Effect of laser surface remelting on the microstructure and properties of Al-Al2Cu-Si ternary eutectic alloy

Bimodal ultrafine eutectic composites (BUECs) exhibit a good combination of strength and plasticity owing to a dual-hierarchy in eutectic length-scales in the microstructure. The present study investigates the variation of phase, morphology, feature length-scales and modality of microstructures obtained in a Al81Cu13Si6 (at. %) ternary alloy after laser surface remelting. A novel approach of varying component bimodal eutectic volume fractions by controlling the cooling rate of the laser solidification process has been presented. The volume fraction of the fine eutectic matrix has a profound effect on the flow strength. Laser remelted microstructures with volume fractions of the fine eutectic varying from 25 to 40% exhibiting compressive flow strengths ranging from 500 to 900 MPa have been obtained. The volume fraction of the fine eutectic decreased with cooling rate and completely ceased to exist at cooling rates greater than \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$4\times {10}^{4}\,^\circ {\rm{C}}/{\rm{s}}$$\end{document}4×104°C/s.

solidification conditions has been studied only in alloys with low Si content (1-2 at. %) 39 . Experimental investigations on the Al-rich part of Al-Cu-Si system using casting methods have resulted in bimodal type structures [5][6][7]41 . The emphasis of the previous works has been on mechanical properties and deformation mechanisms of as-cast structures fabricated with minimal process control 5,7,8 . There is at present little information about processing conditions conducive for bimodal eutectic structure formation.
In the present work, we report the effect of processing conditions on the phase, morphology, modality, volume fraction and feature length-scales of solidification microstructure of Al-Cu-Si BUECs. In the past, the primary method for the control of volume fraction involved varying the alloy composition 4,8,14 . However, the processing conditions and mechanical properties of Al 81 Cu 13 Si 6 BUECs with varying volume fractions have not been studied till date. We leveraged the versatile cooling rate capabilities of laser surface remelting to vary critical BUEC microstructure attributes within the melted zone. The hardness and compressive flow strength of the microstructures was investigated. The possible solidification paths under non-equilibrium conditions were discussed to explain the variation in microstructure phase, morphology and volume fraction. This work also presents the upper bound of cooling rate range favourable for BUEC formation; the fine nanocrystalline matrix ceased to form beyond this limit.

Results
Microstructure of as-received samples. The substrate for laser surface remelting was arc cast with a composition of Al 81 Cu 13 Si 6 . Figure 1 shows the SEM micrograph of the as received material and the EDX mapping of the primary elements, Al, Cu and Si. The SEM micrograph of the substrate revealed a ternary eutectic structure with θ -Al 2 Cu and Si phases dispersed uniformly in α -Al matrix. The length scales of the θ -Al 2 Cu phase ranged from a few 100 nanometers to a few 10 microns, at different regions in the material.
Microstructure of laser treated samples. The phase, morphology, length scale and modality of the solidification microstructure varied significantly at different locations within the laser trace. Figure 2 shows the SEM micrographs of the microstructures obtained at various regions within the melt pool, i.e. along the boundary (Fig. 2d-g) and center-line of the laser trace (Fig. 2a-c). The corresponding laser process parameters were power (P): 1000 W, spot size (S S ): 0.8 mm and scanning velocity (S V ): 25.4 mm/s (sample A). A bimodal eutectic structure with two distinct phases with different length-scales of lamellar spacing was observed near the boundary. The coarse eutectic (binary) consisted of α -Al and θ -Al 2 Cu phases while the fine eutectic (ternary) consisted of α -Al, θ -Al 2 Cu and Si phases. Figure 3 shows the SEM micrograph and EDX mapping of the Al, Cu and Si in the bimodal eutectic structure (sample A). Si was primarily confined to the fine matrix. The coarse eutectic had a discontinuous morphology with interphase spacings in the 100-400 nm range; the coarsest structures were obtained along the trace boundary. The volume fraction of the fine eutectic was highest along the trace boundary and gradually decreased towards the interior regions of the laser trace. In the interior regions of the laser trace, extending to the top, the microstructure of previously fine ternary eutectic structure transformed to a binary structure of Si dispersed in an α -Al matrix. The bimodal nature of eutectic length-scales ceased to appear in this region. Thus, this microstructure can be considered as a 'dual-binary eutectic' . Figure 4 shows the SEM micrograph and EDX mapping of the microstructure without an interdendritic ternary eutectic phase. The volume fraction of the binary Al -Si structure was about 5-10%, significantly lower than the volume fraction of the ternary structure observed at the melt pool boundary (40-45%). Figure 5 shows the high-angle annular dark-field (HAADF) TEM images of the two microstructure types. In the binary/ternary bimodal eutectic (Fig. 5a,b), the lamellar spacing of Al-Al 2 Cu eutectic was larger as compared to the dual-binary eutectic (Fig. 5c,d). The absence of the θ -Al 2 Cu phase in the Si rich regions indicated that Cu was restricted to the coarse eutectic phase (Fig. 5d). The finest Al-Al 2 Cu eutectic structure was obtained near the top of the laser trace, on the laser beam axis.
