower (the sample is deformed with lower number of revolutions), the differences in the imposed strain in different areas of a deformed sample are significant. Nevertheless, progressively larger deformation makes the strain throughout the sample more uniform, including in the peripheral corner areas. Therefore the distribution of microhardness along the horizontal direction is more uniform in the sample after 10 revolutions, than in the one after 1 revolution for all the measured horizontal hardness traces. Notable also is the occurrence of several local deviations from the above mentioned trends, especially in the sample after 1 revolution. Whereas the total strain is a substantial factor influencing mechanical properties of a deformed material, differences between the imposed strain throughout the cross-section of a sample are not sufficiently significant to substantially influence microhardness values [37]. A more significant factor influencing mechanical properties, and therefore hardness, is the macroscale structure at the measured locations. Considering the fact that the deformed samples were originally pre-sintered and consolidated from powders, the deviations in the values of microhardness are most probably caused by remaining porosity. This conclusion is also supported by the fact that the local deviations were mostly observed in the sample deformed with 1 revolution, the structure
L. Kunčická et al. / Materials Science & Engineering A 646 (2015) 234–241
of which exhibited residual isolated pores, while the homogenously compacted structure of the sample after 10 revolutions resulted in relatively homogenous microhardness values. A comparison of the average microhardness values for all the states is in accordance with the calculations of average strains. Higher microhardness values were obtained for samples with higher average imposed strains. Whereas the microhardness of the pre-sintered sample was 29.6 HV, after swaging it increased to 50.1. For the samples processed with HPT, the average microhardness values were even higher, namely 60.7 HV after 1 revolution and 72.8 HV after 10 revolutions. The microhardness values increased due to the imposed shear strain, which imparts deformation strengthening, as well as due to more complete consolidation of the pre-sintered samples. The observed changes in microhardness values are also related to the size and distribution of alumina particles in the structures of the deformed samples. As was reported by Rahimian et al. [30], hardness of Al/Al2O3 composites increases with decreasing size of Al2O3 particles. Decreasing size and redistribution of alumina particles in our composite caused by increasing imposed shear strain likely contributed to the increase in hardness for the HPT processed samples, especially the sample deformed with 10 revolutions. Decreasing grain size certainly had a non-negligible influence as well. Rahimian et al. reported results for 10 wt% alumina, the same level as in our experiments. However, their smallest particle size was 3 μm, while virtually all of the particles in our experiments are sub-micron. In addition, our average grain size was significantly smaller as a result of HPT. For 10 HPT revolutions we achieve 100% density, compared with their 97% relative density. Using the approximate conversion for wrought aluminum from the ASTM standard E140-02 to express Vickers hardness in Brinell units, our maximum measured hardness for 10 revolutions of HPT was 77.5 BN, notably higher than the maximum of 70 BN reported by Rahimian for their samples containing 10 wt% of 3 μm alumina particles.
4. Conclusion Evaluation of microstructural characteristics and microhardness of Al/Al2O3 composite fabricated from pre-sintered Al powder samples by 4 swaging passes, 1 HPT revolution and 10 HPT revolutions showed a substantial influence of both the methods on consolidation of the structure. However, HPT caused better consolidation, completely eliminated remaining porosity in the sample after 10 revolutions. In this sample the grain size was also smallest and the alumina particles were most highly refined and redistributed. Newly formed grains were observed in the structures of all the deformed samples. Samples processed by HPT exhibited a slight tendency to form a 〈011〉 C fiber texture, while the swaged sample exhibited a strong 〈111〉 texture fiber. Microhardness values were increased by straining in all the samples when compared to the pre-sintered state. Microhardness was the highest for the sample subject to the greatest strain, 10 revolutions of HPT, and for this case, exceeding the levels reported by others performing conventional compaction with a comparable volume fraction of alumina. This study has shown a favorable influence of severe plastic deformation compaction via HPT on consolidation of particles and the elimination of porosity. Our future work will be focused on examination of redistribution of reinforcing phases by the influence of severe shear strain and the resulting effects upon mechanical properties. This work shows that favorable results can be
241
obtained by processing of Al powder particles with already present Al oxides to fabricate an Al/Al2O3 dispersion strengthened composite without the need to add additional alumina particles.
Acknowledgments This paper was created in the Project no. LO1203 “Regional Materials Science and Technology Centre-Feasibility Program” funded by Ministry of Education, Youth and Sports of the Czech Republic. The authors from the Colorado School of Mines are grateful for the support of the George S. Ansell Department of Metallurgical and Materials Engineering Proposal Development Fund.
