Direct Laser Deposition of a Single-Crystal Ni 3 Al-Based IC221W Alloy

Transcription

1 Direct Laser Deposition of a Single-Crystal Ni 3 Al-Based IC221W Alloy WEIPING LIU and J.N. DuPONT Single-crystal (SX) nickel aluminide alloys have potential for structural applications where hightemperature strength and oxidation resistance are required. In this work, SX deposits of the Ni 3 Albased IC221W alloy were produced on a SX Ni-base superalloy substrate by means of the laser-engineered net shaping (LENS) process. The microstructure of the deposits was characterized. The effects of processing parameters on the SX solidification in the melt pool and on the fabricability by LENS were investigated. A simple relationship between the ratio of the temperature gradient to the growth velocity and the processing parameters was derived, which can be used qualitatively to guide the proper selection of processing conditions to maintain the columnar dendritic growth during the laser deposition. On the basis of analyses and experiments, the effects of processing parameters on the susceptibility to stray grain formation and solidification cracking are discussed. I. INTRODUCTION NI 3 AL-BASED alloys are attractive materials for hightemperature structural applications, for example, as turbine elements or other heat- and oxidation- or wear-resistant components. [1 4] This is due to their high strength retention at elevated temperatures, combined with relatively low density, good oxidation, and corrosion resistance. For example, the elevated temperature strength and creep resistance of recently developed Ni 3 Al alloys are shown to be superior to most commercial superalloys. [2] The high-temperature corrosion and oxidation resistance of the aluminide alloys also makes them good candidates for applications as high-temperature coatings. Due to the absence of grain boundaries and enhanced mechanical properties, single-crystal (SX) nickel aluminide alloys have significant potential for gas turbine applications where high-temperature strength and oxidation resistance are required. [5] Laser engineered net shaping (LENS) is a solid free-form fabrication process based on laser cladding, which involves laser processing fine metallic powders into fully dense, threedimensional shapes directly from a computer-aided design (CAD) model. A variety of metallic and composite materials have been deposited by LENS processing. [6 12] The LENS process is able to fabricate complex prototypes in near-net shape, leading to significant time and machining cost savings. This process also has potential for precision repair, fast tooling, and small-lot production. [6] As a novel process for repair applications, the LENS process has several potential advantages over conventional processes. It exposes the part to far less heat than conventional arc welding techniques, and the heat-affected zone is much smaller, thus significantly reducing any structural and mechanical degradation to the part while repairing a specific area. The LENS process also WEIPING LIU, formerly Research Scientist, Department of Materials Science and Engineering, Lehigh University, is Research Metallurgist, Indium Corporation of America, Clinton, NY Contact wliu@indium.com J.N. DuPONT, FASM Associate Professor and Director, Joining and Laser Processing Laboratory, is with the Department of Materials Science and Engineering, Lehigh University, Bethlehem, PA Manuscript submitted April 26, can offer exceptional material properties and interface characteristics to the repaired part for property enhancements. For example, refined microstructures in the repaired area can be produced due to the rapid solidification conditions associated with the laser processing. Functionally graded materials and coatings can be fabricated to improve the interface characteristics or the surface properties of the repaired part. [10] The melt-pool solidification conditions during LENS processing generally result in a dendritic microstructure with columnar or equiaxed growth morphology. [7,11] The columnar dendritic growth is caused by epitaxial growth from the partially remelted grains in the previously deposited layer or substrate, which takes place without nucleation. Recent studies