Nanoscaled eutectic NiAl-(Cr,Mo) composites with exceptional mechanical properties processed by electron beam melting

Eutectic NiAl-(Cr,Mo) composites are promising high temperature materials due to their high melting point, excellent oxidation behavior and low density. To enhance the strength, hardness and fracture toughness, high cooling rates are beneficial to obtain a fine cellular-lamellar microstructure. This can be provided by the additive process of selective electron beam melting. The very high temperature gradient achieved in this process leads to the formation of the finest microstructure that has ever been reported for NiAl-(Cr,Mo) in-situ composites. A very high hardness and fracture toughening mechanisms were observed. This represents a feasibility study towards additive manufacturing of eutectic NiAl-(Cr,Mo) in-situ composites by selective electron beam melting.


Materials and methods
A NiAl-28Cr-6Mo (at.-%) in-situ composite was cast using the Bridgman process with a growth rate of 1 mm/ min to create bars of 11.9 mm diameter 21 .
A cross-sectional slice of the ingot with a size of 11.9 × 50 × 5 mm 3 was prepared and placed onto a standard EBM steel base-plate in approximate central position of the build area. The exact position of the specimen with respect to the electron beam was determined by using in-situ electron optical imaging 22 and used to calculate the sample coordinates for subsequent remelting. Remelting was performed under high vacuum conditions at room temperature in an in-house developed EBM system with an electron beam gun by pro-beam GmbH & Co. KGaA (Gilching, Germany) and by using a beam current of 7 mA and an acceleration voltage of 60 kV. Four quadratic samples with each a side length of 10 mm and a spacing of 2 mm were molten in adjacent position onto the specimen. While the scan velocity was varied from 1,000, 500, 250 to 125 mm/s, the scan strategy was always a standard EBM hatch pattern with a line spacing of 100 µm and a rotation of 180° between adjacent lines. The samples were molten in direct succession without temporal interruption, starting with the highest scan velocity (i.e. lowest energy input) to minimize the effect of specimen heat-up on the solidification conditions. The microstructures were analysed using a Zeiss Crossbeam 540 FIB/SEM (Zeiss GmbH, Oberkochen, Germany). Since the phases present in the material exhibit a significant difference in atomic number, imaging back-scattered electrons (BSE) was used. Chemical composition was analysed via energy dispersive X-ray spectroscopy (EDS) with an INCA PentaFET-X3 from Oxford Instruments.
Atom probe specimen were produced by in-situ lift-out using a Zeiss Crossbeam 540 FIB/SEM as described in 23,24 . The APT experiments were carried out in a CAMECA LEAP 4000X HR (CAMECA Inc, Madison, Wisconsin) using pulsed voltage and laser mode to trigger field evaporation. Voltage pulsing was used for more accurate determination of the chemical composition while laser pulsed experiments, yielding larger datasets, were used to show the three dimensional nanostructure for all samples. The voltage was regulated to detect an ion on average in 1% of the pulses, with the pulse voltage set to 20% of the direct current standing voltage. For the laser pulsed experiments, a UV-laser with 355 nm wavelength was used, at a pulse energy of 75 pJ. The experiment (base) temperature was 49 K and a pulse repetition rate of 200 kHz for both voltage pulsed and laser pulsed experiments. The data processing was done using the commercial software IVAS 3.6.8 from CAMECA, implementing a reconstruction algorithm published in 25,26 .
The microhardness was determined with a NanoXP Nanoindenter from Keysight (Keysight Technologies Inc., Santa Rosa, CA, USA) using a Berkovich indenter tip geometry. The maximum indentation depth was 500 nm and the hardness was averaged over the depth range from 400 to 500 nm.