References [1] V.V. Stolyarov, Y.T. Zhu, T.C. Lowe, R.K. Islamgaliev, R.Z. Valiev, Mater. Sci. Eng. A 282 (2000) 78. [2] H. Jiang, J.A. Valdez, Y.T. Zhu, I.J. Beyerlein, T.C. Lowe, Compos. Sci. Technol. 60 (2000) 1753. [3] R. Kocich, A. Macháčková, L. Kunčická, F. Fojtík, Mater. Des. 71 (2015) 36. [4] P. Gudlur, A. Forness, J. Lentz, M. Radovic, A. Muliana, Mater. Sci. Eng. A 531 (2012) 18. [5] A. Russell, K.L. Lee, Structure–Property Relations in Nonferrous Metals, Wiley, New Jersey, 2005. [6] T.C. Lowe, J. Lipkin, J. Mech. Phys. Solids 39 (1991) 417. [7] R. Kocich, A. Macháčková, F. Fojtík, Int. J. Mech. Sci. 64 (2012) 54. [8] S. Zherebtsov, E. Kudryavtsev, S. Kostjuchenko, S. Malysheva, G. Salishchev, Mater. Sci. Eng. A 536 (2012) 190. [9] H. Jiang, Y.T. Zhu, D.P. Butt, I.V. Alexandrov, T.C. Lowe, Mater. Sci. Eng. A 290 (2000) 128. [10] R. Kocich, M. Kursa, A. Macháčková, Acta Phys. Pol. A 122 (2012) 581. [11] F. Djavanroodi, M. Ebrahimi, Mater. Sci. Eng. A 527 (2010) 1230. [12] L. Kunčická, R. Kocich, J. Drápala, V.A. Andreyachshenko, in: Proceedings of the 22nd International Met. Mater. Conference, 2013, pp. 391. [13] I.P. Semenova, A.V. Polyakov, G.I. Raab, T.C. Lowe, R.Z. Valiev, J. Mater. Sci. 47 (2012) 7777. [14] R. Kocich, L. Kunčická, M. Mihola, K. Skotnicová, Mater. Sci. Eng. A 563 (2013) 86. [15] R. Kocich, A. Macháčková, L. Kunčická, Mater. Sci. Eng. A 612 (2014) 445. [16] Y. Beygelzimer, V. Varyukhin, S. Synkov, D. Orlov, Mater. Sci. Eng. A 503 (2009) 14. [17] J. Huang, Y.T. Zhu, D.J. Alexander, X. Liao, T.C. Lowe, R.J. Asaro, Mater. Sci. Eng. A 371 (2004) 35. [18] H.S. Arora, H. Singh, B.K. Dhindaw, Int. J. Adv. Manuf. Technol. 61 (2012) 1043. [19] I.V. Alexandrov, Y.T. Zhu, T.C. Lowe, R.K. Islamgaliev, R.Z. Valiev, Nanostruct. Mater. 10 (1998) 45. [20] M. Das, G. Das, M. Ghosh, M. Wegner, V. Rajnikant, S. GhoshChowdhury, T. K. Pal, Mater. Sci. Eng. A 558 (2012) 525. [21] A. Nouri, P.D. Hodgson, C.E. Wen, Acta Biomater. 6 (2010) 1630. [22] H. Ahmadzadeh, B.R. Freedman, B.K. Connizzo, L.J. Soslowsky, V.B. Shenoy, Acta Biomater. 22 (2015) 83–91. [23] P. Gudlur, A. Boczek, M. Radovic, A. Muliana, Mater. Sci. Eng. A 590 (2014) 352. [24] R. Kocich, S.P.D. Metody, Od Mikrometrů K Nanometrům, 1st ed., Marionetti Press, Ostrava, 2011. [25] T.G. Langdon, Acta Mater. 61 (2013) 7035. [26] M.A. Abdulstaar, E.A. El-Danaf, N.S. Waluyo, L. Wagner, Mater. Sci. Eng. A 565 (2013) 351. [27] O.N. Senkov, D.B. Miracle, J.M. Scott, S.V. Senkova, J. Alloy. Compd. 365 (2004) 126. [28] R. Lapovok, D. Tomus, J. Mang, Y. Estrin, T.C. Lowe, Acta Mater. 57 (2009) 2909. [29] A.P. Zhilyaev, T.G. Langdon, Prog. Mater. Sci. 53 (2008) 893. [30] M. Rahimian, N. Parvin, N. Ehsani, Mater. Sci. Eng. A 527 (2010) 1031. [31] R.D. Doherty, B. Verlinden, J. Driver, I. Samajdar, Thermo-Mechanical Processing of Metallic Materials, Elsevier, Amsterdam, 2007. [32] R. Derakhshandeh Haghighi, A. Jenabali Jahromi, Mater. Des. 32 (2011) 3377. [33] F. Otto, J. Frenzel, G. Eggeler, J. Alloy. Compd. 509 (2011) 4073. [34] D. Orlov, P.P. Bhattacharjee, Y. Todaka, M. Umemoto, N. Tsuji, Scr. Mater. 60 (2009) 893. [35] M. Jahedi, M. Knezevic, M.H. Paydar, J. Mater. Eng. Perform. 24 (2015) 1471. [36] D.J. Lee, E.Y. Yoon, L.J. Park, H.S. Kim, Scr. Mater. 67 (2012) 384. [37] Y. Cao, Y.B. Wang, R.B. Figueiredo, L. Chang, X.Z. Liao, M. Kawasaki, W.L. Zheng, S.P. Ringer, T.G. Langdon, Y.T. Zhu, Acta Mater. 59 (2011) 3903.
Contents lists available at ScienceDirect
Materials Science & Engineering A journal homepage: www.elsevier.com/locate/msea
Synthesis of an Al/Al2O3 composite by severe plastic deformation Lenka Kunčická a,b, Terry C. Lowe c,n, Casey F. Davis c, Radim Kocich a,b, Martin Pohludka a a b c
Regional Materials Science and Technology Centre, VŠB-TU Ostrava, 17. listopadu 15, 70833 Ostrava-Poruba, Czech Republic Department of Material Forming, Faculty of Metallurgy and Materials Engineering, VŠB-TU Ostrava, 17. listopadu 15, Ostrava-Poruba 70833, Czech Republic The George S. Ansell Department of Metallurgical & Materials Engineering, Colorado School of Mines, Golden, CO 80401, USA
art ic l e i nf o
a b s t r a c t
Article history: Received 7 June 2015 Received in revised form 20 August 2015 Accepted 22 August 2015 Available online 24 August 2015
The influence of severe plastic deformation on fabrication of an aluminum/alumina composite made from pre-sintered Al powders was evaluated. Spherical powder particles with diameters up to 45 μm were pre-compacted using cold isostatic pressing (CIP), subsequently vacuum sintered for 60 min at 500 °C, and subjected to processing using either 4 passes of swaging or 1 or 10 revolutions of high pressure torsion (HPT). For all the samples we calculated average strain and measured microhardness. Results showed an increase in microhardness with an increase of the imposed strain. However, the microhardness values and uniformity were also influenced by possible residual porosity. The smallest and most uniform grain size was achieved for the samples processed by HPT, especially after 10 revolutions (average diameter of 0.22 μm). Residual porosity was completely eliminated only after 10 HPT revolutions. Texture evaluations showed 〈111〉 fiber texture development after swaging, while grain orientations after HPT were more random. & 2015 Elsevier B.V. All rights reserved.