Hardness and flow strength. The Vickers hardness was measured at several locations within the laser trace and as-received sample. The variation of measured hardness (and flow strength estimated from hardness) with volume fraction of the fine eutectic was also studied. Figure 6 shows the methodology adopted to study the effect of volume fraction on the flow strength of the bimodal microstructure (sample A). The projected indent area was measured at each indent location. Furthermore, an SEM image adjacent to the indent was captured to estimate the volume fraction of the fine eutectic structure. The varying contrast between the coarse and fine eutectics was leveraged to estimate volume fractions at multiple locations (with an accuracy of ±5%). The SEM micrograph at each location was transformed to a binary image using image processing functions in MATLAB TM and the volume fraction of the black pixels (fine eutectic region) was obtained. The flow strength was estimated as indentation hardness (in GPa) divided by a factor of 2.7 42 . The laser process parameters were varied to obtain microstructures with different volume fractions of the fine eutectic (see Materials and methods section). Figure 7a shows the variation of flow strength with volume fraction of the fine eutectic. The flow strength varied from 500 to 900 MPa for fine eutectic volume fractions ranging from 25 to 40%. Figure 7b shows the variation of interlamellar spacing in the coarse eutectic with volume fraction of the fine matrix. The Al-Al 2 Cu eutectic became coarser at higher volume fractions of the fine eutectic. The flow strength of the ternary (as-received) and binary eutectic regions were

Discussion
The cooling rate at different locations within the melt pool can be drastically different 43,44 . This resulted in a non-uniform solidification microstructure, in terms of phase, morphology and feature length scales. The highest cooling rates were realized on the centerline of the fusion zone while the lower cooling rates were obtained along the boundary. Thus, fine grained microstructures were obtained at the top of the laser trace along the beam axis and the radial coarsening of grains was observed away from this region 45 . The modification of grain size was also accompanied by a change in the eutectic phase and morphology, a direct result of non-equilibrium solidification at high cooling rates (>10 2 °C/s).
Solidification paths for ternary alloy. The solidification of ternary alloys is complex, with several phases and morphologies possible based on the alloy composition and growth conditions 46 . The formation of phases with specific morphologies under non-equilibrium conditions can be treated by evaluating competitive growth through interface response functions (interface temperature). Upon satisfying thermodynamic stability criteria, the phase and morphology with the highest interface temperature grows 47 . Three types of eutectic structures attain stability during laser surface melting of Al 81 Cu 13 Si 6 ternary alloy and they each grow under a different range of overlapping processing conditions. They are Al-Al 2 Cu-Si ternary eutectic, Al-Al 2 Cu binary eutectic and Al-Si binary eutectic.
In the ternary phase diagram (liquidus projection) of Al-Cu-Si, the binary eutectic point appears as univariant equilibrium lines which meet at ternary invariant point. In the present work, the composition of Al 81 Cu 13 Si 6 coincides with a ternary invariant point which signifies that the four phases (Al, Al 2 Cu, Si and liquid) are in equilibrium at the eutectic temperature of 524 °C. The substrate was arc-cast at relatively lower cooling rates compared to laser melting and can be considered to be closer to the liquidus projection phase diagram. The solidification microstructure of the as-received sample was certainly Al-Al 2 Cu-Si ternary eutectic which was confirmed from the SEM micrograph.