by Gaumann et al. [13,14] using a laser metal forming technique indicated that it is possible to deposit a SX Nibase superalloy clad by epitaxial growth onto a SX Ni-base superalloy substrate. Single crystalline solidification could be achieved in the melt pool of a nickel-base alloy by stabilizing the epitaxial, columnar dendritic growth and thereby avoiding nucleation and growth of equiaxed grains when a SX nickel-base substrate was used. However, so far no attempt has been made to deposit SX nickel aluminide alloys. Ni 3 Al has the same fcc structure as Ni, and both have a similar value of lattice parameter. The objective of this research is to investigate the use of LENS for maintaining the SX structure of Ni 3 Al-based alloy deposits on a SX Ni-base superalloy and identify the effects of processing parameters on the SX fabricability. II. EXPERIMENTAL PROCEDURE A Ni 3 Al-based alloy powder, which was designated as IC221W and has a nominal composition of Ni-8.0Al-7.7Cr- 3.0Zr-0.003B (in wt pct), was used in this investigation. The IC221W alloy was developed by the Oak Ridge National Laboratory (ORNL, Oak Ridge, TN) to provide a nickel aluminide alloy with enhanced solidification cracking resistance and thus improved weldability. [4] The alloy powder was provided by ORNL and had a mesh size of 100/ 325 (particle sizes between 45 and 150 m). A typical commercial METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 36A, DECEMBER

2 SX nickel-base superalloy, the CMSX-4 alloy, was chosen as the SX substrate alloy. The nominal composition of this alloy is Ni-9Co-6.5Cr-5.6Al-1Ti-6W-6.5Ta-3Re-0.6Mo-0.1Hf (in wt pct). The as-grown crystal obtained from Concorde Castings (Eastlake, OH) was cut into specimens of approximately 9 (width) 4.5 (thickness) 20 (length) mm, with the 001 orientation normal to the surfaces of the specimen. The LENS processing was performed along the [100] crystallographic direction on the (001) substrate surface. The orientation of the substrate was determined by the Laue backreflection technique. Single- and multitrack deposits and multilayer, wall-shaped deposits (up to 30 layers) were produced for different purposes in the study. For comparison and convenience of microstructure analysis, a pure nickel plate (Ni 200) was also used as a polycrystalline (PX) substrate material to make PX Ni 3 Al-based alloy deposits in this investigation. The PX IC221W deposit samples had a size of mm 6 layers. Successive layers were deposited with the bead lines of two adjacent layers at an angle of 90 deg. The substrate surfaces to be laser processed were ground with 600-grit SiC paper and cleaned in methanol before laser surface melting. An Optomec (Albuquerque, NM) LENS 750 system was used to make the deposits for this investigation. The LENS machine consists of a continuous wave Nd:YAG laser that is focused with a plano-convex focusing lens, a four-nozzle coaxial powder feed system, a controlled-environment glove-box, and a motion control system. The Nd:YAG laser has a 0.5 to 1-mm-diameter circular beam at the focal zone with the Gaussian intensity distribution and a maximum output power of 750 W. The powder delivering nozzles are designed and arranged in such a way that the powder streams converge at the focal point of the laser beam. Figure 1 schematically shows the LENS process as a means of producing SX deposits on an SX substrate. To make a LENS deposit, a CAD model of a three-dimensional component is first sliced into a series of layers of finite thickness using computer software. Each of these layers is then translated into a series of line patterns in order to deposit the layers. The laser beam is used as a heat source to create a molten pool on the substrate, and the powder is injected into the melt pool by an inert gas flowing through the powder-feed system. The first layer of the component is bonded to the substrate. The substrate together with the component under fabrication is moved along the line patterns in the plane of the current layer with the motion control system. Fig. 1 Schematic representation of the LENS process as a means of producing SX deposits. After completing a layer, the laser focal point and powderdelivering nozzles are incremented