Keywords: Metal matrix composites (MMCs) Powder processing Electron microscopy Hardness
1. Introduction Metal matrix composites (MMC) are among the main trends in recent advanced materials research, especially with the emergence of powder-based additive manufacturing methods. These composites consist of a metal matrix reinforced with load bearing constituents that are designed to exploit the favorable properties of each component. The reinforcing elements can either be dispersed particles [1], fibers [2], or layers [3]. The most widely studied MMC system is Discontinuous Reinforced Aluminum (DRA), which contains particles or short fibers. The popularity of DRA is because aluminum is among the most inexpensive lightweight matrix metals. In addition, because of its low melting point, it is easy to combine with many different reinforcing particles or fibers. The MMC on which this research is focused is aluminum strengthened with alumina (Al2O3) particles. This combination of constituents provides higher strength, stiffness, hardness and chemical and thermal stability compared to aluminum [4]. The already favorable mechanical properties of composites can further be improved by refinement of structural units, which results in a higher volume of grain boundaries (GBs) in the deformed structure [5]. This contributes not only to strengthening via introduction of obstacles for movement of dislocations, but also to improvement of plasticity, since when the grain size is small n
Corresponding author. E-mail address: [email protected] (T.C. Lowe).
http://dx.doi.org/10.1016/j.msea.2015.08.075 0921-5093/& 2015 Elsevier B.V. All rights reserved.
enough, plastic deformation by grain boundary sliding becomes possible. Fine scale substructural features (e.g. cell, microbands) can be introduced by imposing a high amount of strain, which can advantageously be performed using methods of severe plastic deformation. The principle of these technologies lies in imposing a significant amount of shear strain into the material, which consequently results into an increase in mechanical properties, especially strength [6]. These methods include dozens of technologies, some of which are based on conventional methods, such as accumulative roll bonding (ARB) [7], multiaxial forging [8] and swaging [3], but also special severe plastic deformation (SPD) technologies, among which belong methods like high pressure torsion (HPT) [9], equal channel angular pressing (ECAP) [10], ECAP with back pressure (ECAP-BP) [11], ECAP with partial back pressure (ECAP-PBP) [12], ECAP-Conform [13], twist channel angular pressing (TCAP) [14], twist channel multi-angular pressing (TCMAP) [15], twist extrusion (TE) [16], repetitive corrugation and straightening (RCS) [17], friction stir processing (FSP) [18] and many more. However, achievement of finer final structural units is easier when the original material already contains small structural units such as found in some particles, including powders and nanopowders [19]. The particle surfaces also introduce additional interfaces, which can be engineered to impart specific properties when compacted into bulk solids. Shear strain imposed into the material during the above-mentioned processes supports shearing of the powder particles, which results in additional refinement of
L. Kunčická et al. / Materials Science & Engineering A 646 (2015) 234–241
structural units. It is also very beneficial for effective rearrangement of particles within the volume of the deformed material. Although strain imposed into the material during one pass (or revolution) is limited, all of the above mentioned processes can be performed repetitively. Therefore, the overall strain that can be accumulated in the material during multiple passes can be very large. However, most materials reach a saturated steady state condition and do not exhibit any further microstructural changes or increases in the properties with increasing strain [20]. This study focuses on preparation of an aluminum/alumina composite using severe plastic deformation of pre-sintered Al powder. Since the Al particles are already coated with a protective alumina layer, no Al2O3 particles were added. In order to evaluate the influence of strain on microstructural changes, we selected processing with severe plastic deformation by HPT and compared it with a more conventional process, during which the imposed strain is not so high (swaging). Before deformation processing, the pre-sintered Al material was subjected to evaluation of oxygen and alumina content, particle diameter volume fraction analysis and selected mechanical properties measurement. After deformation, we performed electron backscatter diffraction (EBSD) analyses and microstructural observations using scanning electron microscopy (SEM) and measurements of microhardness for the deformed samples. Analyses were also supplemented with calculations of degree of deformation.
2. Materials and methods 2.1. Material preparation We selected an Al powder with a distribution of particle sizes up to 45 μm. An SEM image of the original powder is shown in Fig. 1. We compacted the particles in our own fabricated rubber die using cold isostatic pressing (CIP) device located at Technical University of Ostrava (VŠB-TUO), pressurizing to 200 MPa for 30 s and then immediately depressurizing. The pre-compacted samples were sintered at 500 °C for 60 min in a Zwick vacuum creep testing furnace at ASCR, Brno. Although the working diameter of the dies for both the deformation technologies is 20 mm, we designed the CIP rubber die with a slightly larger diameter (25 mm), since we expected shrinkage of the loose powder during the compaction. After having performed the compression and sintering procedures, we sliced the pre-sintered material into 6 mm thick samples which we subsequently subjected to plastic deformation processing at VŠB-TUO. The deformation by HPT was performed on two
Fig. 1. SEM image of the original powder particles.
