Magnetostrictive performance of additively manufactured CoFe rods using the LENS ( TM ) system

Magnetostrictive materials exhibit a strain in the presence of a variable magnetic field. While they normally require large, highly oriented crystallographic grains for high strain values, metal additive manufacturing (3D printing) may be able to produce highly textured polycrystalline rods, with properties comparable to those manufactured using the more demanding free standing zone melting (FSZM) technique. Rods of Co75.8Fe24.2 and Co63.7Fe36.3 have been fabricated using the Laser engineered net shaping (LENSTM) system to evaluate the performance of additively manufactured magnetic and magnetostrictive materials. The 76% Co sample showed an average magnetostriction (λ) of 86 ppm at a stress of 124 MPa; in contrast, the 64% Co sample showed only 27 ppm at the same stress. For direct comparison, a Co67Fe33single crystal disk, also measured as part of this study, exhibited a magnetostriction value of 131 and 91 microstrain in the [100] and [111] directions, respectively, with a calculated polycrystalline value (λs) of 107 microstrain. Electron back scattered diffraction (EBSD) has been used to qualitatively link the performance with crystallographic orientation and phase information, showing only the BCC phase in the 76% Co sample, but three different phases (BCC, FCC, and HCP) in the 64% Co sample.


I. INTRODUCTION
Magnetostrictive materials strain due to changes in an applied magnetic field.In order to optimize this strain performance, materials need to be highly textured, requiring slow crystal growth methods such as the Bridgman and Free-standing Zone Melting (FSZM) techniques, resulting in very large grains.While these large grains are useful for the magnetic-component of their performance, they also degrade the mechanical performance, giving an unobstructed path to fracture.In addition, such fabrication is expensive and challenging.FeGa-and FeAl-based magnetostrictors were nevertheless shown to have advantages over traditionally brittle transduction materials (i.e., Terfenol-D, PZT, etc.), and have even shown the ability to be rolled into thin sheets. 1,2While these newer materials possess the capability of being used in structural applications, the traditional growth techniques relegate these materials to specialty applications only.Interestingly, it is not the grain size that is important, but a Author to whom correspondence should be addressed.Electronic mail: Nicholas.j.jones1@navy.mil2158-3226/2018/8(5)/056403/5 8, 056403-1 © Author(s) 2017 the texture that is required for magnetostrictive performance.By reducing grain size but maintaining columnar growth, it seems possible to reduce the fabrication costs of these materials, and improve their mechanical performance as well.Additive manufacturing techniques may provide a possible path towards structural usage and the application of smart materials in unique areas of interest.Additive manufacturing (AM) is a method of material fabrication in which each layer of material is deposited individually and sequentially.While much of the research in the past has been on polymeric systems, metals-based systems have seen much recent development.These systems are more challenging to use and normally require much smaller build plates, due to general requirements for an inert atmosphere, elevated temperature chambers, small particle sizes, and the usage of a laser or electron beam as the energy source for melting metals, although not all systems fit these characteristics.Electron-beam systems, being controlled by magnetic lenses, provide a much faster control than mechanical translation stages used for laser processing.Dehoff et al. have shown the ability to vary microstructures from isotropic polycrystal to nearly single crystal microstructures inside the same build using electron beam, powder bed fusion systems. 3,4While these properties are highly desirable for magnetic applications, many of these commercial powder bed systems only allow use of verified structural powders, and the powder bed systems also require a large amount of material to begin a build.The large quantity of powder used for the powder bed renders these techniques prohibitively expensive.In addition, since the composition in the powder bed is fixed, designing alloys through these manufacturing techniques for research is not efficient.Laser-engineered net shaping (LENS TM ) 5 is similar to the above-mentioned techniques, but requires much less starting material, and allows for compositional variations during the build.LENS is a directed-energy AM method in which powder is fed into a melt pool created by a laser.Rather than melting consecutive layers in a bed of powder, the powders are melted at the location of interest only, requiring only as much powder as is need for the part.Geng et al. have shown its use in rapidly evaluating the magnetic properties of FeCo samples of varied compositions. 6For rapid, initial validation of AM for a new class of material with laboratory quantities of material, the LENS technique is a much more viable technique to begin with.
In this study, two Co 1-x Fe x rods were printed using the LENS technique to evaluate the performance of this class of magnetostrictive materials using non-traditional material fabrication techniques.Cobalt-rich FeCo alloys have shown notable magnetostriction values, up to 150 microstrain in Fe 30 Co 70 columnar growth samples 7 and up to 260 microstrain in thin-film form; 8

