In situ particulate-reinforced aluminum matrix composites (PRAMCs) display advantages such as low cost, light density, high specific intensity and stiffness, corrosion resistance and excellent anti-wear properties. Therefore, they have received special attention in aeronautics, astronautics, automobiles and other technical applications [1–5]. Nevertheless, drawbacks still exist, such as the lower tensile ductility of PRAMCs  synthesized by direct melt reaction since the mutual exclusivity between metal material ductility and strength is an obstacle to their manufacture and application . To fabricate strong and ductile PRAMCs, researchers use fabrication techniques such as a direct magnetochemistry melt reaction or plastic deformation after casting to gain a finer particle distribution . Apart from improvements in particle distribution, of equal importance is the fact that it is generally known that grain refinement of the matrix is a practical way to improve toughness and strength without losing ductility. Laser surface melting is a surface treatment process and a practical method to produce an ultrarefined grain surface with applications in automotive, aerospace and high-speed railway systems . Flow vortices generated in a molten pool during processing may change the particle distribution. Laser surface treatment is a more cost-effective method than melt spinning, which is a rapid solidification technique to produce alloys that provide ultrafine and homogeneous microstructures in some cases. Researchers  have indicated that the aluminum alloy surface can be modified and wear and corrosion properties can be improved by laser melting. Therefore, laser melting has been applied to aluminum alloys and also needs to be investigated for aluminum matrix composites.
Laser-melting technology uses a high-power laser beam focused onto a metal surface to create a moving molten pool. As the laser beam moves, the enlarged molten pool cools rapidly and solidifies (up to 106 K/s). Although laser melting raises the service life of aluminum alloy compounds considerably, some problems still exist with the technology. Li et al. and Kumar Ghosh et al. [10,11] indicate that defects such as pores and thermal cracks tend to occur under room temperature processing conditions. Other problems also arise because of different physical properties between the substrate and the alloyed zone . The research of Hamatani et al.  indicates that the heat-affected zone (HAZ) will be softened easily by re-crystallization, tempering and/or dislocation recovering after melting. It has also been reported that cracks have been observed and grew through the HAZ . The width of the heat-affected zone is based on the extent of heat conduction from the melting zone to the substrate. It is generally accepted that laser welding has a higher welding speed and creates a narrower HAZ width compared with other welding methods, because of its high thermal gradients and cooling and solidification rates. Therefore, if heat transfer during laser melting can be controlled or limited, rapid thermal energy conduction from the HAZ and the HAZ width could be treated smaller.
Liquid nitrogen, which has a boiling temperature of 80 K, can contribute extremely high cooling rates compared with air during solidification. Initial alloy temperatures before melting can also be reduced significantly. Vaporized nitrogen may also optimize welding conditions to achieve a reduced HAZ width and a stable surface with no defects . Few reported attempts exist on laser melting in situ aluminum matrix composites under liquid nitrogen conditions.
Therefore, in this work, we investigate the effects of laser melting on a ZrB2p/6061Al composite in air and under liquid nitrogen cooling conditions. The in situ ZrB2p/6061Al composite was produced via a melt reaction, and the microstructure and distribution of ZrB2 particulates before and after laser melting are described by X-ray diffractometry (XRD), light optical microscopy (LOM), field emission-scanning electron microscopy (FESEM), energy-dispersive spectroscopy (EDS) and high-resolution transmission electron microscopy (HRTEM). The relationship between processing parameters and the resulting solidification microstructures and ZrB2 particulate mechanisms during processing is discussed.
Materials and experimental methods
Material and sample preparation
Raw materials are commercial AA6061, which was used as a base alloy in this study (shown in Table 1), inorganic zirconium potassium fluoride salt (K2ZrF6) and potassium fluoborate (KBF4) powder. To ensure dehydration, zirconium potassium fluoride and potassium fluoborate were preheated in an electric oven at 473 K for 4 h and then cooled, ground and screened. The dried K2ZrF6 and KBF4 were mixed with a mass ratio of 52:48 and kept in aluminum foils. Simultaneously, the AA6061 alloy ingots were melted in a resistance furnace and kept at 1,123 K. The aluminum foil-containing mixtures were added into the melt that was agitated by mechanical stirring. Continuous melt temperature detection was by a thermocouple during the melt reaction. When the reaction finished after approximately 10 min, the melt was refined and degassed with hexachloroethane (C2C16). Finally, the melt was cast into a copper module at 993 K.