235
samples at 5 GPa and 0.5 RPM. One sample was subject to 1 revolution, and the other one to 10 revolutions. Swaging was performed on a cylindrical shaped sample in 4 swaging steps, during which the diameter was reduced from the initial 20 mm to the final 10 mm. All deformation processing was conducted at room temperature without any heat treatment and sintering processes. The deformed samples were subsequently analyzed and the changes in microstructures induced by the plastic deformation technologies were compared. Back-Scattered Electron (BSE) SEM analyses observing the initial powder particles, pre-sintered material and eventually deformed structures were performed at VŠBTUO and Colorado School of Mines (CSM), CO, USA, while electron backscattered diffraction (EBSD) microstructure and texture examinations were carried out at CSM. Prior to the analyses, the samples were ground and polished mechanically and eventually ion milled with a JEOL CP Cross Section Polisher. The locations selected for EBSD microstructural observations were in the midradius on the transverse cross-section for all the samples. The EBSD scan step for the pre-sintered sample was 0.5 μm, while for the swaged sample it was 0.3 μm, 0.1 μm for HPT after 1 revolution and 50 nm for HPT after 10 revolutions, respectively. Porosities, particle size, and grains size in all the states of the material, including the pre-sintered state, were evaluated using images obtained by BSE-SEM observations, and further processed and evaluated using ImageJ software. X-ray analyses of phase composition were performed using the Philips X'Pert system at CSM, CO, USA. Copper with Kα1 wavelength of 1.54054 Å and Kα2 wavelength of 1.54439 Å was used as the X-ray source. A Vickers microhardness measuring device at VŠB-TUO was used to measure microhardness. All the samples were tested under a load of 250 g for the time of 7 s. Hardness measurements of all the samples were performed across their diameters in order to investigate possible non-uniformities within their cross-sections. Average microhardness values from all the measured individual values were calculated as well. Young's and shear moduli of the pre-sintered sample were measured by the IMCE n.v. company in Belgium using a RFDA measuring system (http://www.imce.eu/) in accordance with the ASTM E 1876 standard in order to be able to evaluate the assumption of porosity throughout the entire volume of the pre-sintered material. All the moduli measurements were performed ten times and the average values were eventually calculated.
3. Results and discussion 3.1. Properties of pre-sintered material To be able to subsequently evaluate the influence of the deformation technologies, selected properties and parameters were observed for the pre-sintered material. First we performed SEM and X-ray analyses to observe the microstructures and evaluate chemical composition, especially to determine oxygen content. Subsequently, we measured microhardness throughout the crosssection of the pre-sintered material and calculated its average value. Eventually, elastic and shear moduli were evaluated. The results of the measurements are summarized in Table 1, while Fig. 2 shows an SEM image of the pre-sintered microstructure. The results of SEM analyses showed approximately 5 wt% of oxygen in the structure. The source of oxygen is the protective aluminum oxide layer, which initially coated the surface of all the particles. The result of X-ray analysis depicted in Fig. 3 confirmed the presence of oxygen in the form of alumina (Al2O3). The most distinctive peaks in the plot are peaks corresponding to the pure Al phase. Nevertheless, noticeable at low diffraction angles, the detail of which is depicted in the corner of Fig. 3, is diffuse
L. Kunčická et al. / Materials Science & Engineering A 646 (2015) 234–241
236
Table 1 Mechanical characteristics of pre-sintered material. Property
Unit
Value
Elastic modulus Shear modulus Poisson's ratio Microhardness Oxygen content Alumina content
[GPa] [GPa] [–] [HV] [wt%] [wt%]
53.98 19.94 0.353 29.58 5.0 10.6
homogenous isotropic behavior during loading [22]. According to the chemical composition and structure of alumina, the content of the individual elements in the chemical compound is 52.9 wt% Aluminum and 47.1 wt% Oxygen. Since the overall oxygen content in this material was determined to be 5.0 wt%, the supposed content of Al2O3 in the sample is 10.6 wt%. The material can therefore be considered as Al-matrix composite with approximately 10 wt% of Al2O3 strengthening phase. Values of elastic modulus similar to ours were observed by Gudlur et al. for Al/Al2O3 composite with residual porosity of 7% and nominal content of alumina 10% [23]. Similarity of the results also confirms the observed and calculated alumina content. 3.2. Imposed strain The first step during our evaluation procedure of the deformed samples was calculation of the imposed strain. This we performed for swaging according to Eq. (1), as well as for HPT according to Eq. (2) [3,9,24].
S0 Sn
(1)
2πrN t
(2)
φ = ln
φ= Fig. 2. SEM image of pre-sintered structure.
Fig. 3. X-ray plot for pre-sintered sample.