upward in the height direction in an amount of the layer thickness. A new layer is subsequently deposited onto the previous layer until the component is fully constructed in the layer-by-layer fashion. The LENS processing was carried out in an argon atmosphere in the glove-box to prevent oxidation. The powder can be recovered and reused after sieving. The layer thickness, hatch spacing, and stand-off distance were set at 0.254, 0.381, and mm, respectively, in the experiments. The oxygen level in the glove-box was kept below 30 ppm during processing. Light optical microscopy (LOM), scanning electron microscopy (SEM) coupled with an X-ray energy-dispersive spectrometer, and X-ray diffraction (XRD) were used for microstructure analysis in the study. Samples for LOM were mounted and polished using standard metallographic techniques and etched with a solution containing CuCl 2 (5 g), HCl (100 ml), and H 2 O (100 ml). X-ray diffraction using the Cu K radiation ( A) was conducted on the PX sample surface perpendicular to the build direction. Calibration for XRD was made using a standard tungsten powder sample. Electron backscattering diffraction (EBSD) mapping analyses were performed on selected samples to identify more accurately the degree of stray grain formation in the SX deposits. III. RESULTS AND DISCUSSION A. Microstructure Analysis of a PX IC221W Deposit Figure 2 shows the microstructure of etched and as-polished cross sections of a PX IC221W alloy deposit on the Ni 200 substrate processed with a laser power of 415 W, a 4.2 mm/s traverse speed, and a g/s powder feed rate. The polycrystalline nature can be clearly seen from the various dendritic growth directions (Figure 2(a)), although the growth direction parallel to the build direction the arrow direction indicated in Figure 2(a) was prevalent in the microstructure. The microstructure of as-deposited IC221W was characterized by dendrites with an interdendritic, eutectic-type constituent (Figures 2(b) through (d)). According to the Ni-Al-Cr phase diagram, [15] the primary solidification phase of the IC221W alloy is a disordered fcc phase (Ni- Al-Cr solid solution), most of which is transformed to an L1 2 -type ordered phase (Ni 3 Al) during subsequent cooling. [4] The SEM backscattered electron micrographs of the as-polished sample show the distribution (Figure 2(c)) and morphology (Figure 2(d)) of the interdendritic, eutectic constituent. The EDS analyses (Figure 3) indicated a clear Zr peak in the interdendritic, eutectic component and very low Zr in the dendrites, suggesting that most of Zr was present in the eutectic. Based on the work by Santella et al., [4] the eutectic-type constituent consists of and Ni 5 Zr phases. Previous work [4] has shown that the increase in the amount of the eutectic component in the IC221W alloy, as opposed to the IC221M (Ni-8.0Al-7.7Cr-1.7Zr-0.03B, wt pct), improves the solidification cracking resistance of the alloy. As shown in Figure 4, the result of XRD conducted on the surface perpendicular to the build direction confirms that the PX IC221W deposit consists mainly of Ni 3 Al phase, as evidenced by the presence of a clear (100) superlattice diffraction peak for the ordered Ni 3 Al intermetallic compound. The reason that the diffractive peak from {200} planes in 3398 VOLUME 36A, DECEMBER 2005 METALLURGICAL AND MATERIALS TRANSACTIONS A

3 Fig. 2 Microstructure of a polycrystalline IC221W deposit: (a) and (b) LOM photomicrographs (etched sample); and (c) and (d) SEM backscattered electron micrographs (as-polished sample). Fig. 3 EDS patterns of the (a) dendrite and (b) eutectic-type constituent in the PX IC221W deposit shown in Fig. 2. METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 36A, DECEMBER

