# Fileset

[Improvement of Oxidation Resistance for Ni-Base Single Crystal Superalloy TMS-238 by Suppression of Sb Segregation at Oxide-Substrate Interface Using CaO.pdf](https://mdr.nims.go.jp/filesets/fd919ff0-401d-4105-8a4f-36d7a745419b/download)

## Creator

Hirotoshi Maezawa, [Chihiro Tabata](https://orcid.org/0000-0001-6597-4998), [Yuji Takata](https://orcid.org/0009-0006-9362-0250), [Jun Uzuhashi](https://orcid.org/0000-0003-2023-8158), [Tadakatsu Ohkubo](https://orcid.org/0000-0003-3548-1951), [Tadaharu Yokokawa](https://orcid.org/0000-0003-1595-6729), Hiroshi Harada, [Kyoko Kawagishi](https://orcid.org/0000-0001-7652-9232), Shinsuke Suzuki

## Rights

This version of the article has been accepted for publication, after peer review (when applicable) and is subject to Springer Nature’s AM terms of use, but is not the Version of Record and does not reflect post-acceptance improvements, or any corrections. The Version of Record is available online at: https://doi.org/10.1007/s11663-023-02948-5[In Copyright](http://rightsstatements.org/vocab/InC/1.0/)

## Other metadata

[Improvement of Oxidation Resistance for Ni-Base Single Crystal Superalloy TMS-238 by Suppression of Sb Segregation at Oxide/Substrate Interface Using CaO](https://mdr.nims.go.jp/datasets/00de9e25-b06e-49d9-a6be-cfa1a16a9d06)

## Fulltext

1  Title 1 Improvement of Oxidation Resistance for Ni-base Single Crystal Superalloy TMS-238 by 2 Suppression of Sb Segregation at Oxide/Substrate Interface Using CaO 3  4 Authors 5 HIROTOSHI MAEZAWA1,2, CHIHIRO TABATA2,3, YUJI TAKATA2, JUN UZUHASHI2, 6 TADAKATSU OHKUBO2, TADAHARU YOKOKAWA2, HIROSHI HARADA2, KYOKO 7 KAWAGISHI2,3, and SHINSUKE SUZUKI1,3,4 8  9 Affiliations 10 1 Department of Applied Mechanics and Aerospace Engineering, Waseda University: 3-4-1 Okubo, 11 Shinjuku, Tokyo, 169-8555, Japan 12 2 National Institute for Materials Science (NIMS): 1-2-1 Sengen, Tsukuba, Ibaraki, 305-0047, Japan 13 3 Department of Materials Science, Waseda University: 3-4-1 Okubo, Shinjuku, Tokyo, 169-8555, 14 Japan 15 4 Kagami Memorial Institute for Materials Science and Technology, Waseda University: 2-8-26 16 Nishiwaseda, Shinjuku, Tokyo, 169-0051, Japan 17 HIROSHI HARADA is now with Superalloy Design Laboratory, 3-18-33, Azuma, Tsukuba, Ibaraki, 18 305-0031, Japan. 19  20 Corresponding Author 21 HIROTOSHI MAEZAWA, hirowolves5@suou.waseda.jp 22   23 mailto:hirowolves5@suou.waseda.jp2  Abstract 24 Impurity Sb degrades the oxidation resistance of Ni-base superalloys after Al2O3 crucible melting. 25 The purpose of this study is to clarify the effect of CaO on the oxidation resistance of Ni-base single 26 crystal superalloy TMS-238 containing Sb and the reaction mechanism. Single crystal alloys 27 containing Sb were cast using an Al2O3 crucible and a CaO crucible. In the cyclic oxidation tests, the 28 mass did not decrease for the alloy after CaO crucible melting. In STEM observations for the alloy 29 after CaO crucible melting, Sb-Ca-O inclusion was found. In 3DAP for the oxide/substrate interface 30 of the alloy after CaO crucible melting, Sb segregation was not detected. Moreover, the CaO rod was 31 inserted in the TMS-238 melt to observe the CaO/melt interface. In EPMA observations of the 32 surface of the CaO rod, Sb was observed. Therefore, the higher oxidation resistance of the alloy after 33 CaO crucible melting than that of Al2O3 crucible melting is due to the suppression of Sb segregation 34 at the oxide/substrate interface by removing Sb from the melt and fixing Sb in the alloy. These occur 35 by the formation of Sb-Ca-O on the surface of CaO, and the formation of Sb-Ca-O inclusions in the 36 alloy, respectively. 37  38  39 Ⅰ. INTRODUCTION 40 Ni-base superalloys are used in turbine blades for jet engines and gas turbines for power generation 41 because of their superior high-temperature properties. To improve the thermal efficiency of jet 42 engines and gas turbines for power generation, it is necessary to increase the turbine inlet 43 temperature. For this reason, rare metals such as Re, Ta, W, and Ru are added to Ni-base superalloys 44 to improve their high-temperature properties.[1–5] Our research group developed the 6th generation 45 Ni-base single crystal (SC) superalloy TMS-238, which has the highest temperature capability in 46 history.[1] However, an increase in material cost due to the addition of rare metals is an issue. 