Phase transformations and magnetic properties

The phase transformation and Curie temperatures of the samples produced in publications II-V are summarized in Table 4.5. The LFMS and DSC curves obtained for the as-built sample (IV-1) and the heat-treated sample (IV-11) in publication IV are shown in Figure 4.9a-b. Here, two phase transformations are observed upon heating: the first-order structural transformation from 14M martensite to cubic, and the second-order phase transformation from cubic ferromagnetic to cubic paramagnetic. The reverse transformations are observed upon cooling.

The phase transformations of the as-built samples (III-OPT and IV-1) in publications III and IV were broad, in the range of ~20–40 ℃, and exhibited a large deviation from the values obtained for similar alloy compositions in the scientific literature. It was suggested in publication III that the observed shift and broadening of the phase transformations could correspond to minor compositional variations, lattice strains, or the presence of multiple phases or martensite structures.

4.2 Characterization of as-built and heat-treated samples 57

Figure 4.9: The phase transformations of an as-built sample (IV-1) and a heat-treated sample (HT, IV-11). (a) LFMS measurement. Zero-magnetization level corresponds to the magnetization of paramagnetic austenite. (b) DSC measurement. (c) The X-ray diffractograms obtained for the heat-treated sample (IV-11) upon heating and cooling over the martensite transformation temperature. Intensities have been scaled and the baseline is used as an offset for each measurement. Recognized peaks are indexed relative to the coordinate system of the parent austenite unit cell. (Modified from publication IV.)

Table 4.6: ANOVA table for the width of the austenite-martensite transformation (ΔTM), including homogenization temperature (Th) and time (th) as continuous predictors.

(Modified from publication IV.) ΔTM (℃)

Source DF Adj SS Adj MS F-value P-value Th 1 16.822 16.822 113.47 0.002 th 1 3.760 3.760 25.36 0.015 Th2 1 0.020 0.020 0.13 0.738 th2 1 0.372 0.372 2.51 0.212 Th*th 1 0.012 0.012 0.08 0.795

Error 3 0.445 0.148 - -

Total 8 21.300 - - -

Figure 4.10: Effect of the applied heat-treatment on the width of the austenite-martensite transformation. (Modified from publication IV.)

All homogenized samples in publication IV showed a recovery of the typical magneto-structural properties and reversible martensitic transformations in relation to the chemical composition of each sample. Notably, the sample annealed at 800℃ for 4 h without homogenization at a higher temperature also showed a recovery of the typical properties.

This implies that the untypical magneto-structural properties of as-built Ni-Mn-Ga are mostly related to the atomic disorder and quenched-in stress from the L-PBF process, as low-temperature annealing itself is not expected to induce a large chemical homogeneity increase. The observed variations in the phase transformation temperatures between samples were small and likely relate to the small differences in the exact chemical compositions of the samples. The ANOVA in Table 4.6 and Figure 4.10 show that increasing either the homogenization temperature or time resulted in a decrease in the width of the austenite-martensite transformation (ΔTM). This change was suggested to attribute to the general homogeneity increase and the gradual dissolving of the 10M martensite phase. However, the observed transitions are wider than the transitions observed in conventional Ni-Mn-Ga single crystals.

Figure 4.9c shows the results of an additional XRD experiment performed on the sample homogenized at 1080 °C for 24 h to investigate the anomaly (two transformation peaks) observed upon heating during the DSC measurement of the same sample; see Figure 4.9b for reference. The observed peak splitting may relate to intermartensitic transformation in Ni-Mn-Ga. The X-ray diffraction patterns were collected from 60–70° angles during heating and cooling of the sample over martensitic and reverse martensitic transformations. The diffraction data revealed the absence of intermediate phases during these transformations; hence, it was suggested that the splitting of the transformation peak upon heating observed in the DSC measurements occurred due to the presence of structural defects such as pores or cracks. The observed anomaly was not observed upon cooling in DSC, which supports this argument.

4.3 Actuation experiments 59

Figure 4.11: The VSM magnetization curves obtained for the as-built sample IV-1 and a sample homogenized at 1080 °C for 24 h (HT, IV-11). (Modified from publication IV.)