scattering, indicative of the semi-crystalline or non-crystalline nature of the alumina phase. Since the above described observations of distinctive peaks are valid only for crystalline structures, it is probable that Al2O3 is also present in the structure in an amorphous state. In a reference plot of X-ray data for pure Al2O3, a set of distinctive peaks starts to rise at the diffraction angle of 28° (not shown here). Although the low intensities in our plot are probably burdened with noise, a slight increase forming a small local peak can be distinguished between the angles of 28° and 29°. This is associated with the presence of Al2O3. As can be seen in Table 1, both the measured moduli are lower than the theoretical values for aluminum. Considering the elastic modulus, this value is approximately 77% of the theoretical one (70 GPa). This is caused mainly by residual porosity, since porosity has the ability to decrease the moduli [21]. The average porosity for the compressed sample before sintering was 20.76%, while after sintering, the porosity decreased to the average value of 7.74%. Fig. 2 clearly shows the presence of voids and pores in the pre-sintered material. The measured Poisson's ratio indicates
where S0 and Sn are the cross-sectional areas of the swaged sample before and after a swaging step [mm2], respectively, r is radius of the sample deformed using HPT in the measured location [mm], N is number of HPT revolutions and t is thickness of the sample [mm]. According to these calculations, the average strain imposed into the material during swaging was 0.27 after the first pass, 0.58 after the second pass, 0.94 after the third pass and eventually 1.39 after the fourth pass. For the HPT processed samples, the imposed strain was significantly different at the observed mid-radius location of both the samples. Whereas the calculated strain was 5.24 for the sample after one revolution, it was 52.36 for the sample after ten revolutions. The differences in the imposed strains are due to the different designs of the deformation technologies. Considering pressure conditions for the two processes, conditions for HPT are more favorable for particle compaction. During HPT the flow of the material is theoretically constrained by the closed dies surrounding the sample. However, the real conditions are quasi-constrained, since there is a slight gap between the two dies and the volume of a sample is slightly larger than the volume between the closed dies [25]. During HPT processing the gap gets filled with flash, which subsequently limits outward material flow. The high axial compressive force during HPT restricts the sample from sliding and rotating and also supports imposing of a high amount of shear strain into the material. On the other hand during swaging the material flow is limited in the radial direction due to the presence of the swaging dies, while the axial direction is not constrained. Therefore, material can flow out from the contact region of the swaging dies in the axial direction, which consequently reduces the amount of strain that can be accumulated in the sample [26]. 3.3. Porosity Microstructures of the samples deformed by swaging, 1 HPT revolution and 10 HPT revolutions are shown in Fig. 4a to 4d, respectively. The best consolidation of the pre-sintered powder material was found after 10 HPT revolutions (Fig. 4d). The microstructure of this sample did not exhibit any residual porosity.
L. Kunčická et al. / Materials Science & Engineering A 646 (2015) 234–241
237
Fig. 4. Microstructures of the deformed samples; swaged (a) (b), after 1 HPT revolution (c), after 10 HPT revolutions (d).
Sporadic isolated pores were still evident in the sample after 1 HPT revolution (Fig. 4c). The volume fraction of pores remaining in the structure of this sample was 0.78%. The swaged sample exhibited the largest residual porosity among the deformed samples, namely 5.67% (Fig. 4b). As can be seen in the microstructure of this sample in Fig. 4a, the pores were homogenously distributed in the swaged structure. According to these results residual porosity decreased with an increase in imposed strain. A similar conclusion was also reported e.g. by Senkov et al. within their study of an Al-based fine powder compound processed by ECAP [27]. Shear strain is especially efficient for reduction/elimination of porosity via rearrangement of powder particles within a sample, since by its influence smaller particles can fill free volumes between larger particles [28]. When comparing the HPT process to the swaging process from the point of view of the amount of the imposed shear strain, this is more significant for HPT. During swaging the axis of the sample is perpendicular to the axis of the imposed strain, while during HPT the axis of the deformation process is identical to the axis of the sample and also to the axis of the compacting force [29]. This combination enables one to impose the highest possible amount of shear strain into the deformed material. Therefore, pores in the structure of the sample deformed with 10 revolutions were eliminated since the particles experienced the most significant rearrangement during processing. 3.4. Grain size The microstructure of the undeformed pre-sintered material has already been shown in Fig. 2. The Figure clearly shows that the initial Al powder particles are all covered with a protective oxide layer. Due to its high melting temperature, alumina cannot melt or evaporate during the sintering process and remains in the
structure. Deformation processing results in straining of the aluminum matrix structure and subsequent development and movement of dislocations. At relatively low strains, brittle structural components (alumina) are largely fractured or experience no deformation, while the more ductile structural matrix (aluminum) is readily deformed [27]. The smallest alumina particles and also their cracked fragments act as obstacles for dislocations movement [30]. At higher strains, dislocations developed inside the grains/particles accumulate at these obstacles. The accumulated energy consequently results into a formation of subgrains, which eventually results into a formation of new grains and therefore grain refinement [31]. Larger alumina particles that remain in the matrix after deformation processing are able to bear load, while the smaller particles are more effective barriers to dislocation motion. Fig. 5a depicts an example of structure processed using ImageJ software which we used to evaluate structural units sizes (presintered sample already shown in Fig. 2), while the histogram of particles size volume fractions is shown in Fig. 5b. The histogram shows that most of the structural units in the pre-sintered material had diameters smaller than 10 μm. As can be seen in all the deformed structures in Fig. 4b–d, some grains have boundaries surrounded with Al2O3 (white particles), whereas some grains have boundaries without any presence of the oxide. These are grains formed during deformation processing. Different orientations of these newly nucleated grains can be distinguished according to their different shades of gray in the SEM images. The microstructure of the swaged sample experienced the lowest imposed strain. As can be seen in Fig. 4b, this strain was sufficient for nucleation of new grains within the ductile Al phase. However, the brittle and hard alumina particles did not experience full consolidation and voids surrounded with oxides still remained evident on some grain boundaries. ImageJ analysis results,
238
L. Kunčická et al. / Materials Science & Engineering A 646 (2015) 234–241
Fig. 5. Structure of pre-sintered material processed with ImageJ software (a); volume fractions of particles according to diameter (b).
Fig. 6. Orientation imaging map; pre-sintered state (a), after swaging (b), after 1 HPT revolution (c), after 10 HPT revolutions (d).