47 Therefore, our research group has been studying the "direct and complete recycling method", where 48 used turbine blades are directly melted and recast to produce recycled materials.[6,7] This leads to the 49 possibility of using various low-purity materials to reduce the cost of ingots production. However, 50 low-purity materials contain impurities.[8–10] Takata et al. revealed that low melting point metallic 51 impurity Sb significantly degrades the oxidation resistance of Ni-base SC superalloy TMS-238, and 52 it was assumed that Sb segregated at the oxide/substrate interface, leading to the acceleration of the 53 spallation of the oxide layers and degradation of the oxidation resistance.[11] To improve the 54 oxidation resistance of Ni-base SC superalloy containing Sb, it is necessary to suppress Sb 55 segregation at the oxide/substrate interface by removing Sb from the melt or fixing Sb in the alloy. 56 Regarding the removal of impurities from the melt, it was found that about 90 pct of Sb can be 57 removed from Ni-base alloy NCF-1 (14/17Cr-6/10Fe-Ni) by MSR (Metal bearing Solution Refining) 58 using 6 pct Ca-CaF2 flux.[12] However, the use of CaF2 has been reduced due to environmental 59 3  concerns.[13] It was also found that about 40 pct of Sb can be removed from crude copper as Sb 60 oxides by using CaO-Na2SO4 flux.[14] Our research group reported the effect on removing impurity S 61 by contacting CaO with Ni-base superalloy melts. The reaction of Al in the melt with CaO forms 62 calcium aluminate and Ca. It was revealed that S is removed by CaS incorporated into calcium 63 aluminate.[15,16] 64 Regarding the fixation of an impurity in the alloy, CaO crucible melting for the Ni-base superalloy 65 is effective for fixing impurity S in the alloy. Ca dissolves into the melt by the reaction of Al and 66 CaO in the molten metal, and S is fixed in the alloy by the formation of CaS inclusions. This 67 suppresses S segregation at the oxide/substrate interface and suppresses the spallation of the oxide 68 layers, thereby improving the oxidation resistance.[17,18] 69 Based on these previous studies, it can be hypothesized that impurity Sb can be removed from the 70 melt and fixed in the alloy as inclusions by using CaO, which can improve the oxidation resistance 71 of the alloys. However, the effect of CaO on the oxidation resistance of Ni-base SC superalloys 72 containing Sb has not been clarified. 73 The purpose of this study is to clarify the effect of CaO on the oxidation resistance of Ni-base SC 74 superalloys TMS-238 containing Sb and the reaction mechanism. The recycling of TMS-238, which 75 contains many rare metals, is beneficial for the use spread of the highest temperature capable 76 superalloy. Cyclic oxidation tests were conducted to reveal whether the oxidation resistance had 77 improved by using the CaO crucible, and microstructures were investigated to reveal whether Sb 78 segregation at the oxide/substrate interface had been suppressed and if Sb inclusions had formed in 79 the alloy. In addition, a CaO rod was inserted into TMS-238 melt, and the CaO rod was analyzed to 80 observe the reaction at the CaO/melt interface to reveal whether Sb had been removed from the melt, 81 and the mechanism was discussed. 82  83  84 Ⅱ. EXPERIMENTAL PROCEDURES 85 A. Evaluation of Oxidation Resistance and Microstructure Investigation of Alloys 86 An ingot of Ni-base single crystal superalloy TMS-238[1] (IHI Master Metal Co., Ltd.) was used in 87 this study. Three types of alloys were fabricated in this study: the alloy without Sb addition that was 88 melted in an Al2O3 crucible is called TMS-238(Al2O3), the alloy with Sb addition that was melted in 89 an Al2O3 crucible is called TMS-238+Sb(Al2O3), and the alloy with Sb addition that was melted in a 90 CaO crucible is called TMS-238+Sb(CaO). 91 A 2 kg ingot was placed in an Al2O3 crucible or a CaO crucible and heated at 1600 °C using high 92 frequency vacuum heating. The pressure within the furnace was kept under 6 × 10-2 Pa. A Ni tube 93 filled with a bulk of 10 ppm of Sb was added to the melt for TMS-238+Sb(Al2O3) and 94 TMS-238+Sb(CaO). The melt was poured into a ZrO2-base mold with a starter and a selector held at 95 4  1500 °C. Single crystal round bars (diameter: 11 mm, length: 135 mm) were cast by withdrawing the 96 mold from the heating chamber to the cooling chamber at 200 mm/h. 97 The alloys were solution heat treated at 1335 °C for 20 h, aging heat treated at 1150 °C for 2 h, and 98 secondary aging heat treated at 870 °C for 20 h. 99 The chemical compositions for each alloy were analyzed using inductivity-coupled plasma optical 100 emission spectrometer (ICP-OES) (Agilent, 720-ES). For ICP-OES, 10 specimens with 0.5 mm in 101 thickness were cut out from the starters of each alloy. The concentrations of trace elements Sb, Ca, 102 and S were measured for each alloy using glow discharge mass spectrometry (GD-MS) (Thermo 103 Scientific, VG9000). For GD-MS, a specimen with 5 mm in height was cut out from the starters of 104 each alloy. 