Rapid solidification is a non-equilibrium phenomenon which may result in the undercooling of melt, formation of metastable phases, and coring or microsegregation in the solid solution 21 . The solidification microstructure within the laser trace was therefore expected to deviate significantly from the equilibrium ternary phase diagram prediction. However, at lower scan velocities (<5 mm/s), a fine Al-Al 2 Cu-Si ternary eutectic was obtained within the laser trace. The coupled zone refers to the range of compositions and growth conditions (non-equilibrium) where the solidification microstructure is wholly eutectic 48 . Within the coupled zone, the eutectic structures grow faster than the competing dendritic phases. This concept is primarily applicable to binary systems but can be extended to ternary systems as well 41 . There exists an invariant coupled zone within which the formation of a ternary eutectic phase is more feasible than competing dendritic structures (Al-Al 2 Cu in this case). Outside the invariant coupled zone, the formation of Al-Al 2 Cu dendrites is observed. A univariant coupled zone can be associated with this binary eutectic reaction. The morphology of the Al-Al 2 Cu eutectic is not lamellar since the solidification process is associated with high growth velocities in the presence of a third component (Si). The destabilization of the planar growth interface under rapid solidification conditions can be treated by an absolute stability condition for undercooled alloy melts, which considers the surface energy effect in addition to the constitutional undercooling criterion 49 . Morphological instability is further induced by the formation of a long range solute boundary layer, resulting from the segregation of the third component (Si) 50 .
At higher cooling rates, the solidification path proceeds along the univariant equilibrium line associated with the following binary eutectic reaction: 2 Al and Al 2 Cu dendrites with primary arm spacings ranging from 100-500 nm coprecipitate as a binary eutectic mixture. When the interdendritic melt composition reaches the invariant ternary eutectic composition at 524 °C, it is quenched, resulting in the formation of a fine nanocrystalline matrix 41 .
The final solidification microstructure closely resembled those reported by earlier studies on Al-Cu-Si bimodal eutectic composites [6][7][8] . However, in this study, the feature length scales and volume fractions of constituent eutectics varied significantly within the laser trace. The previous investigations report a small subset of the structures obtained in this study owing to limitations in their solidification processing capabilities. The dendritic arm spacing of the coarse binary eutectic and the volume fraction of the fine matrix were dependent on the solidification conditions (cooling rate). The volume fraction of the fine matrix decreased with finer length scales of the coarse binary eutectic. A shift in primary phase boundary of the coupled zone at high growth velocities has been observed in binary systems 51 . Similarly, the coupled zone associated with the ternary eutectic reaction shifts as the growth velocity increases. The growth velocity increases with distance from the trace bottom, explaining the observed microstructure modality and phase distribution. The evolution of a ternary fine eutectic matrix was favored at lower growth velocities.
Effect of cooling rate on solidification microstructure. The cooling rate of the solidification process had a significant effect on the morphology and phase distribution of the solidification microstructure. However, this change in morphology and elemental distribution was restricted to the phase containing Si, i.e. the ternary Al-Cu-Si and binary Al-Si phases. The cooling rate of the solidification process was experimentally obtained to study its effect on the microstructure phase, morphology and modality. The average cooling rate and associated microstructures (captured by SEM) are shown in Fig. 8. The samples were laser treated by a laser power of 500 W, spot size of 0.8 mm, and different scanning speeds of 2 54, 5.28, 20.32, and 25.4 mm/s. The ternary eutectic structure ceased to form at a cooling rate of ×°4 10 C/s 4 . The bimodal nature of the microstructure was not observed at higher cooling rates; with Si globules precipitating along the Al − Al 2 Cu eutectic colony interfaces. This cooling rate also signifies the limit of the invariant coupled zone of the ternary eutectic solidification process. The decrease in spacing of Al − Al 2 Cu eutectic colonies was continuous regardless of the existence of the fine ternary eutectic matrix. The consistent dendritic morphology of the Al − Al 2 Cu eutectic colonies and absence of any primary phase indicates that the solidification conditions were still confined to the univariant coupled zone associated with the Al − Al 2 Cu binary eutectic reaction.  dependence between the coarse eutectic spacing and fine eutectic volume fraction was observed over a wide range of solidification conditions, realized by varying laser scanning speed and laser power. This result revealed that any attempt to refine the microstructure would result in reduction of volume fraction of fine eutectic. The heterogeneity of length scales in the solidification microstructure reduced and eventually disappeared as the cooling rate was increased. The bimodal nature suppresses shear localization and crack nucleation, improving compressive plastic strain 5 . The shear stress across interfaces between the coarse and fine eutectic colonies is effectively dissipated by the rotation of the coarse eutectic colonies 6 . These effects are likely to be less pronounced at lower volume fractions of the fine eutectic. The high compressive plastic strain which results from this heterogeneity of length scales is expected to be compromised in structures solidified at high cooling rates. Thus, the strength and compressive plasticity can be tuned over a wide range by varying the cooling rate of the solidification process.