Specimens were cut into 40 × 40 × 20 × mm3 cubes by wire cutting. The specimen surfaces were ground with SiC emery paper from 180# to 800# and cleaned with alcohol to ensure that their surfaces were flat and uniform.
Laser surface melting
The specimen was treated using a neodymium-yttrium aluminum garnet (Nd-YAG) laser system (GSI, Model JK2003SM) (Figure 1(a)) at the Tianjin Key Laboratory of Advanced Joining Technology (TKL-AJT), Tianjin, China. The system has a maximum output power of 2,000 W, a focal length of 160 mm and a focused beam diameter of 0.6 mm. Figure 1 and Table 2 provide the experimental configuration applied in this work. The bottom and body of the specimen were submerged in liquid nitrogen whereas the upper specimen surface was exposed to air for laser beam treatment. During laser melting, shielding nitrogen gas from air and vaporized liquid nitrogen accelerates nitrogen solidification in the melted composite. Thus, argon gas was used as shielding gas to decrease porosity by preventing nitrogen from entering the laser-melting zone. The laser beam diameter was 1.0 mm, the scanning power density varied from 200 W/mm2 to 400 W/mm2, and the scan velocity was 4 mm/s. To prevent overfiring and grain growth of the melted zone, laser beam movement should avoid overlap in any area of the specimen (Figure 1(b)). For comparison, a specimen exposed at room temperature was also treated by laser surface melting.
Metallographic and substructural analysis
After surface melting, the specimen was cut vertically in the laser scanning direction. Its cross-section and melted surface were ground and polished by standard metallographic techniques. After etching with 0.5 % HF, specimens were cleaned using water and alcohol. The microstructure of the as-received and laser-melted specimen were studied by LOM (Olympus, Model GX51) and FESEM (Hitachi, Model S4800) and using attached EDS (Oxford, Model X-Max) to determine the specimen composition. To obtain detailed morphology of the laser-melted zone, microsections approximately 2 mm thick were wire cut from the specimen and polished to a thickness slightly less than 40 μm, and the ion beam thinner was used to prepare samples for HRTEM (JEOL, JEM-2100F). Phases in the as-received and treated specimen were determined by XRD at 40 kV and 20 mA and a scanning range from 20° to 80° at a speed of 8° per minute. In terms of mechanical properties, Vickers hardness tests were carried out using a Reichert microhardness tester with 100 g load on polished cross-sections.
Experimental results and discussion
Figure 2 shows the XRD pattern of an as-prepared ZrB2p/6061 in situ aluminum matrix composite fabricated from 6061Al-K2ZrF6-KBF4 by direct melt reaction. The diffraction peaks confirm that the composite consists of α-Al, Mg2Si and ZrB2 phases. Owing to the excess KBF4 power used during the melt reaction, no diffraction peaks of other phases relating to this melt reaction are visible [14,15]: (1) (2)
The AlB2 in eq. (1) and Al3Zr in eq. (2) interact and form ZrB2: (3)Thus, possible reactions can be illustrated as: (4)Part of the K3AlF6 gas ran off from the melt after reaction, while the rest was degassed and stripped with oily KAlF4 by hexachloroethane.
Figure 3 shows the distributions and morphologies of the ZrB2 particles in 6061Al matrix composites. As shown in Figure 3(a), the theoretical particle volume fraction in the composite is 5 %, whereas the actual volume fraction is slightly less than the calculated one because of insufficient kinetic processes of the in situ reaction . With the 5 % ZrB2 volume fraction, particles agglomerate to form clusters to a certain extent. From Figure 3 (c and d), it can be seen that the morphology of ZrB2 particles is elliptical or nearly hexagonal and of hundreds of nanometers in size. Its EDS pattern demonstrates the existence of Al, Zr and B in spectrum 1 (Figure 3(d)). The detailed substructure of a cross-section of the composites was determined by HRTEM.