105 Cyclic oxidation tests were performed as follows. Round bars of each alloy were cut into 106 cylindrical specimens (diameter: 9 mm, height: 5 mm) and the surfaces of the specimens were 107 ground with SiC P600 and washed with acetone. Cyclic oxidation tests at 1100 °C for 1 h and 108 air-cooling for 1 h in laboratory air were performed on all specimens. This process was repeated for 109 200 cycles, and the change in mass was measured after several cycles. 110 Microstructures were investigated by using scanning electron microscopy (SEM), 3D atom probe 111 (3DAP), and scanning transmission electron microscopy (STEM) techniques. A needle-shaped 112 specimen for 3DAP was prepared using focused ion beam (FIB)-SEM dual-beam system (FEI, 113 Helios G4UX) from the TMS-238+Sb(CaO) performed the oxidation test at 1100 °C for 1 h. The 114 elemental distribution at the oxide/substrate interface was measured using 3DAP (CAMECA, 115 LEAP5000XS). Circular backscattered (CBS)-SEM observations of the starter of 116 TMS-238+Sb(CaO) before heat treatment were performed, and a thin foil specimen was also 117 prepared using FIB-SEM. The STEM system with energy-dispersive X-ray spectroscopy (EDS) (FEI, 118 Titan G2 80-200 ChemiSTEM) was used to obtain high-angle annular dark-field (HAADF)-STEM 119 images, EDS elemental maps, and nano-beam electron diffraction (NBED) analysis. 120  121 B. Evaluation of Reaction at CaO/Melt Interface by Inserting CaO Rod in Melt 122 CaO rod insertion experiments were performed using the high frequency vacuum heating furnace in 123 Figure 1 as follows. Similar experiments previously performed for removal of S from Ni-base 124 superalloy melt.[15,16] Dense CaO rods (diameter: 15 mm, length: 80 mm, porosity: <1 pct) were used 125 to observe the reaction at the CaO/melt interface. Ingots of TMS-238 and pure Ni (Nilaco Co.) were 126 used to compare the reaction depending on the different amounts of Al in the melt. Additive Sb was 127 prepared by filling a Ni tube with a bulk of Sb equal to 100 ppm of the weight of the alloys. 128 A 500 g ingot was placed in an Al2O3 crucible and melted in the high frequency vacuum heating 129 furnace. The pressure in the furnace was adjusted to less than 50 Pa at the start of heating (Figure 130 2(a)). The furnace was preset so that the melt heated up to 1500 °C after 56 min from the start of 131 5  heating. The CaO rod was preheated by suspending above the melt with a 40 mm gap between the 132 bottom surface and the melt (Figure 2(b)). Sb was added by putting the Ni tube containing Sb into 133 the melt at exactly 65 min from the start of heating (Figure 2(c)). The melt was heated up to 1600 °C 134 within 1 min from the addition of Sb (Figure 2(d,e)). This temperature was kept throughout the rest 135 of the experiment. 66 min from the start of heating is defined as the holding time t=0 s. In the 136 experiments using TMS-238, two types of experiments were carried out in which heating was 137 stopped at t=0 and 240 s. The reason why heating was stopped at t=0 s is that Sb may evaporate from 138 the melt and be deposited on the CaO rod before being inserted into the melt, and the CaO rod before 139 being inserted into the melt needed to be analyzed to distinguish Sb reacted in the melt from Sb 140 deposited by evaporation. In the experiment using pure Ni, one type of experiment was carried out in 141 which the heating was stopped at t=240 s. These experiments will be referred to as 0s(TMS-238), 142 240s(TMS-238), and 240s(pure Ni), respectively. In the experiment of 0s(TMS-238), the CaO rod 143 was pulled up from the preheating position when the heating was stopped at t=0 s (Figure 2(d)). In 144 the experiments of 240s(TMS-238) and 240s(pure Ni), the CaO rod was inserted into the melt, and 145 the length of the rod below the surface of the melt was 35 mm at t=0 s (Figure 2(e)). The CaO rod 146 was kept in the melt and pulled out from it when the heating was stopped at t=240 s (Figure 2(f)). 