4 the XRD pattern appeared as the strongest one in place of the peak from {111} planes is that the columnar dendritic grains are preferentially oriented with their 100 directions being approximately parallel to the build direction (Figure 2(a)). Due to an insufficient quantity of the Ni 5 Zr phase in the sample, this phase did not appear in the XRD pattern. B. Fabricability of SX IC221W Deposits For the purpose of investigating the fabricability of SX deposits, single track deposits were made along the [100] crystallographic direction on the (001) surface of the SX superalloy substrate. Figure 5 shows the microstructure of Fig. 4 XRD pattern of the polycrystalline IC221W deposit. the central part in the transverse section (perpendicular to the beam traveling direction) of deposits made at a laser power (P) of 450 W and traverse speeds of 5, 10, and 15 mm/s (with the same powder-feed rate of g/s). It is noted that the area of the transverse section of the clad decreases with the increase in traverse speed. The following relation can be used to explain this observation. [21] The area (S) of the clad in the transverse section equals the net deposition rate divided by the beam travel speed (V b ), i.e., S j # R p r # V b where is the deposition efficiency (the fraction of the powder actually deposited into a clad over the total powder fed from the nozzles), R p is the powder mass feed rate (g/s), and is the density of the powder (g/mm 3 ). The deposition efficiency ( ) tends to increase with increasing laser heat input per unit length of clad (P/V b ). [21] Therefore, according to Eq. [1], the area of clad increases with decreasing travel speed when a fixed powder feed rate is used. The typical value of the deposition efficiency is 0.1 to 0.2; the thickness and width of the deposited layers are in the range of 0.2 to 0.4 mm and 0.8 to 1.5 mm, respectively. As shown in Figure 5, the transverse-section microstructure in the deposits is composed of three regions: (A) epitaxial dendritic growth in the [001] direction, (B) epitaxial dendritic growth in the [100] direction (the beam traveling direction), and (C) the region of stray grains. It should be pointed out that the SX nature is maintained in both the A and B regions. The only difference in A and B is that the [1] Fig. 5 Transverse-section microstructures of single-track deposits made at a laser power of 450 W and a traverse speed of (a) 5, (b) 10, and (c) 15 mm/s VOLUME 36A, DECEMBER 2005 METALLURGICAL AND MATERIALS TRANSACTIONS A

5 primary growth direction of the dendrites changes from the [001] to [100] direction according to the melt-pool geometry. [18,20] As can be seen in Figure 5, the C region becomes larger as the traverse speed increases. Figure 6 shows the EBSD maps, obtained by orientation image microscopy (OIM), of the selected areas in the deposits shown in Figure 5. The results suggest that the propensity for stray grain formation increases with increasing processing speed under the experimental conditions. This can be explained from the following two aspects. First, recent studies [14,16,22] have shown that the formation of stray grains is related to the extent of constitutional supercooling (CS) ahead of the solidification interface. In these investigations of autogenous (no filler metal used) laser welding and laser surface remelting of SX nickel-base superalloys, the existence of constitutional supercooling was indicated to be a viable mechanism for stray grain formation. Thus, the extent of CS can be used as a measure of the propensity for stray grain formation via nucleation ahead of the solidification front. According to solidification theories, [17] the extent of CS is related to the ratio G/V, where G is the temperature gradient in the liquid at the solidification front and V is the solidification growth velocity at the solid/liquid interface. In the conventional CS theory, V is taken as the velocity of the initially planar interface. In the present case, V is taken as the dendrite-tip growth velocity. The smaller the G/V ratio, the larger the extent of CS in the liquid ahead of the solidification front becomes. It should be pointed out that, according to solidification theories, the extent of CS also is related to the liquid composition in the melt pool. In the present study, the liquid composition in the melt pool is a function of the processing parameters due to the effect of dilution of the substrate (especially for the first several layers). However, it was found that the composition had only a minor effect under the present experimental conditions due to a relatively small difference in the degree of dilution for the variation range of the processing parameters. Therefore, the following discussion is focused on the effect of the G/V ratio on the extent of CS. Fig. 6 EBSD maps of the selected areas in the deposits shown in Figs. 5(a) through (c). The color scale indicates the misorientation angle of the grain from the reference orientation (at zero degree), i.e., (001)/[100]. METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 36A, DECEMBER