Penetration depth after laser melting
With an increase in laser power, the top surface became rougher and a wavy topography in the form of ripples resulted. The effects of laser power on the depth of the melted zone are shown in Figure 4. The penetration depth increases accordingly with laser power. When the laser power was less than 199 W/mm2, no melted surface existed on the composites. Only when the power of the laser was higher than 243 W/mm2 and 199 W/mm2, the specimen under liquid nitrogen cooling and air started melting. For the specimen cooled by liquid nitrogen, the requirement for a higher laser power is attributed to the ultra-low temperature and better specimen thermal conductivity.
Morphology of laser-melted ZrB2p/6061Al composites
Figure 5 shows six optical metallographs of an ingot of laser-melted specimen under room temperature and liquid nitrogen conditions, respectively. The melted specimen shows a finer laser-melted zone than the substrate without laser melting (Figure 5(g)). Furthermore, the specimen under liquid nitrogen cooling shows ultrafine grains compared with other specimens. This resulted from the extra undercooling of specimens under liquid nitrogen compared with room temperature. For specimens treated in air and liquid nitrogen conditions, the laser-melted matrix shows a finer grain near the top surface than the edge of the bottom and middle. The cellular dendrite size in a certain metal and laser parameters depend on the degree of undercooling and the temperature gradient. As a relevant estimation, based on the research of Kurz , the cellular spacing under an ultrahigh-temperature gradient can be expressed as: (5)
where ∆T0 is the degree of undercooling, DL is the liquid diffusion coefficient, Γ is the Gibbs–Thomson coefficient, k0 is the solute distribution coefficient, GL is the temperature gradient and v is the dendrite growth velocity. According to eq. (5), the higher thermal gradient is likely to generate a narrower cellular spacing. The temperature gradient was calculated in  and result illustrates that the highest temperature gradient exists at the top of the molten pool, whereas the lowest exists at the bottom of the molten pool. In addition, findings from this work are consistent with results from other experiments, which demonstrates that the average cellular spacing decreases from the bottom to the top of the molten pool. Based on this phenomenon, it can also be concluded that the dynamics of the fluid flow leads to the flotation of solute particles, which leads to heterogeneous nucleation.
Figure 6 shows SEM micrographs of the ZrB2 particle distribution of specimens before melting and that melted under liquid nitrogen cooling. A relatively better distribution of ZrB2 particles and small particle clusters is observed after laser melting. During processing, thermocapillary flow induced by surface tension variations largely determines the movement behavior of particles in the melt. The fluid flow intensity increases with increases in melting power, which results in a rotational flow vortex pattern  (Figure 7). Since the centrifugal forces act on the rotating flow, ZrB2 particle clusters disperse. With the precipitation of an α-Al phase, ZrB2 particles are pushed by the boundary of the α-Al grain because of the insufficient wettability between matrix and particles. The particles distributed near the grain boundary after solidification of the melted zone and narrower cellular spacing largely contribute to the finer particle distribution.
Mechanical behavior of ZrB2 particles in processing
Until recently, researchers found that cracks are easily generated and are visible in PRAMC during laser cladding. Reference  indicates that for the Nd-YAG laser, the power intensity in the laser beam center is relatively higher than that at the periphery. Similarly, the laser power is largely absorbed at the material surface of the material, whereas the bottom absorbs less. Therefore, a higher temperature is generated in the top center of the molten pool. Consequently, this induces a thermal stress because of the non-uniform nature of the temperature distribution in the molten pool. Zhou et al.  investigated the stress behavior during rapid heating and solidification of metal matrix composites and presumed that contraction stress was generated during this process. The contraction and thermal stress formed the residual stress that induced a crack when the stress was higher than the strength of the material . However, in this work, no obvious crack is detected since a different composite fabrication method is used. The in situ metal matrix composite fabricated in this work exhibits more advantages than ex situ composites (where particles are added into the melt as powders), such as clean interfaces, strong interfacial bonding and a better particle distribution, so there is less possibility of the generation of surface defects such as cracking . What is more, laser cladding usually fabricates composites that contain a higher volume fraction of particles. The possibility of a clustering effect, which resists the flow of molten aluminum matrix, increases with an increase in the amount of particles. A comparatively lower volume fraction of particles (5 %) also contributes to a decrease in cracking tendency in this work.