147  148  149 Fig. 1—Schematic diagram of the high frequency vacuum heating furnace for the CaO rod insertion 150 experiments (adapted from the reference[16]). Some parts were modified for the experiments in this 151 study. 152  153 6   154 Fig. 2—Flow and schematic diagrams of the CaO rod insertion experiments. (a) Start of heating ingot, 155 (b) preheating CaO rod, (c) adding Sb, (d) pulling up CaO rod from preheating position and finish of 156 heating (0s(TMS-238)), (e) inserting CaO rod, and (f) pulling out CaO rod from melt and finish of 157 heating (240s(TMS-238) and 240s(pure Ni)). 158  159 After the CaO rod insertion experiments, two sector-shaped specimens were cut out from the upper, 160 middle, and lower parts of the alloys, respectively, as shown in Figure 3(a). These specimens were 161 dissolved in hydrofluoric acid and nitric acid, and the elemental concentrations in the alloys were 162 measured. The concentrations of Ni, Co, Cr, Mo, W, Al, Ta, Hf, Re, and Ru in TMS-238 and Al in 163 pure Ni were measured using ICP-OES (Agilent, 720-ES) and the concentrations of Sb were 164 measured using inductivity-coupled plasma mass spectrometer (ICP-MS) (Agilent, 7700). 165 CaO rods used in the experiments were analyzed using electron probe micro analyzer (EPMA) 166 (Shimadzu Co., EPMA-1610) and x-ray photoelectron spectroscopy (XPS) (JEOL Ltd., JPS-9010). 167 As shown in Figure 3(b), CaO rods were cut 5 mm from the bottom, filled with resin, and polished. 168 Backscatter electron (BSE) images and elemental maps at the interface between the CaO rod and 169 resin were obtained. Spectra for the binding energy of the bottom surface of the CaO rod were also 170 obtained. 171  172 7   173 Fig. 3—Schematic diagram of the specimen preparation. (a) Specimens for ICP-OES and ICP-MS of 174 the alloys after CaO rod insertion experiments and (b) specimens for EPMA and XPS analyses of the 175 CaO rods. 176  177  178 Ⅲ. RESULTS 179 A. Evaluation of Oxidation Resistance and Microstructure Investigation of Alloys 180 Table I shows the nominal composition of TMS-238[1] and the chemical compositions of the 181 samples prepared for this study. Sb concentrations in both TMS-238+Sb(Al2O3) and 182 TMS-238+Sb(CaO) were higher than that in TMS-238(Al2O3), and Ca concentration was high in 183 TMS-238+Sb(CaO). In GD-MS, 1 ppm was the smallest order of magnitude in this study. Therefore, 184 note that S contents were similar in all alloys. 185  186 Table I.  Chemical compositions of the alloys (Ni bal.)[1] 187 Sample wt pct  ppm Co Cr Mo W Al Ta Hf Re Ru  Sb Ca S TMS-238 (nominal) 6.5 4.6 1.1 4.0 5.9 7.6 0.1 6.4 5.0  - - - TMS-238 (Al2O3) 6.3 4.4 1.1 4.0 5.8 7.7 0.10 6.4 5.0  0.11 <0.1 1.9 TMS-238 +Sb(Al2O3) 6.5 4.5 1.1 4.1 6.1 7.9 0.12 6.4 5.0  8.1 0.02 2.1 TMS-238 +Sb(CaO) 6.5 4.5 1.1 4.1 6.0 7.9 0.11 6.4 5.0  9.5 5.5 1.0  188  Figure 4 shows the results of the cyclic oxidation test. There was no obvious difference in the initial 189 oxidation for all alloys. However, after 50 cycles, as the number of cycles increased, the masses of 190 TMS-238(Al2O3) and TMS-238+Sb(Al2O3) decreased. The loss of mass for TMS-238+Sb(Al2O3) 191 8  was larger than that for TMS-238(Al2O3), which was also seen in the study by Takata et al.[11]. 192 Remarkably, the mass did not decrease for TMS-238+Sb(CaO). 193  194  195 Fig. 4—Results of the cyclic oxidation tests of TMS-238(Al2O3), TMS-238(Al2O3), and 196 TMS-238+Sb(CaO) for 200 cycles at 1100 °C in laboratory air. 197  198 Figure 5 shows the result of 3DAP of the oxide/substrate interface of TMS-238+Sb(CaO) after an 199 oxidation test at 1100 °C for 1 h. Figure 5(a) shows the 3D atom map. The alloy containing Ni and 200 Al, and the oxide layers containing Al and O were measured. The concentration of S was also 201 measured to take the influence of both Sb and S segregation at the oxide/substrate interface on the 202 oxidation resistance into account. Figure 5(b) shows the concentration profiles of Ni, Al, O, and 34 203 Da, which were drawn under the same condition as the study by Tabata et al.[17]. The reason why the 204 peak at 34 Da shows the S concentration is described in Tabata et al.[17]. The amount of S 205 segregation was almost equal to that of the alloy melted in the CaO crucible shown in the study by 206 Tabata et al.