6 The dendrite-tip growth velocity (V hkl ) along the growth direction [hkl] in the melt-pool solidification is related to the laser beam travel speed (V b ) by the following equation: [18] cos u ` Vhkl ` ` Vb `. cos hkl where is the angle between the normal 1 S n 2 to the solidification front and the travel direction of the beam, and hkl is the angle between the solidification front normal and the dendrite growth direction defined by [hkl] (Figure 7). Under the substrate orientation conditions used in the present study, the maximum velocity (V max ) of epitaxial dendrite growth is equal to the beam travel velocity (V b ), which corresponds to the [100] growth region where dendrites grow epitaxially along the [100] direction [20], i.e., the beam travel direction. As shown in Figure 5, the stray grains also were mostly observed to occur in this growth region. For a qualitative analysis, the quasi-steady temperature distribution in the melt pool can be approximated by the Rosenthal solution to the three-dimensional temperature field induced by a point heat source moving along the positive x direction in a semi-infinite plate, which is given by [19] T T 0 hp # exp c V b 1R x2d 2pkR 2a where T 0 is the preheating temperature of the substrate, the laser absorption coefficient, P the incident laser power, k the thermal conductivity of the material, the thermal diffusivity of the material, and R (x 2 y 2 z 2 ) 0.5. The temperature gradient along the growth direction in the [100] Fig. 7 Schematic representation of the angular relationships between the solidification interface normal (n) and the x-y-z reference system and between the normal (n) and the [hkl] dendrite growing direction (V). [2] [3] growth region is the derivative of T with respect to x, which can be expressed as G 100 0T 0x hp 2pkR c V b 2 2a 1R x2 x R d. exp c V b 1R x2 d 2a From Eq. [4], the ratio G 100 /V 100 in the [100] growth region of the melt-pool solidification interface can be obtained: For simplicity, consider the [100] growth of dendrites located at the x-axis on the solidification interface (i.e., at the trailing point of the melt pool). In this case, y z 0 and R x. Therefore, Eqs. [3] and [5] can be simplified, respectively, as T T 0 hp # 1 [6] 2pk x G 100 hp # 1 [7] V 100 2pkV b x 2 By inserting Eq. [6] into Eq. [7], the following equation is derived: G 100 2pk # 1T T [8] V 100 hpv b where T is the temperature at the solidification front and can be represented by the liquidus temperature of the alloy if the dendrite tip undercooling is neglected. From Eq. [8], it follows that an increase in the beam travel speed (V b ) decreases the G/V ratio, leading to an increased extent of CS. For example, at a laser power of 450 W and a roomtemperature preheat (20 C), when the beam travel speed increases from 5 to 15 mm/s, the G/V ratio at the melt-pool trailing portion decreases from 262 C s mm 2 to 87.3 C s mm 2 ( 0.4, k 0.02 W mm 1 C 1, and T 1390 C are used in these calculations). Therefore, the increased extent of CS associated with the higher processing speeds contributed to the increase in the propensity for stray grain formation observed in Figure 5. In order to further demonstrate the effects of the processing speed and the role of CS in promoting the stray grain formation under the present experimental conditions, laser surface melting experiments of the SX substrate alloy were conducted without any addition of powder particles. Figure 8 shows the transverse-section microstructures of melt tracks made at a laser power of 450 W, a substrate preheat of 200 C, and a beam travel speed of 5, 10, and 15 mm/s, respectively. In Figure 8(a), an essentially stray-grain-free microstructure is observed. As the travel speed reaches 10 mm/s, a significant area fraction of stray grains can be seen in the microstructure. Further increase in the beam travel speed to 15 mm/s leads to an increased area fraction of stray grains. The corresponding G/V ratios at the meltpool trailing portion for the 5, 10, and 15 mm/s travel speeds are calculated to be 198, 99, and 66 C s mm 2, respectively. A few solidification cracks also can be seen, located [4] G 100 hp x ar 2 V 100 2pkR 2a x V b R b V # b exp c 1R x2d 2a [5] 3402 VOLUME 36A, DECEMBER 2005 METALLURGICAL AND MATERIALS TRANSACTIONS A