Figure 8 (a) and (b) show typical bright-field TEM images of laser-melted composite cooled by air and liquid nitrogen, respectively, and an elliptical-shaped ZrB2 phase. Dislocations forming around the ZrB2 particles can be detected in the liquid nitrogen-cooled specimen, whereas in the specimen treated at room temperature, the interface of particles and the matrix is almost complete. This most likely occurs because of the thermal mismatch stress between the ZrB2 particles and the AA6061 matrix during high-speed liquid nitrogen cooling. ZrB2 particles also contribute significantly to the increase in tensile strength during processing as a dislocation strengthening effect. In their research on thermal expansion of metal matrix composites, Vaidya et al.  discovered that when the temperature changes during processing, the radial σr and tangential σθ stresses generated at the matrix–particle interface (Figure 9) can be calculated from the equations below: (6) (7)
where is the interfacial stress and can be calculated using the following equation: (8)r is the distance between the force point in the matrix and the center of the particle, a is the particle radius (Figure 9), Vp is the volume fraction of the reinforcement, Δα is the difference in coefficient of thermal expansion (CTE) between the matrix and particles (CTEAA6061 = 24.5×10–6/K, = 6.8×10–6/K) and ΔT is the temperature change during laser melting. Reference  indicates that when the Al matrix temperature is above 573 K, |σr–σθ| > σy (σy is the yield strength of the matrix) and that when cooled, plastic deformation may happen at the matrix–particle interface because of the difference in CTE between matrix and particle. This leads to the generation of a high density of dislocations at the matrix–particle interface in the composite because of the mismatched strains between them. The high density of dislocations hinders dislocation movement at the matrix–particle interface and increases the microhardness and tensile strength of the composite after laser melting.
Results from Vickers hardness tests of the treated specimen are shown in Figure 10. The specimen under liquid nitrogen reaches a higher value of hardness than that under air. The microhardness of the melted layer generated in liquid nitrogen and air is 60–75 HV and 56–65 HV, respectively. According to the Hall–Petch relationship between the yield stress and grain size, the hardness of the liquid nitrogen-cooled specimen increases relative to other specimens and the substrate. The high density of dislocations generated around the ZrB2 particles during laser processing also contributes to the increase in hardness. Because of the ultrahigh cooling speed of the specimen in liquid nitrogen, the effect of solid–solution strengthening is more obvious than in the air-cooled specimen.
The ZrB2p/6061Al composites were fabricated from the Al-K2ZrF6-KBF4 system by a direct melt reaction. FESEM analysis indicates that the morphology of the nano-scaled ZrB2 particles is rectangular and nearly hexagonal shaped and the particles exist as clusters in certain degree. The interface between the ZrB2 particles and the aluminum matrix is continuous with no visible cracks. After laser melting, the size of the cellular dendrite of the specimen under liquid nitrogen cooling is narrowest compared with any other specimen since it has the highest degree of undercooling. Similarly, the laser-melted matrix shows a finer grain near the top surface than the edge of the bottom and middle because of the higher temperature gradient. SEM analysis shows that the nano-ZrB2 particle distribution changed from clusters to well-distributed particles along the edge of the temperature gradient after laser melting. TEM analysis shows a generation of dislocations along the particle–matrix interface after laser melting, which induces a thermal mismatch stress and leads to an increase in hardness of the melted zone of 50.4 % from 60 HV to 75.2 HV along with well-distributed ZrB2 particles and refined grains.