Results
Microstructure. The e-beam remelted layer exhibits similar microstructural features for all tested specimen. www.nature.com/scientificreports/ solid liquid interface in order to get planar lamellae. The coarse as-cast material with phase distances of 2.039 ± 1.213 µm and primarily solidified NiAl-dendrites differs from the fine-structured remelted layer. Figure 2 shows the decrease in cell size and phase distance with increasing scan velocity. For high scan rates at constant beam power, the energy input into the material is lower and the melt pool is smaller. As a result the layer thickness decreases, the solidification rate increases and finer phases form. For 1,000 mm/s the phase distance is 9 ± 2 nm, whereas for 125 mm/s coarser phase distance of 82 ± 19 nm are measured. In comparison to the directional solidification in a Bridgman process from Raj et al. 13 , where the phase distances are in a range from 0.8 to 4.1 µm a much finer microstructure is obtained. A linear relationship in the double logarithmic diagram is found for both processes, which is in agreement with the Jackson-Hunt Model, that describes the lamellar and rod-like growth in eutectic alloys 27 .
Previous investigations on NiAl-28Cr-6Mo reveal lamellar microstructures for all withdrawal rates 8,9,28 . In this study, the two-phase microstructure does not seem to be continuously lamellar. Figure 3 shows a coral-like arrangement of the cross-linked (Cr,Mo)-phase inside the NiAl-matrix. APT measurements support the trend observed in Fig. 2 of refining microstructure with increasing scan velocity. In Fig. 3a,b coarser cell boundaries as well as the shift in phase orientation at cell boundaries can be observed in the lower part of the tomographies. chemical phase analysis. EDS-measurements reveal a homogeneous composition throughout the remelted layer. There are no concentration gradients towards the surface or the as-cast material. The average composition of the remelted layer with a scan speed of 250 mm/s is near eutectic (Table 1). For other samples a minor trend of Cr-evaporation with decreasing scan speed can be observed. The lack of Al originates from evaporation during the casting process, since both as-cast and e-beam remelted layer exhibit similar composition. Despite of non-eutectic composition, the remelted layer exhibits a eutectic microstructure. The fast solidification enables the creation of a eutectic-like microstructure in a hypereutectic alloy, as it is described by Shang et al. 12 .
The phase specific chemical composition is determined using the partitioning ratio k i (Eq. (1)) of the NiAland (Cr,Mo)-phases. The partitioning ratio of Ni and Al as well as Cr and Mo is similar at different scan rates (Fig. 4). The solubility of foreign atoms increases with faster scan velocities. The solubility of Cr in NiAl at the highest scan speed is five times the solubility at the slowest scan speed. This is consistent with the trend observed by Sheng et al. 16 . However, the amount of foreign atoms is quiet low, compared to those in as-cast NiAl-(Cr,Mo) observed in 29 . The NiAl-phase exhibits a lack of Al and a light surplus of Ni in comparison with the stoichiometric ratio. The (Cr,Mo)-phase shows a lack of Cr due to evaporation and therefore a surplus of Mo. Minor contaminations of C and Si were introduced during the casting process without purpose. They strongly segregate in the (Cr,Mo)phase. Especially for C, a preferred accretion at the phase and cell boundaries can be observed.
Microhardness. The hardness of remelted layer approaches 7.7-12.7 GPa, which is significantly harder than the as-cast material (5.2 GPa for indents in the NiAl-phase and 6.7 GPa for indents in the (Cr,Mo)-phase). The highest hardness is found for the finest microstructure. Figure 5 displays the hardness as a function of the inverse square root of the phase distance. A linear relationship can be found in this plot, similar to grain boundary Another possible impact increasing the hardness difference of the as-cast material and the electron beam remelted layers is the presence of high residual stresses. High cooling rates in electron beam melting can create thermally induced stresses, especially for high melting intermetallic materials. However, the influence of residual stresses on hardness measurements is not so high and typically below 10% 31 and this probably plays only a minor role here. crack propagation. The as-cast material was not preheated in advance of electron beam remelting, so high thermal stresses caused by thermal shock, fast solidification and volume shrinkage were introduced. At slow scan velocities the energy input is high, so the whole sample will heat up and thermal stresses and crack formation decreases. Figure 6a exhibits large cracks that cut the remelted layer perpendicular. They stop in the coarse as-cast microstructure where the crack is deflected along the phase boundaries or energy is dissipated by plastic deformation of the (Cr,Mo)-phase. These crack-hindering mechanisms are already described elsewhere 8,9,13,14,32 . Figure 6b,c also show crack deflection in the e-beam remelted layer along the columnar texture parallel to the