7 Fig. 8 Transverse-section microstructures of the melt tracks made at a laser power of 450 W and a 200 C substrate preheat and a beam travel speed of (a) 5, (b) 10, and (c) 15 mm/s. mainly along the boundaries of stray grains, as shown in Figures 8 (b) and (c). This is due to the fact that the misoriented stray grains have high-angle grain boundaries that contain a low-melting eutectic liquid during the final stage of solidification. The second aspect for the stray grain formation in the laser deposition processing is related to the incompletely melted powder particles in the melt pool. In the LENS process, the traverse speed determines the laser/materials interaction time t b D b /V b (where D b is the beam diameter acting on the sample). The interaction time between the laser beam and the material decreases with the increase in the traverse speed (when the laser power is constant), thus reducing the temperature of the melt pool and hence the melting of the powder particles. This also can be reflected from the fact that the laser heat input per unit length of clad ( P/V b ) is reduced during processing with a higher traverse speed but the same laser power. The partially melted powder particles in the melt pool can increase the nuclei density and act as heterogeneous nucleation and growth sites for equiaxed grains that form on the surface, thus promoting the formation of stray grains. As shown in Figure 5(c) (marked with arrows), the stray grains at the surface of the clad were caused by the growth from the partially melted powder particles. Similar observations at the top of deposits are shown in Figure 9, more clearly at higher magnifications. Fortunately, the surface stray grains can be remelted by properly choosing the processing parameters when a subsequent layer is deposited onto the previous layer. However, the layer of the stray grains at the top of a finished product has to be removed by machining. It is noted from Figure 5 that the relative thickness of the [001] growth region with respect to the clad height increases with the increasing processing speed. This is related to the change in the melt-pool geometry. As the processing velocity increases, the melt-pool shape becomes more elongated and the slope of the solidification interface in the beam travel direction is decreased, resulting in a relatively larger area with the angle larger than 45 deg where the dendrites grow in the primary [001] direction. [20] In addition, at the deposit/substrate interface shown in Figure 5, there exists a very thin layer of planar growth solidification (white contrast), which has been well established to result from a relatively high temperature gradient and a very low growth velocity at the boundary region. Fig. 9 LOM photomicrographs showing dendritic growth from the partially-melted powder particles (indicated by arrows) at the top of deposits. METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 36A, DECEMBER

8 The processing velocity also influences the solidification cracking susceptibility of the nickel aluminide alloy deposits. As mentioned previously, the IC221W alloy possesses improved solidification cracking resistance with respect to other nickel aluminide alloys currently available. [5] However, as shown in Figure 5(c), cracking occurred in the deposit when a high traverse speed (15 mm/s) was used. As previously pointed out, the cracking seems to be related to the isolated stray grains observed in the SX portion of the deposit. The crack also can be partly attributed to the increased thermal stresses resulting from a higher cooling rate associated with the high processing speed. Therefore, a processing speed preferably less than 10 mm/s should be used for the SX deposition of IC221W alloy when the beam power and powder feed rate are 450 W and