L. Kunčická et al. / Materials Science & Engineering A 646 (2015) 234–241
calculated based on four longitudinal and four transversal observations, showed the average grain diameter of 2.75 μm. This value was the largest among the deformed structures. With progressing plastic deformation, higher imposed strains result in more significant grain refinement and also in a redistribution of the microstructural components [32]. Both can be observed in the samples deformed by HPT. The structure of the sample after 1 revolution still exhibited clusters of coarse oxides. Nevertheless, redistribution of oxide layers into smaller individual particles is already evident in certain locations. Besides formation of new grains, this structure also exhibited light gray areas with higher amount of Al2O3. These are most probably remnants of particles and grain boundaries located slightly below the observed surface, which were also scanned by the electron probe. The average grain diameter after 1 HPT revolution was 0.47 μm. The influence of high shear can be seen the most in the sample deformed by 10 revolutions. The final grain refinement was the most significant in this sample. The structure was most highly homogenized, with alumina redistributed into a more homogenous pattern. The high strain also influenced the arrangement of grains and oxide particles into bands. It is evident that the spacing of oxide particle bands was usually 1–2 μm, although the average grain diameter, calculated from four different locations was only 0.22 μm, somewhat smaller compared to as the sample after 1 HPT revolution. Several bands of newly formed sub-micron grains can therefore be observed between the bands of oxides originating from the powder particles surfaces. 3.5. Grains orientations and texture All the conditions of the individual processes also influence the final grain shape. Orientation imaging maps (OIMs) for the deformed samples, as well as of the undeformed pre-sintered sample are depicted in Fig. 6a–d. As can be seen in Fig. 6a, the orientation of the structural units in the undeformed sample were random,
239
however this structure also contained a high amount of pores as could be seen in Fig. 2. Grain orientations for the samples processed with HPT were more or less random (Fig. 6c and d), while the situation for the swaged sample was entirely different (Fig. 6b). The strain imposed into the HPT samples was remarkably higher than for the swaged sample. The accumulated strain energy caused significant refinement of the original structural units, as could be seen in Fig. 4. The newly formed grains were small and more or less equiaxed with random orientations. On the other hand, the structural units after swaging were mostly oriented in the 〈111〉 direction with some grains possessing an approximately 45° inclination to the 〈001〉 direction. Similar grain orientations were observed for swaged Cu by Otto et al. [33]. This fiber structure was imparted by the above mentioned geometrical conditions during swaging, when axial material flow is encouraged by the constraints provided by the surrounding swaging dies. The influence of the deformation processes on grains orientation can be understood by additional texture analyses, as described next. Inverse pole figure (IPF) analyses were performed to evaluate the differences between the influences of the two processes on orientations of grains. [001] IPFs for all the states are shown in Fig. 7a–d. In the unprocessed pre-sintered sample, a slight notion of 〈111〉 fiber formation can be seen (Fig. 7a). However, the texture is more or less random – considering the texture intensity scale, the minimum value was slightly less than 1x random, while the maximum value was slightly higher than 1.3x random. This makes the difference between the minimum and maximum value smaller than 0.5. The most significant texture formation was found in the swaged sample, in which a strong 〈111〉 fiber texture developed (Fig. 7b). The maximum intensity value was as high as almost 9x random. During the swaging process, our pre-sintered material was consolidated and at the same time the newly formed grains oriented in their preferred orientations along the unconstrained
Fig. 7. [001] inverse pole figures; pre-sintered state (a), after swaging (b), after 1 HPT revolution (c), after 10 HPT revolutions (d).
L. Kunčická et al. / Materials Science & Engineering A 646 (2015) 234–241
240
axial direction. We previously observed such texture development for another type of swaged Al sample [3], although the material investigated in the mentioned study was cast. The IPFs for the samples processed with HPT confirmed the conclusions made on the basis of OIMs. The grains did not exhibit any strong texture formation. Although both the samples were inclined to form a 〈011〉 C fiber texture component, the maximum intensities were only 1.51x random for the sample after 1 revolution and 1.85x random for the sample after 10 revolutions. Similar texture formation was also observed by Orlov et al. [34] for a HPT processed CP Al sample, although the intensity observed in the mentioned study after a similar amount of the imposed strain was slightly higher. This is most probably caused by consumption of some portion of the imposed strain on consolidation of the individual particles and densification of the pre-sintered structure of our material. Also notable is the only slight difference between the textures for both the HPT samples. This is most probably caused by a saturation of texture which occurs after a certain amount of imposed shear strain [9]. After reaching this state, the grain orientations do not exhibit any further rearrangement with additional straining. 3.6. Microhardness Microhardness for all the deformed samples was measured at individual points along a line passing through the axis of a sample on its transverse cross-sectional cut. The measurements were made at 1 mm spacings starting at a distance of 0.5 mm from the edge of the samples. For the samples deformed with HPT the analyses were performed along bottom, top and central diametral lines. Results of all the HPT measurements are depicted in Fig. 8, while an example of an indentation for the swaged sample is shownin Fig. 9. The values measured along the diameter of the swaged sample did not exhibit any decreasing/increasing trend towards the central axis of the sample. As was already shown in Section 3.3, the pores in the structure of this sample were more or less homogenously distributed. Therefore, the measured microhardness values did not exhibit any remarkable trends and only the average value was calculated for this sample. The results of measurements for the samples processed with HPT exhibited trends typical for this type of processing [35]. Along the horizontal direction, the lowest microhardness values were generally observed in the central area of a sample in the vicinity of its rotational axis. This is caused by the character of HPT. During the HPT process the die rotates along the central axis, which is identical to the axis of the deformed sample. Therefore, the distribution of the imposed strain from the axis towards the surfaces
Fig. 8. Microhardness values for HPT processed samples.