The findings of this study will enable more precise control of mechanical properties of rapid solidification processing (e.g. laser surface melting, suction casting, metal based additive manufacturing etc.) of eutectic ternary alloys. The large variation in flow strength and possibly ductility (not addressed in this work) over a narrow range  of cooling rates could result in BUECs finding application in materials processing environments with dynamic requirements. Further studies are required to establish the effects of varying temperature gradients and growth velocities on the morphology and length-scales of BUECs. These parameters could offer more precise control of microstructure attributes such as feature length-scales and volume fraction; thus, providing a robust processing technique for BUECs with a wide range of mechanical properties.

Summary and Conclusions
The variation of morphology, constituent phases and length-scales of an arc-melted Al 81 Cu 13 Si 6 alloy upon laser remelting have been studied. Three distinct eutectic structures were obtained based on location within the laser trace: Al-Al 2 Cu-Si ternary eutectic, Al-Al 2 Cu binary eutectic and Al-Si binary eutectic. The extent of departure from equilibrium solidification (quantified by cooling rate) determined the solidification microstructure, which was either a (i) ternary eutectic, or (ii) binary (Al-Al 2 Cu)/ternary (Al-Al 2 Cu-Si) bimodal eutectic, or (iii) 'dual-binary (Al-Al 2 Cu)/Al-Si eutectic' . The formation of bimodal eutectics was favored at cooling rates under ×°4 10 C/s 4 . The flow strength of bimodal eutectics was effectively varied between 500 to 900 MPa, a result of differing fine eutectic volume fractions within the laser trace. The control of volume fraction and constituent phases will provide greater flexibility in processing materials with diverse mechanical properties.

Materials and Methods
Sample preparation and laser surface remelting. A eutectic Al 81 Cu 13 Si 6 (at. %) ternary alloy ingot was fabricated using an arc melter (Materials Preparation Center, Ames Laboratory, Iowa State University). Argon gas environment was employed to protect the sample from oxidation during arc melting. Specimens with a thickness of 20mm × 10mm × 2 mm were cut from the as-cast ingot.
After grinding and polishing, the specimens were laser surface remelted using solid state disk laser (TRUMPF Laser HLD 4002) at a wavelength of 1030 nm. Previous investigations on laser surface remelting of Al alloys primarily used CO 2 laser (λ = 10.6 μm) 40,52 . However, the absorption coefficient of Al at this wavelength is low (2-3%) 53 . Lower wavelength laser irradiation resulted in a more efficient remelting process (absorption coefficient of 10%), permitting the use of lower powers and higher scan speeds. The absorption was further improved by coating the sample with graphite (Bonderite L-GP G aerosolized graphite lubricant) prior to remelting. The laser power (P) was varied between 500 and 1000 W. The laser spot diameter (S s ) and scan speed (S y ) ranged from 0.8 to 1.2 mm and 2.54 to 101.4 mm/s respectively. Argon shielding gas (flow rate of 9.4 L/min) was used during the laser melting process to prevent oxidation. A two-color pyrometer (IMPAC ISR 12-LO/GS) was used to measure the temperature-time history at the top of the melt pool. The temperature history was obtained at approximately the middle of laser scan path, to avoid anomalous free-edge heat transfer effects. Materials characterization and hardness testing. The eutectic microstructures of the as-cast and laser treated Al-Cu-Si specimens were characterized using an FEI Helios 650 scanning electron microscope (SEM) in the secondary electron mode. The TEM samples were prepared by mechanical cutting, grinding, polishing followed by focussed ion beam milling. High angle annular dark field (HAADF) images and corresponding bright field images were captured using a JEOL 3100 R05 TEM with an operating voltage of 300 kV. Indentation experiments were conducted at room temperature on a hardness tester. A Vickers indenter (Clark CM-800 microhardness tester) was employed to perform the indentation experiments in the load controlled mode, with the peak load ranging from 250 mN to 2 N. Image processing. The coarse and fine eutectic phases were mapped to white and black pixels respectively using imadjust and medfilt2 MATLAB functions. The image was further processed to fill gaps using erosion and dilation operations with a disk structuring element. The imcomplement and imfill functions were used to fill any remaining holes. The volume fraction of the fine eutectic phase was evaluated as the number of black pixels divided by the total number of pixels.