g/s, respectively. Furthermore, based on Eq. [8], it appears that a lower laser power is advantageous for obtaining a larger G/V ratio to avoid the stray grain formation in the melt-pool solidification. However, similar to the case with a high traverse speed, a low laser power will decrease the temperature in the melt pool and may enhance the nuclei density at the surface by increasing the number of incompletely melted powder particles. Also, a decrease in laser power while keeping the traverse speed unchanged will increase the cooling rate and consequently the cracking susceptibility, [12] because the cooling rate is inversely proportional to the laser heat input per unit length of clad (P/V b ). Moreover, an adequate laser power is necessary for completely remelting the surface layer containing stray grains in a previous layer and for ensuring the epitaxial SX growth as well. As a general rule, based on Eq. [8] and the preceding analysis, the combination of a relatively low processing velocity and a reasonably low laser power should be advantageous for producing SX deposits free of stray grains and cracks. Finally, according to Eq. [8], any preheat of the substrate will lead to a decreased G/V ratio, thus promoting the stray grain formation. Therefore, preheating of the substrate for the purpose of reducing the cracking susceptibility should be avoided in order to fabricate SX alloy deposits. On the basis of the previous experiments and analyses, SX deposits consisting of multilayer, overlapped tracks and of wall-shaped, multilayer single tracks were made. Figure 10 shows the transverse-section microstructures of two-layer deposits processed with a laser power of 420 W, a 5 mm/s traverse speed, and a powder feed rate of g/s. In Figure 10(a), the deposit consisted of two layers of single tracks. It can be seen that significant remelting of the underlying layer occurred during the SX deposition. Both the B and C regions in the first layer were remelted during the deposition of the second layer, leading to a perfect connection of the [001] primary dendrites at the layer boundary, indicated by arrows in Figure 10(a). Shown in Figure 10(b) is the microstructure of the overlapping area in a two-layer deposit consisting of multiple tracks. As can be seen, at the overlapping area, indicated by the arrow in Figure 10(b), some dendrites grow along the primary [010] direction due to the thermal conduction into the previously deposited adjacent track. As mentioned previously, the change in the primary growth direction among the 001 directions does not change the SX nature. Figure 11 shows the microstructure of a wall-shaped, 12-layer deposit processed with a 4 mm/s traverse speed, a starting laser power of 400 W, and a powder-feed rate of 0.02 g/s. During the production of multilayer deposits, especially in the case of wall-shaped deposits, the temperature gradients in the melt pool decrease as the number of layers increases, due to the change of the thermal conduction conditions in the deposit. Therefore, the laser power needs to be adjusted (decreased) accordingly to keep the G/V ratio above the critical value for maintaining the columnar dendritic Fig. 10 (a) Transverse-section microstructure of a two-layer, single-track deposit; and (b) microstructure of the overlapping area in a two-layer, multipletrack deposit (laser power: 420 W; traverse speed: 5 mm/s) VOLUME 36A, DECEMBER 2005 METALLURGICAL AND MATERIALS TRANSACTIONS A

9 Fig. 11 Microstructure of a wall-shaped, 12-layer deposit (laser power: from 400 to 285 W; traverse speed: 4 mm/s): (a) half of the transverse section, (b) bulk of the longitudinal section, and (c) top of the longitudinal section. growth as the number of layers increases. This can be accomplished in the LENS process by means of a melt-pool sensor (MPS) subsystem that provides closed-loop control during deposition. The MPS subsystem in the LENS machine is designed to automatically adjust the laser power to maintain a constant surface area of the melt pool, which is useful for eliminating the melt-pool size variations that can be caused by any changes in thermal and geometrical conditions during processing. In depositing the sample shown in Figure 11, the laser