[17]. Figure 5(c,d) shows the mass spectrum. Peaks at 60.45 Da, 61.45 Da, 120.9 Da, and 207 122.9 Da, which indicate 121Sb++, 123Sb++, 121Sb+, and 123Sb+, could not be recognized in the spectrum. 208 Generally, the chemical sensitivity of 3DAP technique is a few per a million atoms.[19] Thus, it is 209 concluded that there is no detectable Sb segregation at the oxide/substrate interface of 210 TMS-238+Sb(CaO) by 3DAP method. 211  212 9   213 Fig. 5—Results of 3DAP for TMS-238+Sb(CaO) oxidized at 1100 ̊C for 1 h. (a) 3D atom map, (b) 214 concentration profiles of Ni, Al, O, and 34 Da with standard deviation (the yellow region on (a) 215 represents the area analyzed to obtain (b)) , (c) mass spectrum (121Sb++ and 123Sb++, 60.0 to 62.0 Da), 216 and (d) mass spectrum (121Sb+ and 123Sb+, 120.0 to 124.0 Da). 217  218 FIB-SEM and STEM observations were performed to confirm the formation of Sb inclusions in the 219 alloy by CaO crucible melting. Figure 6 shows the results for the starter of TMS-238+Sb(CaO) 220 before heat treatment. Figure 6(a) shows the CBS-SEM image, and the inclusion, which looks black 221 when observed using SEM, was found near the grain boundary and contained Sb with EDS attached 222 to the FIB system. Note that inclusions that look white contained Re etc. and are likely not to affect 223 the fixation of Sb in the alloy. The inclusion, which looks black when observed using SEM, near the 224 grain boundary was lifted out and thinned to the TEM specimen using the FIB-SEM system. Figure 225 6(b) shows the cross-sectional HAADF-STEM image of the inclusion. Note that the surfaces of the 226 inclusion and the alloy were capped with Pt as protection against FIB damage. The center of the 227 inclusion (the darkest contrast) was confirmed to be a void. Figure 6(c) shows the STEM-EDS maps 228 shown by at pct. Sb, Ca, Zr, and O were found in the inclusion. Figure 6(d) shows the structure of 229 the inclusion was determined by using NBED technique and the NBED pattern. The pattern shows 230 10  the halo ring, indicating that the inclusion is amorphous. 231  232  233 Fig. 6—Results of FIB-SEM and STEM observations for the starter of TMS-238+Sb(CaO) before 234 heat treatment. (a) CBS-SEM image, (b) HAADF-STEM image, (c) STEM-EDS maps, and (d) 235 NBED pattern. The plot on (b) represents the point analyzed to obtain (d). 236  237 B. Evaluation of Reaction at CaO/Melt Interface by Inserting CaO Rod in Melt 238 Table II shows the Sb contents in the alloys of 0s(TMS-238), 240s(TMS-238), and 0s(pure Ni) 239 obtained by ICP-MS. Moreover, Table III shows the composition of 240s(TMS-238) and Al content 240 in the alloy of 0s(pure Ni) obtained by ICP-OES. There was no significant difference between the 241 nominal composition of TMS-238 and the chemical composition of the alloy of 240s(TMS-238). 242 The alloy of 240s(pure Ni) contains Al. 243  244 Table II.  Sb contents in the alloys (ppm) 245 11  Sample Sb 0s(TMS-238) 93.7 240s(TMS-238) 107.4 240s(pure Ni) 69.6  246 Table III.  Chemical compositions of the alloys (Ni bal., wt pct) 247 Sample Co Cr Mo W Al Ta Hf Re Ru 240s(TMS-238) 4.4 4.5 1.1 4.1 5.9 7.7 0.11 6.9 5.0 240s(pure Ni) - - - - 0.025 - - - -  248 Figure 7 shows the results of EPMA analysis for the CaO rods. For 240s(TMS-238), Al was 249 detected on the surface and in the particle boundaries of the CaO rod. For 240s(pure Ni), Al was 250 detected in the particle boundaries of the CaO rod. The particle boundaries were formed because the 251 CaO rods was made by sintering CaO particles. The results for Sb are after interference corrections. 252 The interference correction method applied in this study is described in Appendix. Sb was detected 253 on the surface of the CaO rods of 240s(TMS-238) and 0s(TMS-238). The Sb concentration for 254 240s(TMS-238) was higher than that of 0s(TMS-238). 255  256 12   257 Fig. 7—Results of EPMA analysis for the CaO rods. BSE images and elemental maps showing the 258 distributions of Al and Sb (after interference correction, explained in Appendix). The unit of the 259 color bars is wt pct. 260  261 Figure 8 shows the results of XPS analysis of the CaO rods before and after the CaO rod insertion 262 experiment. The dotted red line shows the peak position of Sb2O3[20], and the single-dashed blue line 263 shows the peak position of Sb[21]. Neither Sb nor Sb2O3 peaks were observed for the as-received 264 CaO rod and the CaO rod of 240s(pure Ni). On the other hand, the Sb2O3 peak was observed for the 265 CaO rods of 0s(TMS-238) and 240s(TMS-238). 