Figure 4.11 shows the VSM hysteresis loops obtained for the as-built sample IV-1 and the heat-treated samples (exemplified by IV-11) in publication IV. The measured saturation magnetization of the as-built sample was 25 Am²/kg. All heat-treated samples exhibited a clear magnetization value increase of nearly ~170% in comparison to the as-built sample. Additionally, the saturation magnetization values obtained for the heat-treated samples were nearly identical, averaging ~68±1 Am²/kg, which was expected because all samples exhibited mostly the same crystal structure and were equally dense.

Additionally, the obtained value is in agreement with the values previously reported in the literature for Ni-Mn-Ga alloys with the same electron concentrations (Heczko &

Straka, 2004). The coercive field of the as-built sample was ~31 mT, which was relatively high compared to the values of ~8 mT and ~4–5 mT obtained for the annealed sample (IV-2) and the homogenized samples (IV-3→IV-11), respectively. Each heat-treated sample exhibited a saturation field in the approximate range of 0.6–0.8 T.

4.3

This section reveals and discusses the results of the third stage in our experimental research, namely the demonstration of the MSM effect in L-PBF-built Ni-Mn-Ga.

In publication V, polycrystalline Ni-Mn-Ga samples were manufactured via L-PBF and subsequently heat-treated near the melting temperature to increase the chemical homogeneity and degree of atomic ordering and to induce a coarse grain structure. The applied L-PBF and heat-treatment parameters and the corresponding sample properties are summarized in Table 3.3, Table 4.1, and Table 4.5, respectively. Figure 4.12a shows a photo of the as-built samples on a high-purity Ni substrate. The samples were built using different combinations of process parameters so that they would exhibit different values of VED, thus inducing different amounts of Mn evaporation. Subsequently, the

heat-treated samples exhibited different chemical compositions with corresponding crystal structures; see Figure 4.12b. Notably, the samples with the largest amount of Mn evaporation also exhibited the largest standard deviations in composition after the heat treatment. The results presented here are consistent with those obtained earlier in publication III, showing that the evaporation of Mn during L-PBF increases with increasing VED. The XRD patterns obtained for the heat-treated samples are displayed in Figure 4.12c. Peaks belonging to the 10M, 14M, and NM martensites are indexed relative to the cubic coordinate system. The unindexed peaks originate from the modulated superstructure. Importantly, each sample exhibited a single martensite phase structure, which implies that the heat treatment near the melting temperature effectively increased the chemical homogeneity and reduced the L-PBF-induced internal stresses. The samples exhibited different phase transformation temperatures, corresponding to the chemical composition and crystal structure of each sample, with a clear increase in the martensite transformation temperature and a decrease in the Curie temperature observable from the sample with the largest amount of Mn (V-1) to that with the least (V-6).

Figure 4.12: General characterization of the Ni-Mn-Ga samples manufactured via L-PBF in publication V. (a) Photo of the samples as-built on a high-purity Ni substrate. The inset shows the applied scanning strategy. (b) Chemical composition (error bars correspond to standard measurement deviations) and the corresponding martensitic crystal structure at ambient temperature for the heat-treated samples. (c) A set of XRD patterns obtained for heat-treated samples at ambient temperature. (d) Optical polarized light image of the heat-treated and polished sample V-2, revealing large grains (outlined with red lines) with twins. BD notes the build direction. The white rectangle marks the part of the sample containing the largest grain, which was further investigated in the magnetic actuation experiments. (Modified from publication V.)

4.3 Actuation experiments 61

Figure 4.12d shows an optical polarized light image of the sample V-2, revealing a coarse grain structure with martensitic twins ranging from a few micrometres to hundreds of micrometres in width. The spherical pores visible in the figure may have formed due to gas entrapment during the L-PBF manufacture; see the discussion of the results obtained in publication III. No crack formation was observed in the produced samples, possibly due to the reduction of the L-PBF-induced internal stresses enabled by the smaller thickness of the samples compared to the thicker samples in publications II-IV.