 D.-B. Chen, Y.-T. Zhao, H.-Y. Zhu, M. Zheng and G. Chen, Trans. Nonferrous Met. Soc. China, 22 (1) (2012) 36–41. Google Scholar
 I. Dinaharan and N. Murugan, Trans. Nonferrous Met. Soc. China, 22 (4) (2102) 810–818.Google Scholar
 G.-R. Li, H.-M. Wang, Y.-T. Zhao, D.-B. Chen and G. Chen, Trans. Nonferrous Met. Soc. China, 20 (4) (2010) 577–583.Google Scholar
 J. Hu, L.C. Kong and G. Liu, Mater. Sci. Eng., A, 486 (1–2) (2008) 80–84.Google Scholar
 L.-H. Zhong, Y.-T. Zhao, S.-L. Zhang, G. Chen, S. Chen and Y.-H. Liu, Trans. Nonferrous Met. Soc. China, 23 (9) (2013) 2502–2508. Google Scholar
 M. Song and D. Xiao, Mater. Sci. Eng., A, 474 (1–2) (2008) 371–375. Google Scholar
 X. Kai, Z. Li, G. Fan, Q. Guo, Z. Tan, W. Zhang, Y. Su, W. Lu, W.-J. Moon and D. Zhang, Scr. Mater., 68 (8) (2013) 555–558. Google Scholar
 Z. Liu, P.H. Chong, A.N. Butt, P. Skeldon and G.E. Thompson, Appl. Surf. Sci., 247 (1–4) (2005) 294–299. Google Scholar
 T.T. Wong, G.Y. Liang and C.Y. Tang, J. Mater. Process Tech., 66 (1–3) (1997) 172–178. Google Scholar
 Y. Li and D. Gu, Mater. Design, 63 (2014) 856–867. Google Scholar
 S. Kumar Ghosh and P. Saha, Mater. Design, 32 (1) (2011) 139–145. Google Scholar
 H. Hamatani, Y. Miyazaki, T. Otani and S. Ohkita, Mater. Sci. Eng., A, 426 (1–2) (2006) 21–30. Google Scholar
 A. Trudel, M. Sabourin, M. Lévesque and M. Brochu, Int. J. Fatigue, 66 (2014) 39–46. Google Scholar
 G.-R. Li, H.-M. Wang, Y.-T. Zhao, G. Chen, Q.-X. Dai and X.-N. Cheng, J. Univ. Sci. Technol. Beijing, 31 (5) (2009) 591–596. Google Scholar
 K. Tian, Y. Zhao, L. Jiao, S. Zhang, Z. Zhang and X. Wu, J. Alloy Compd., 594 (5) (2014) 1–6. Google Scholar
 H.-J. Yang, Y.-T. Zhao, G. Chen, S.-L. Zhang and D.-B. Chen, Trans. Nonferrous Met. Soc. China, 22 (3) (2012) 571–576. Google Scholar
 W. Kurz and D.J. Fisher, Acta. Mater., 29 (1) (1981) 11–20. Google Scholar
 L.-X. Ying, L.-Q. Wang, G.-C. Chen, L. Gu, D.-Z. Zheng and W.-Z. Li, Heat Treat. Metals, 7 (2004) 24–28. Google Scholar
 D. Donghua and G. Dongdong, Mater. Design, 55 (2014) 482–491. Google Scholar
 S. Zhou, X. Zeng, Q. Hu and Y. Huang, Appl. Surf. Sci., 255 (5) (2008) 1646–1653. Google Scholar
 P.B. Kadolkar, T.R. Watkins, J. Th, M. De Hosson, B.J. Kooi and N.B. Dahotre, Acta. Mater., 55 (4) (2007) 1203–1214. Google Scholar
 R.U. Vaidya and K.K. Chawla, Compos. Sci. Technol., 50 (1) (1994) 13–22. Google Scholar
About the article
Published Online: 2016-12-29
Published in Print: 2017-01-01
Funding: This work was supported by the National Natural Science Foundation of China (Grant Nos. 51405334 and 51275342), State Key Laboratory of Advanced Welding & Joining, Harbin Institute of Technology (AWJ-Z14-03).