266  267 13   268 Fig. 8—Results of XPS analysis of the CaO rods. (a) As-received, (b) 0s(TMS-238), (c) 269 240s(TMS-238), and (d) 240s(pure Ni). The dotted line shows the peak position of Sb2O3[20], and the 270 single-dashed line shows the peak position of Sb[21]. 271  272  273 Ⅳ. DISCUSSION 274 A. Effect on Improvement of Oxidation Resistance 275 The effect on the improvement of oxidation resistance by CaO crucible melting is discussed here. In 276 the cyclic oxidation tests shown in Figure 4, TMS-238+Sb(Al2O3) showed the largest weight loss, 277 while TMS-238+Sb(CaO) had little change in the mass, indicating that CaO crucible melting 278 improved the oxidation resistance for Ni-base superalloy containing impurity Sb. Because there is no 279 obvious difference in the initial oxidation of all samples, the difference in weight loss is considered 280 to be the difference in the amount of spallation of the oxide layers. Takata et al. reported that the 281 oxidation resistance of Ni-base superalloy containing Sb after Al2O3 crucible melting had 282 significantly degraded and Sb segregated at the oxide/substrate interface.[11] Takata et al. reported that 283 1.1 ppm of Sb had a negative effect on the oxidation resistance of the alloy, and when the Sb content 284 was at 3.8 ppm, Sb segregation was observed. The Sb concentration is about 0.15 at pct at 285 14  oxide/substrate interface. In this study, the amount of Sb in the alloy melt using a CaO crucible was 9.5 286 ppm, which is significantly larger than the Sb-containing alloy melt using an Al2O3 crucible in the 287 previous study[11]. However, for the sample melt using a CaO crucible, the Sb peak was not detected at 288 the oxide/substrate interface in the mass spectrums from Figure 5(c). Therefore, Sb segregation at 289 oxide/substrate interface is considered to have facilitated a spallation of oxide layers in the samples 290 melted in an Al2O3 crucible. Thus, it is thought that CaO crucible melting suppressed Sb segregation at 291 the oxide/substrate interface, resulting in the prevention of spallation of the oxide layers and the 292 improvement of oxidation resistance. 293 It should also be noted that S segregated at the oxide/substrate interface as the concentration profiles 294 shown in Figure 5(b). However, the amount of S segregation at the oxide/substrate interface was 295 thought to have been reduced by CaO crucible melting as reported in the study by Tabata et al.[17]. If 296 CaO only had an effect on Sb segregation, it would be expected that oxidation resistance would 297 recover from TMS-238+Sb(Al2O3) to TMS-238(Al2O3) because S segregation would still remain. 298 However, the oxidation resistance improved from TMS-238+Sb(Al2O3) to TMS-238+Sb(CaO) by 299 using CaO in this study. Therefore, CaO crucible melting suppressed both S and Sb segregations at the 300 oxide/substrate interface of the Sb-containing alloy. When only the suppression of Sb segregation is 301 considered, it is thought that the oxidation resistance of the Sb-containing alloy improved to the same 302 level or more than that of the Sb-free alloy at least. In addition, the suppression of S segregation is 303 considered, it is thought that the oxidation resistance of the Sb-containing alloy improved to the same 304 level of the Sb-free alloy melted in a CaO crucible. 305  306  307 B. Reaction at CaO/Melt Interface 308 The process of Ca formation contributing to the formation of inclusions is discussed here. Ca and 309 Al were detected in the CaO rod of 240s(TMS-238) as the elemental maps of CaO rods shown in 310 Figure 7 and Ca was detected in TMS-238+Sb(CaO) as the results of GD-MS for the alloys shown in 311 Table I. These results suggest that, as reported in the previous studies[16,18,22,23], the following 312 reaction occurred, in which Al in the melt reacted with CaO to form calcium aluminate and Ca. 313 Through this reaction, Ca dissolved into the melt.  314 3CaO + 2Al = 3Ca + Al2O3#[1]  CaO+Al2O3=CaAl2O4##[2]  Next, the reaction and the effect on removing Sb are discussed. As shown in Figure 7, Sb was not 315 detected in the as-received CaO rod but concentrated on the surface of the CaO rod of 0s(TMS-238), 316 indicating that Sb evaporated from the melt, and the deposit may have been observed on the surface 317 of the CaO rod before inserting. Moreover, Sb concentrated on the surface of the CaO rod of 318 15  240s(TMS-238), with a higher Sb concentration than that of 0s(TMS-238). Therefore, CaO has the 319 effect on removing Sb from Ni-base superalloy melt. However, the free surface area of the melt was 320 large relative to the volume of the melt, resulting in a large amount of Sb evaporation. Therefore, it 321 was not possible to distinguish the amount of Sb removal by CaO and the amount of Sb evaporation. 