II. EXPERIMENTAL PROCEDURE
Two cylinders ∼7 mm in diameter and ∼50 mm in height were 3D printed using the LENS technique with compositions of Co 63.7 Fe 36.3 and Co 75.8 Fe 24.2 .For convenience, these will be referred to as the 64% and 76% Co samples, respectively.To print the rods, Fe and Co powders (> 99.9% purity) were fed from two different powder hoppers at the desired Co/Fe ratio.The substrate was rastered in the x-y plane and the print head translated in the z-direction (direction of growth) to deposit the individual layers with laser powers of 320 -450 W and scanning speeds of ∼50 mm/s.The as-received surface of the samples was rather rough for the application of strain gages (Fig. 1a), so the samples were removed from the substrate plate and their surfaces machined smooth (Fig. 1b).The magnetostriction of the samples was characterized under compression as a function of applied stress (up to 124 MPa), using a dead-weight apparatus similar to that shown in Ref. 13.Both stress and magnetic field were along the rod axis.Magnetostriction was measured using two MicroMeasurements strain gages (WK-06-500GB-350) attached to opposite sides of the rod; the saturation magnetostriction value was determined at 700 Oe.
Following the magnetostriction measurements, sections at each end of the rods were cut off for further analysis using resonant ultrasound spectroscopy (RUS), vibrating sample magnetometry (VSM), and scanning electron microscopy (SEM).RUS samples were polished in the shape of rectangular parallelepipeds (2-3 mm on a side) from these rod end slices.Disk slices were used to measure magnetization versus field hysteresis loops.For SEM, disk samples were cut in half across the diameter and hot mounted in a conductive bakelite resin with carbon filler, with a final vibratory polishing step using a 0.4 micron colloidal silica suspension.Both the planar (circular face) and the vertical (rectangular face) cross sections were imaged.A Hitachi SU 6600 SEM with an EDAX Apollo 40 electron detector was used at 30 kV to determine crystal orientation/phase and composition, using electron backscattered diffraction (EBSD) and energy dispersive x-ray spectroscopy (EDS), respectively.For EBSD, a 70 • tilt was used along with a step size of 1 µm; EDAX Orientation Imaging Microscopy (OIM) Analysis TM software was used to analyze the images.
To be able to interpret the performance of these AM rods, a (100) Co 67 Fe 33 single crystal sample was fabricated in order to determine both λ 100 and λ 111 .While this composition does not directly match either rod, the magnetostriction values in this range vary quite monotonically, so the performance can be used to evaluate the rod values, nonetheless.Appropriate quantities of high purity Fe (99.95 wt.%) and Co (99.9 wt.% metals basis) were cleaned and arc melted under an argon atmosphere several times.The buttons were then remelted and the alloy drop cast into a copper chill cast mold to ensure compositional homogeneity throughout the ingot.The crystal was grown in a resistance furnace from the as-cast ingot in an alumina Bridgman style crucible.The ingot was heated under a pressure of 5.0 x 10 -6 torr up to 1500 • C to degas the crucible and charge.The chamber was then backfilled to a pressure of 2.8 x 10 3 kPa with high purity argon.This over-pressurization was done near melting to diminish gas pockets from being trapped in the cone region of the crystal and also to minimize the amount of evaporation from the melt during crystal growth.The ingot was further heated to 1600 • C and held at this temperature for 1 hour to allow thorough mixing before withdrawing the sample from the heat zone at a rate of 5mm/hr.Grain growth was achieved by annealing the sample at 1050 • C for 0.5 hr followed by slowly cooling to 860 • C and dwelling for 7 days.The sample was then slow cooled at a rate of 10 • C/min.Etching in equal parts HNO 3 and H 2 O was done to make grain boundaries more distinct.A large grain was selected and oriented by Laue back-reflection and the sample cut from the ingot, yielding a disk with a (100) face that was 4.98 mm in diameter and 2.51 mm in thickness.The magnetostriction constants, λ 100 and λ 111 , were measured using a using a Kyowa strain gage (KFL-1-120-C1-11) applied along <100> and <110> directions, respectively, on the (100) face.The sample was rotated 360 • about an axis normal to the (100) face in a constant, in-plane saturating 20 kOe magnetic field.(Note: λ is the magnetostriction coefficient, whereas (3/2) λ is the saturation magnetostriction, as described in Ref. 14).These results were not form effect corrected.