Fig. 9. Example of a microhardness measurement indentation (swaged sample).
of the dies is not uniform. The smallest amount of shear strain is imposed into the central area of the sample, while towards the vertical walls of the dies the strain increases within the volume of the sample and decreases again in the areas close to the die walls. Numerical simulations performed by Lee et al. [36] showed that even a dead zone with unprocessed material can occur in the corners of the dies. This is most probably the reason why all the microhardness curves for top and bottom areas of the deformed samples in Fig. 8 exhibited a drop of the values at the corners of the samples. However, with an increasing number of revolutions the imposed strain increases also in the corner regions. Towards the surfaces of the sample that are in contact with the top and bottom horizontal dies surfaces, the deformation along a horizontal line in the vicinity of the surfaces is inhomogeneous as well. As mentioned above, the imposed strain in the area close to the sample axis as well as in the corners of the dies is low. However, between these two areas the imposed strain increases due to the influence of friction of the sample with the dies. Increasing friction supports shearing of the grains, which subsequently leads to their refinement and eventual increase in hardness. This conclusion is consistent with the work of Jahedi et al. [35]. When the overall imposed strain is lower (the sample is deformed with lower number of revolutions), the differences in the imposed strain in different areas of a deformed sample are significant. Nevertheless, progressively larger deformation makes the strain throughout the sample more uniform, including in the peripheral corner areas. Therefore the distribution of microhardness along the horizontal direction is more uniform in the sample after 10 revolutions, than in the one after 1 revolution for all the measured horizontal hardness traces. Notable also is the occurrence of several local deviations from the above mentioned trends, especially in the sample after 1 revolution. Whereas the total strain is a substantial factor influencing mechanical properties of a deformed material, differences between the imposed strain throughout the cross-section of a sample are not sufficiently significant to substantially influence microhardness values [37]. A more significant factor influencing mechanical properties, and therefore hardness, is the macroscale structure at the measured locations. Considering the fact that the deformed samples were originally pre-sintered and consolidated from powders, the deviations in the values of microhardness are most probably caused by remaining porosity. This conclusion is also supported by the fact that the local deviations were mostly observed in the sample deformed with 1 revolution, the structure
L. Kunčická et al. / Materials Science & Engineering A 646 (2015) 234–241
of which exhibited residual isolated pores, while the homogenously compacted structure of the sample after 10 revolutions resulted in relatively homogenous microhardness values. A comparison of the average microhardness values for all the states is in accordance with the calculations of average strains. Higher microhardness values were obtained for samples with higher average imposed strains. Whereas the microhardness of the pre-sintered sample was 29.6 HV, after swaging it increased to 50.1. For the samples processed with HPT, the average microhardness values were even higher, namely 60.7 HV after 1 revolution and 72.8 HV after 10 revolutions. The microhardness values increased due to the imposed shear strain, which imparts deformation strengthening, as well as due to more complete consolidation of the pre-sintered samples. The observed changes in microhardness values are also related to the size and distribution of alumina particles in the structures of the deformed samples. As was reported by Rahimian et al. [30], hardness of Al/Al2O3 composites increases with decreasing size of Al2O3 particles. Decreasing size and redistribution of alumina particles in our composite caused by increasing imposed shear strain likely contributed to the increase in hardness for the HPT processed samples, especially the sample deformed with 10 revolutions. Decreasing grain size certainly had a non-negligible influence as well. Rahimian et al. reported results for 10 wt% alumina, the same level as in our experiments. However, their smallest particle size was 3 μm, while virtually all of the particles in our experiments are sub-micron. In addition, our average grain size was significantly smaller as a result of HPT. For 10 HPT revolutions we achieve 100% density, compared with their 97% relative density. Using the approximate conversion for wrought aluminum from the ASTM standard E140-02 to express Vickers hardness in Brinell units, our maximum measured hardness for 10 revolutions of HPT was 77.5 BN, notably higher than the maximum of 70 BN reported by Rahimian for their samples containing 10 wt% of 3 μm alumina particles.
4. Conclusion Evaluation of microstructural characteristics and microhardness of Al/Al2O3 composite fabricated from pre-sintered Al powder samples by 4 swaging passes, 1 HPT revolution and 10 HPT revolutions showed a substantial influence of both the methods on consolidation of the structure. However, HPT caused better consolidation, completely eliminated remaining porosity in the sample after 10 revolutions. In this sample the grain size was also smallest and the alumina particles were most highly refined and redistributed. Newly formed grains were observed in the structures of all the deformed samples. Samples processed by HPT exhibited a slight tendency to form a 〈011〉 C fiber texture, while the swaged sample exhibited a strong 〈111〉 texture fiber. Microhardness values were increased by straining in all the samples when compared to the pre-sintered state. Microhardness was the highest for the sample subject to the greatest strain, 10 revolutions of HPT, and for this case, exceeding the levels reported by others performing conventional compaction with a comparable volume fraction of alumina. This study has shown a favorable influence of severe plastic deformation compaction via HPT on consolidation of particles and the elimination of porosity. Our future work will be focused on examination of redistribution of reinforcing phases by the influence of severe shear strain and the resulting effects upon mechanical properties. This work shows that favorable results can be
241
obtained by processing of Al powder particles with already present Al oxides to fabricate an Al/Al2O3 dispersion strengthened composite without the need to add additional alumina particles.
Acknowledgments This paper was created in the Project no. LO1203 “Regional Materials Science and Technology Centre-Feasibility Program” funded by Ministry of Education, Youth and Sports of the Czech Republic. The authors from the Colorado School of Mines are grateful for the support of the George S. Ansell Department of Metallurgical and Materials Engineering Proposal Development Fund.