322 Here, the removal reaction of Sb at the CaO/melt interface is discussed. Sb was not observed on the 323 CaO rod of 240s(pure Ni). Al was also detected in the particle boundaries of the CaO rod of 324 240s(pure Ni). The alloy of 240s(pure Ni) contained small amounts of Al as shown in Table III. It is 325 considered that the Al entered the melt from the Al2O3 crucible and entered the particle boundaries of 326 the CaO rod but was not enriched on the surface of the CaO rod. Therefore, the reaction of Eq. [1] 327 did not occur on the surface of the CaO rod. On the other hand, the inclusion in the alloy after CaO 328 crucible melting consists of Sb, Ca, O, etc., and Al did not concentrate in the inclusion as shown in 329 Figure 6. These results suggest that the removal reaction of Sb at the CaO/melt interface requires 330 simple substance Ca, which is formed on the surface of the CaO rod by the reaction between CaO 331 and Al. The reaction at the CaO/melt interface considered in this study is shown in Figure 9. First, 332 Ca, which was formed by the reaction shown in Eq. [1], was not only dissolved into the melt but also 333 existed on the surface of the CaO rod. Next, Sb reacted with Ca and O on the surface of the CaO rod, 334 and was removed from the melt through the formation of Sb-Ca-O layer. Since the CaO crucible 335 (CaO: >99.9 vol.%) and the CaO rod (CaO: 98.9, MgO: 0.58, SiO2: 0.34, Al2O3: 0.02, Fe2O3: 0.02 336 vol.%) have similar compositions, it is thought that Sb-Ca-O layer was formed on the surface of the 337 CaO crucible as well. 338  339  340 Fig. 9—Schematic diagrams of the reaction at the CaO/melt interface considered in this study. (a) Al 341 in the melt touches the surface of CaO, (b) Ca-Al-O layer is formed and Ca is dissolved in the melt 342 by the reduction of CaO, and (c) it is assumed that Sb-Ca-O layer is formed on the surface of CaO. 343  344 C. Reactions in Melt and Alloy During Solidification 345 The formation process of the inclusion in the melt is discussed. The reaction layer consisting of Sb, 346 Ca, and O formed at the CaO/melt interface as shown in the previous section, suggesting that the 347 16  complex oxides, which contain Sb, Ca, and O, were formed in the melt at 1600 °C (Figure 10(b-1)). 348 Moreover, the Sb-Ca-O inclusion existed near the grain boundary of the starter of 349 TMS-238+Sb(CaO) as shown in Figure 6, suggesting that Sb, Ca, and O were distributed to the 350 liquid phase (Figure 10(a-2)) or that the complex oxides formed in the melt gathered in the final 351 solidification part (Figure 10(b-2)). Note that the inclusion shown in Figure 6 also presented signs of 352 Zr, which most likely came from the ZrO2-base mold (66.2 vol pct of ZrO2, 32.7 vol pct of SiO2, 353 etc.), that was used to cast the alloy. Various oxides that are components of the crucibles and molds 354 used to fabricate the alloys may be included in these inclusions. 355 Next, the effect on suppressing Sb segregation at the oxide/substrate interface is discussed. Figure 356 10(d,e) shows the mechanism behind the improvement of oxidation resistance of Ni-base superalloy 357 containing impurity Sb using CaO. The oxidation resistance of TMS-238+Sb(Al2O3) is degraded 358 because Sb diffuses and segregates to the oxide/substrate interface during oxidation tests shown in 359 Figure 4, and this results in the spallation of the oxide layers, as Takata et al. revealed[11]. On the 360 other hand, the oxidation resistance of TMS-238+Sb(CaO) improved. This is because the amount of 361 Sb segregation at the oxide/substrate interface decreased as shown in Figure 5. The suppression of 362 Sb segregation at the oxide/substrate interface is likely caused by the removal of Sb from the melt 363 and the fixation of Sb in the alloy, analogous to the previous studies on S[17,18]. In this study, it is 364 likely that the formation of Sb-Ca-O inclusions, which is shown in Figure 6, fixes Sb to the alloy, 365 similar to the effect on fixing impurity S revealed by Tabata et al.