power changed from the starting 400 W for the deposition of the first layer to 285 W for the deposition of the last layer. As also seen in Figure 11, SX deposits of multiple layers can be fabricated successfully by performing the automatic control of the laser power in the LENS process. The SX nature is well preserved in the bulk of the deposit, i.e., except the region of stray grains in the top layer shown in Figures 11(a) and (c), as evidenced by the consistent primary growth of dendrites along the [001] direction in Figures 11(a) and (b). IV. CONCLUSIONS Single-crystal deposits of the Ni 3 Al-based IC221W alloy were successfully produced on a SX Ni-base superalloy substrate by means of the LENS process. The processing parameters influenced the SX fabricability from both the propensity for stray grain formation and the solidification cracking susceptibility. A simple relationship between the G/V ratio and the processing parameters (P, V b, T 0 ) was derived, which can be used qualitatively to guide the proper selection of processing conditions to maintain the columnar dendritic growth during laser deposition. As a general rule, the combination of a relatively low processing velocity and a reasonably low laser power is advantageous for producing SX deposits free of stray grains and cracks. It is important to perform automatic control over the laser power during the LENS processing of SX deposits of multiple layers in order to maintain the SX growth conditions. The microstructure of the IC221W deposits consists of mainly -Ni 3 Al and some (disordered solid solution) and Ni 5 Zr phases. ACKNOWLEDGMENTS The authors gratefully acknowledge the financial support of this work by the National Science Foundation through Award Nos. DMI and DMI made through its Division of Manufacturing & Industrial Innovation. They thank Dr. V.K. Sikka, Oak Ridge National Laboratory, for providing the IC221W alloy powder used in this study. Thanks also are due to Dr. Andrew Deal, Lehigh University, for his assistance with the OIM work. METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 36A, DECEMBER

10 REFERENCES 1. N.S. Stoloff, C.T. Liu, and S.C. Deevi: Intermetallics, 2000, vol. 8, pp V.K. Sikka: in High Temperature Aluminides & Intermetallics, S.H. Whang, C.T. Liu, D.P. Pope, and J.O. Stiegler, eds., TMS, Warrendale, PA, 1990, pp Physical Metallurgy & Processing of Intermetallic Compounds, N.S. Stoloff and V.K. Sikka, eds., Chapman & Hall, New York, NY, M.L. Santella and Z. Feng: Analysis of Weld Solidification Cracking in Cast Nickel Aluminide Alloys, Trends in Welding Research, Proc. 4th Int. Conf, H.B. Smartt, J.A. Johnson, S.A. David, eds., Gatlinburg, TN, June 1995, ASM INTERNATIONAL, Materials Park, OH, 1995, pp K.M. Flores and R.H. Dauskardt: Scripta Mater., 1997, vol. 36, pp W. Hofmeister, M. Griffith, M. Ensz, and J. Smugeresky: JOM, 2001, vol. 53 (9), pp G.K. Lewis and E. Schlienger: Mater. Design, 2000, vol. 21, pp P.A. Kobryn, E.H. Moore, and S.L. Semiatin: Scripta Mater., 2000, vol. 43, pp P.C. Collins, R. Banerjee, S. Banerjee, and H.L. Fraser: Mater. Sci. Eng. A, 2003, vol. A352, pp W. Liu and J.N. DuPont: Scripta Mater., 2003, vol. 48, pp W. Liu and J.N. DuPont: Metall. Mater. Trans. A, 2003, vol. 34A, pp W. Liu and J.N. DuPont: Metall. Mater. Trans. A, 2004, vol. 35A, pp M. Gaumann, S. Henry, F. Cleton, J.D. Wagniere, and W. Kurz: Mater. Sci. Eng. A, 1999, vol. A271, pp M. Gaumann, C. Bezencon, P. Canalis, and W. Kurz: Acta Mater., 2001, vol. 49, pp Handbook of Ternary Alloy Phase Diagram, P. Villars, A. Prince, and H. Okamoto, eds., ASM International, Materials Park, OH, 1995, vol. 3, p J.M. Vitek, S.A. David, and L.A. Boatner: Sci./Technol. Weld Joining, 1997, vol. 2, pp M.C. Fleming: Solidification Processing, McGraw-Hill, New York, NY, 1974, pp M. Rappaz, S.A. David, J.M. Vitek, and L.A. Boatner: Metall. Mater. Trans. A, 1989, vol. 20A, pp D. Rosenthal: Trans. ASME, 1946, vol. 43 (11), pp W. Liu and J.N. DuPont: Acta Mater., 2004, vol. 52, pp R.R. Unocic and J.N. DuPont: Metall. Mater. Trans. B, 2004, vol. 35B: J.W. Park, S.S. Babu, J.M. Vitek, E.A. Kenik, and S.A. David: J. Appl. Phys., 2003, vol. 94, pp VOLUME 36A, DECEMBER 2005 METALLURGICAL AND MATERIALS TRANSACTIONS A