Page 107 - Read Online
P. 107
Page 32 of 45 Mooraj et al. J Mater Inf 2023;3:4 https://dx.doi.org/10.20517/jmi.2022.41
doped samples are shown in Figure 16A and B. Moorehead et al. used DED to produce a library of Mo-Nb-
[194]
Ta-W HEAs, which were then examined via XRD . The XRD patterns are shown in Figure 16C. In this
case, all the compositions tested showed a full BCC phase.
However, energy dispersive spectroscopy (EDS) analysis showed that Nb was segregated to the
interdendritic region. In order to tune the phase structure in a CoCrFeNiNb alloy system, our group used
x
DED by placing pre-alloyed CoCrFeNi powder in one powder feeder and pure Nb powder in another.
Then, the feed rates from each feeder were adjusted to produce a graded material with increasing Nb along
the build direction. The CoCrFeNiNb compositions were selected as going from x = 0 to x = 1, where x was
x
increased by about 0.1 for every 1 mm increase in height. Synchrotron XRD (SXRD) at the Cornell High
Energy Synchrotron Source was then performed along the build direction to analyze the phase composition
in each region. The beam size was maintained as 0.5 × 0.5 mm such that the measured phase compositions
correspond accurately to the designed compositions. As the Nb content increased, an FCC/Laves dual-
phase structure formed with the Laves phase volume fraction increased from 0% to 57% as the Nb content
increased from 0 to 20 at. %. Figure 16D shows the SXRD results taken for each composition.
In addition to XRD analysis, EBSD can offer a means to probe phases at higher spatial resolution (around
200 nm for EBSD vs. about 1 mm for XRD), which may be especially important for gradient compositional
[185]
libraries where the compositional change can be quite drastic over small length scales . EBSD also has the
advantage of being equipped onto SEM facilities. Thus, EDS analysis can often be carried out in parallel
such that phase and composition can be resolved almost simultaneously. For experiments that include many
phases and samples, there is a need to automate the phase analysis process to make the process more
efficient [195,29] . Many groups have used high-throughput SXRD to rapidly identify phases in combinatorial
[196]
material libraries . The majority of these studies used thin films produced by magnetron
sputtering [29,195,197,198] . However, there are almost no studies of large bulk materials that used similar high-
throughput methods for phase structure analysis.
It should be noted that a bottleneck for the previously mentioned methods is the human intervention
needed during data analysis. This analysis can require impractical time commitments when the number of
compositions reaches hundreds or thousands. Thus, the development of automated systems for analyzing
XRD, EBSD, and EDS data is crucial to ensure that experimental results of high-throughput experiments
can be achieved in a timely manner. Machine learning has recently shown impressive results in this field by
correctly indexing phases within an EBSD pattern without requiring human input to guess at the present
[199]
phases . Although this process has not been attempted for HEAs, it shows great promise to be applied to
new materials. Extending this practice further to carry out more in-depth analysis, such as Rietveld
refinement for XRD patterns, will greatly accelerate the development of future HEAs.
Magnetic property measurement
Tang et al. added Ho to a FeCoNi(CuAl) alloy to investigate the effect of the addition of rare earth (RE)
0.8
[200]
element on the magnetic properties of this system . The initial parent alloy showed a fully FCC structure,
and adding Ho led to the formation of a secondary BCC phase. Increasing the atomic fraction of Ho led to
an increase in the volume fraction of the BCC phase fraction and no change in the lattice parameter of the
FCC phase. As shown in Figure 17A, increasing the Ho content led to lower energy losses via eddy currents
and hysteresis until x = 0.05. Once x = 0.07, the energy losses increased substantially. The decrease in
hysteresis loss is due to lower Cu segregation at the BCC-FCC phase boundaries with increasing Ho. This
[200]
segregation leads to a lower magnetic domain pining effect, decreasing hysteresis losses . At x = 0.07, the
Cu and Ho tend to segregate heavily to phase boundaries, leading to higher hysteresis losses. The BCC