References [1] V.V. Stolyarov, Y.T. Zhu, T.C. Lowe, R.K. Islamgaliev, R.Z. Valiev, Mater. Sci. Eng. A 282 (2000) 78. [2] H. Jiang, J.A. Valdez, Y.T. Zhu, I.J. Beyerlein, T.C. Lowe, Compos. Sci. Technol. 60 (2000) 1753. [3] R. Kocich, A. Macháčková, L. Kunčická, F. Fojtík, Mater. Des. 71 (2015) 36. [4] P. Gudlur, A. Forness, J. Lentz, M. Radovic, A. Muliana, Mater. Sci. Eng. A 531 (2012) 18. [5] A. Russell, K.L. Lee, Structure–Property Relations in Nonferrous Metals, Wiley, New Jersey, 2005. [6] T.C. Lowe, J. Lipkin, J. Mech. Phys. Solids 39 (1991) 417. [7] R. Kocich, A. Macháčková, F. Fojtík, Int. J. Mech. Sci. 64 (2012) 54. [8] S. Zherebtsov, E. Kudryavtsev, S. Kostjuchenko, S. Malysheva, G. Salishchev, Mater. Sci. Eng. A 536 (2012) 190. [9] H. Jiang, Y.T. Zhu, D.P. Butt, I.V. Alexandrov, T.C. Lowe, Mater. Sci. Eng. A 290 (2000) 128. [10] R. Kocich, M. Kursa, A. Macháčková, Acta Phys. Pol. A 122 (2012) 581. [11] F. Djavanroodi, M. Ebrahimi, Mater. Sci. Eng. A 527 (2010) 1230. [12] L. Kunčická, R. Kocich, J. Drápala, V.A. Andreyachshenko, in: Proceedings of the 22nd International Met. Mater. Conference, 2013, pp. 391. [13] I.P. Semenova, A.V. Polyakov, G.I. Raab, T.C. Lowe, R.Z. Valiev, J. Mater. Sci. 47 (2012) 7777. [14] R. Kocich, L. Kunčická, M. Mihola, K. Skotnicová, Mater. Sci. Eng. A 563 (2013) 86. [15] R. Kocich, A. Macháčková, L. Kunčická, Mater. Sci. Eng. A 612 (2014) 445. [16] Y. Beygelzimer, V. Varyukhin, S. Synkov, D. Orlov, Mater. Sci. Eng. A 503 (2009) 14. [17] J. Huang, Y.T. Zhu, D.J. Alexander, X. Liao, T.C. Lowe, R.J. Asaro, Mater. Sci. Eng. A 371 (2004) 35. [18] H.S. Arora, H. Singh, B.K. Dhindaw, Int. J. Adv. Manuf. Technol. 61 (2012) 1043. [19] I.V. Alexandrov, Y.T. Zhu, T.C. Lowe, R.K. Islamgaliev, R.Z. Valiev, Nanostruct. Mater. 10 (1998) 45. [20] M. Das, G. Das, M. Ghosh, M. Wegner, V. Rajnikant, S. GhoshChowdhury, T. K. Pal, Mater. Sci. Eng. A 558 (2012) 525. [21] A. Nouri, P.D. Hodgson, C.E. Wen, Acta Biomater. 6 (2010) 1630. [22] H. Ahmadzadeh, B.R. Freedman, B.K. Connizzo, L.J. Soslowsky, V.B. Shenoy, Acta Biomater. 22 (2015) 83–91. [23] P. Gudlur, A. Boczek, M. Radovic, A. Muliana, Mater. Sci. Eng. A 590 (2014) 352. [24] R. Kocich, S.P.D. Metody, Od Mikrometrů K Nanometrům, 1st ed., Marionetti Press, Ostrava, 2011. [25] T.G. Langdon, Acta Mater. 61 (2013) 7035. [26] M.A. Abdulstaar, E.A. El-Danaf, N.S. Waluyo, L. Wagner, Mater. Sci. Eng. A 565 (2013) 351. [27] O.N. Senkov, D.B. Miracle, J.M. Scott, S.V. Senkova, J. Alloy. Compd. 365 (2004) 126. [28] R. Lapovok, D. Tomus, J. Mang, Y. Estrin, T.C. Lowe, Acta Mater. 57 (2009) 2909. [29] A.P. Zhilyaev, T.G. Langdon, Prog. Mater. Sci. 53 (2008) 893. [30] M. Rahimian, N. Parvin, N. Ehsani, Mater. Sci. Eng. A 527 (2010) 1031. [31] R.D. Doherty, B. Verlinden, J. Driver, I. Samajdar, Thermo-Mechanical Processing of Metallic Materials, Elsevier, Amsterdam, 2007. [32] R. Derakhshandeh Haghighi, A. Jenabali Jahromi, Mater. Des. 32 (2011) 3377. [33] F. Otto, J. Frenzel, G. Eggeler, J. Alloy. Compd. 509 (2011) 4073. [34] D. Orlov, P.P. Bhattacharjee, Y. Todaka, M. Umemoto, N. Tsuji, Scr. Mater. 60 (2009) 893. [35] M. Jahedi, M. Knezevic, M.H. Paydar, J. Mater. Eng. Perform. 24 (2015) 1471. [36] D.J. Lee, E.Y. Yoon, L.J. Park, H.S. Kim, Scr. Mater. 67 (2012) 384. [37] Y. Cao, Y.B. Wang, R.B. Figueiredo, L. Chang, X.Z. Liao, M. Kawasaki, W.L. Zheng, S.P. Ringer, T.G. Langdon, Y.T. Zhu, Acta Mater. 59 (2011) 3903.