Discussion
The nanostructured phase alignment of NiAl-and (Cr,Mo)-phases is achieved by the very fast solidification rate. The eutectic solidification as well as the low miscibility of both phases enables phase distances below 10 nm. It should be noted that phase distances were determined from BSE images, leading to systematic overestimation of the values. The relationship between the phase distance and the solidification rate according to Fig. 2 can be described by (Eq. (3)): with λ representing the phase distance in m. For electron beam remelting, the velocity v is the scan rate of the e-beam in m/s, whereas for directional solidification v represents the withdrawal rate in m/s. From Fig. 2 the parameters A and B have been determined to be 2.12 × 10 −8 1/s and − 0.96 for e-beam remelting. For Bridgman casting according to 13 , the parameters are 2.17 × 10 −12 1/s and − 1.28 . Shang et al. found that the Jackson-Hunt model is applicable not only for planar eutectics, but also for cellular eutectic alloys like in this study 12 . An accurate measurement of the temperature or the cooling rate in the harsh environment of electron beam melting is currently not possible. To overcome this gap, material specific numerical approaches can be used to simulate the process and deduce information about solidification conditions 33 .   www.nature.com/scientificreports/ The preferred lamellar microstructure could not be formed like it is present in as-cast alloys due to fast solidification 29 . Nucleation rate is increased, diffusion is suppressed, leading to prevention of lamellar growth. The (Cr,Mo)-phases in Figs. 2 and 3 are in fact cross-linked rods with feather-like arrangement inside the cells. The cross-linkage is anisotropic and cross sections of (Cr,Mo)-rods are not spherical. This indicates a tendency for the aimed lamellar arrangement. The phases are small at the cell center and coarsen towards the cell boundary. The solidification slows down towards the cell boundary, due to the recalescence released at the solid-liquid interface as described in 29 . Another influence on phase morphology is given by the melting range. Eutectic alloys, such as 33.2Ni-33.2Al-33.6Cr, solidify with a sudden liquid-solid transition 11 . The lack of Cr and Al as well as the presence of Si and C cause deviation off the eutectic. Local concentration gradients will provide undercooling at the solid-liquid interface, leading to instability and non-planarity of the interface. In this case, a continuous lamellar morphology is not possible. On top of that, the substitution of 6 at.-% Cr with Mo affects the eutectic point 27,34 . Figure 1 shows columnar cells growing along the temperature gradient. This directed microstructure is often observed in powder bed electron beam melting processes. As reported in many researches 19,35,36 , this texture leads to anisotropic behaviour of mechanical properties. In the case of this work, columnar cells consists of smaller, more equiaxed cells (see Fig. 2), leading to frequent interruption of the columnar cells. With the hardness measurements performed in this work, no influence of the orientation on the hardness has been detected. In general, influences of the orientation on nanoindentation measurements are often strongly visible on the different pile-up pattern, however not so strong on the absolute hardness numbers, as has been reported for example on lamellar TiAl alloys 37,38 . Other mechanical properties like fracture toughness will show stronger anisotropy, as can be seen in experiments on NiAl single crystals 39 .
For as-cast NiAl-28Cr-6Mo more than 20 MPam 1/2 was reported in 8,9 . In this work, fracture toughness was not determined, but similar crack deflection and crack hindering mechanisms were found. Coarse phases at the cell boundaries represent a preferred crack path. In case of a columnar textured material processed by electron beam remelting, the cracks are deflected alongside those columnar cells. Depending on the orientation of the columnar cell structure, anisotropic fracture behaviour is expected. Higher fracture toughness will appear for crack propagation perpendicular to the columnar texture. The fine feather-like phases exhibit crack deflecting potential to prevent crack growth perpendicular to the phases. In addition, similar crystal structures and lattice parameters of the NiAl-and the (Cr,Mo)-phase lead to a strong interfacial bonding. Figure 6d indicates, that cracks will rather be deflected along the NiAl-(Cr,Mo) interface than propagate perpendicular to the phases and interphases. Plastic deformation of the more ductile (Cr,Mo)-phase was not observed. Since cracks, which are formed in the remelted layer are stopped in the base material, fracture toughness of remelted material is expected to be smaller, but toughening mechanisms are still present.
For common powder bed selective electron beam melting, the sample usually is preheated to control residual stresses and annihilate crack formation in the material during processing 18 .

Summary and conclusion
This study is a first promising approach towards additive manufacturing of nanoscaled NiAl-(Cr,Mo) in-situ composites. The effects of fast solidification on microstructure and mechanical properties of the in-situ composite NiAl-28Cr-6Mo were analysed and the following results and conclusions were found: www.nature.com/scientificreports/ • E-beam remelted composites consists of cellular microstructure with a feather-like arrangement of (Cr,Mo)rods in NiAl-matrix. The preferred lamellar growth is supressed by fast solidification. • A non-eutectic alloy can solidify with a eutectic-like microstructure for high solidification rates.
• The nanostructured material with phase distances less than 10 nm exhibits superior hardness up to twice the hardness of the as-cast material, that can be described by the Hall-Petch relation. • Crack hindering mechanisms on different microstructural scales are present. The preferred crack path is the coarse cell boundary and the coral-like (Cr,Mo) network shows crack deflection.