III. RESULTS AND DISCUSSION
The magnetostriction of the 3D printed rods is shown in Fig. 1c.Despite being rather close in composition and printed in the same run, they exhibit strikingly different performances, with the 76% Co rod approaching 100 ppm and the 64% Co rod being under 30 ppm.Neither rod saturated with the available applied stress, however, the slope of the 76% Co rod appears to be flattening out with increasing compressive stress.There is more scatter in the two strain gages for the 76% Co rod, however this is not unexpected and can be attributed to the larger magnetostriction values, since both have similar microstructures, as is shown below.It should be noted that, as the applied compressive stress increases, the values should approach the saturation magnetostriction value, (3/2) λ.
In contrast, the measured saturation magnetostriction values of the Co 67 Fe 33 single crystal disk were (3/2) λ 100 = 197 ppm and (3/2) λ 111 = 136 ppm.The isotropic polycrystalline value, calculated as λ s = (3/5) λ 111 + (2/5) λ 100 , is 107 ppm, with a full saturation value, (3/2) λ s , of 161 ppm.The 76% Co rod is approaching these values, however the compressive stress required for this rod was not achievable at this time; it is, however, expected that the 76% Co sample should have a lower saturation value than the calculated 161 ppm, since the magnetostriction should decrease with increasing cobalt content.
A VSM was used to measure the saturation magnetization of both samples near the substrate and at the free surface; both ends showed similar performance.The specific saturation magnetization, σ s , was ∼181 emu/g for the 64% Co sample, and 200 emu/g for the 74% Co sample.The 76% Co sample is directly in line with literature values, while the 64% Co sample is ∼35 emu/g lower than expected; 15 this decreased performance can be attributed to the presence of non-equilibrium phases as is shown below, due to the solidification in the AM process.
The elastic properties (compressional and shear moduli) of the two compositions were measured with RUS, and are reported here in GPa.The 64% Co sample was tested at both ends of the rod.A larger void density was observed for the free-rod end.The density (calculated from geometry and mass) of this end was 8,301 kg/m 3 , compared with that of the substrate end of 8,324 kg/m 3  To further clarify the above results, both EBSD and EDS were performed on the rods, to determine crystal orientation and compositional variations inside the sample.Since these samples are in the cobalt rich region of the phase diagram, three structures were used to map the samples: body centered cubic (BCC), face centered cubic (FCC), and hexagonal close packed (HCP).Both rods showed a weakly textured microstructure with varying crystallite sizes and no observable columnar growth (see Fig. 2; the individually colored pixels are voids or other areas where no crystallographic match could be found).Interestingly, the lower cobalt sample (64%) showed a microstructure with all three phases toward the substrate end, with an FCC matrix, smaller BCC regions, and HCP islands.These phases also showed chemical segregation; the HCP islands were rich in Co and the BCC regions were poor in Co.At the free end, the matrix became BCC with a more refined grain size, with HCP islands and large regions of FCC.The phase mixture of this sample can be linked to both the poor magnetostriction performance and the lower saturation magnetization value, since both γ (FCC) and ε (HCP) CoFe have only a small positive or negative strain response, 16,17 and may also explain why the elastic constants at the free end are closer to the single crystal value.In contrast, the 76% Co rod had a microstructure primarily of BCC; this fits with both the magnetostriction and magnetization data, which are more in alignment with a BCC isotropic polycrystal.It is curious that the sample with the higher cobalt content was the more homogeneous one, possessing a BCC structure.Threedimensional crystal structure analysis and higher cobalt percentage elastic constant measurements may further assist in understanding the varied behavior and observed magnetostriction.

IV. CONCLUSION
Additively manufactured, magnetostrictive Co 1-x Fe x samples have been fabricated using LENS TM .While both the 76% and 64% Co samples exhibited notable magnetostriction, the 76% Co sample had much higher values, approaching calculated polycrystalline values.This discrepancy in performance was attributed to the presence of multiple crystalline phases in the 64% Co sample along with chemical inhomogeneity, while the 76% Co sample was generally single phase with more uniform grain size.These results are also in alignment with RUS and VSM measurements looking at elastic constants and magnetization, respectively.By adjusting the printing parameters, it may be possible to get better aligned, columnar growth of a single BCC phase; this may be achievable by reducing the x-y motion of the LENS TM stage and only operating in the z-direction, or by switching to an electron beam powder bed fusion system.Such optimized crystallography would improve the observed magnetostrictive response, and enable such a technology for use in building active structures, or static structures with active components.Further crystallographic analysis and elastic constant measurements may help optimize the printing parameters.

FIG. 1 .
FIG. 1. CoFe rods fabricated using LENS TM showing (a) the as-received surface condition (1/8 inch markings on ruler) and (b) the machined surface finish with an applied strain gage.Red and blue were used to discriminate between the substrate and the free-rod ends, respectively.(c) Magnetostriction of the 76% Co and 64% Co LENS TM fabricated rods as a function of axial compressive stress.The filled-in squares are an average of the two strain gages values; the dashed lines are guides to the eye.
. The samples had high-Q spectra and fitted the resonances well, which means that the average elastic properties through the sample were isotropic.The various crystallites detected through EBSD are orders of magnitude smaller than the wavelength of the waves producing the resonances.The elastic constants found for the 64% Co sample were c 11 = 227.4and c 44 = 67.6 for the free end and c 11 = 271.9and c 44 = 72.7 for the substrate end.It is interesting to compare these values with the elastic constants we measured for a 64 at.%Co single crystal CoFe sample, which were found to be c 11 = 225.6,c 12 = 142.16, and c 44 = 135.6.In addition, the constants measured for the 76 at.%Co LENS TM sample at the substrate end were both higher, with c 11 = 291.13and c 44 = 81.47.

FIG. 2 .
FIG. 2. Bottom (substrate end) cross-section EBSD images of LENS TM fabricated Co 1-x Fe x rods showing (a) the inverse pole figure (IPF) for the 64% Co with (b) its corresponding phase map, as well as (c) the IPF for the 76% Co sample.