[17,18]. According to previous 366 research, CaS did not affect the mechanical properties such as creep and fatigue strength[1,6]. Moreover, 367 Sb-Ca-O inclusions found within the alloy do not affect to the spallation of the oxide layers because 368 Sb-Ca-O inclusions were not observed at the oxide/ substrate interface, and its size were smaller than 369 the CaS inclusions. Moreover, in terms of Sb contents, Sb content in TMS-238+Sb(CaO) is slightly 370 higher than that of TMS-238+Sb(Al2O3). It is most likely because some of the Sb remained in the melt 371 due to the formation of Sb-Ca-O inclusions inside the melt. In the previous studies[17,18], S contents 372 after CaO crucible melting were also slightly higher than those after Al2O3 crucible melting. It is 373 thought that these were due to the formation of CaS inclusions. Therefore, the improvement of 374 oxidation resistance can be attributed to the prevention of spallation of the oxide layers, which is due 375 to the suppression of Sb segregation at the oxide/substrate interface by removal of Sb and fixation of 376 Sb in the alloy by CaO. 377  378 17   379 Fig. 10—Schematic diagrams of the reactions in melt and alloy during solidification and oxidation 380 test considered in this study. (a-1) Sb, Ca, and O exist in the melt, (a-2) Sb, Ca, and O are distributed 381 to the liquid phase, (b-1) the complex oxides are formed in the melt, (b-2) the complex oxides gather 382 in the final solidification part, and (c) the inclusions are formed in the final solidification part. 383 During the oxidation test, (d) Sb is fixed in the alloy and (e) Sb segregates at the oxide/substrate 384 interface. 385  386  387 V. CONCLUSIONS 388 In this study, the following has been made clear through the CaO crucible melting experiments and 389 the CaO rod insertion experiments for Ni-base single crystal superalloy TMS-238 containing 390 impurity Sb. 391 1. CaO crucible melting improves oxidation resistance of the alloy containing Sb. This is because 392 CaO has the effect on removing Sb from the melt and fixing Sb in the alloy. This leads to the 393 suppression of Sb segregation at the oxide/substrate interface, and the prevention of spallation of the 394 oxide layers. 395 2. The removal of Sb from the melt occurs by the formation of Sb-Ca-O on the surface of CaO. 396 Sb-Ca-O is formed by the reaction of Sb with O and Ca, which is formed by the reduction of CaO by 397 Al. 398 3. The fixation of Sb in the alloy occurs by the formation of Sb-Ca-O inclusions in the alloy. Ca 399 18  dissolves into the melt by the reaction of CaO and Al, which then reacts with O and Sb to form 400 Sb-Ca-O inclusions. 401  402  403 ACKNOWLEDGMENTS 404 This research was financially supported by the Council for Science, Technology and Innovation 405 (CSTI), Cross-ministerial Strategic Innovation Program (SIP), ‘‘Materials Integration for 406 revolutionary design system of structural materials’’ (Funding agency: JST) and The Technical 407 Association of Refractories, Japan. The authors are grateful to Dr. Makoto Osawa and Dr. Toshio 408 Osada at NIMS, and Mr. Takahide Horie, a graduate student at Waseda University, for the helpful 409 discussions, Ms. Kyoko Suzuki at NIMS, for the support of the microstructural investigation, and Mr. 410 Lei Tan, a graduate student at Waseda University, for the support of the CaO rod insertion 411 experiment. 412  413 CONFLICT OF INTEREST 414 On behalf of all authors, the corresponding author states that there is no conflict of interest. 415  416 REFERENCES 417 [1] K. Kawagishi, A. Yeh, T. Yokokawa, T. Kobayashi, Y. Koizumi, and H. Harada: Superalloys 2012, 418 2012, pp. 189–95. 419 [2] P. Caron: Superalloys 2000, 2000, pp. 737–46. 420 [3] L. Liu, J. Meng, J. Liu, M. Zou, H, Zhang, X. Sun, and Y. Zhou: J. Mater. Sci. Technol. 421 (Shenyang, China), 2019, vol. 35, pp. 1917-24. 422 [4] A. Heckl, S. Neumeier, M. Göken, and R.F. Singer: J. Mater. Sci. Eng. A., 2011, vol. A528, pp. 423 3435–44. 424 [5] A.C. Yeh and S. Tin: Scr. Mater., 2004, vol. 52, pp. 519–24. 425 [6] S. Utada, Y. Joh, M. Osawa, T. Yokokawa, T. Kobayashi, K. Kawagishi, S. Suzuki, and H. 426 Harada: Superalloys 2016, 2016, pp. 589–99. 427 [7] H. Harada, K. Kawagishi, T. Kobayashi, T. Yokokawa, M. Osawa, M. Yuyama, S. Suzuki, Y. Joh, 428 and S. Utada: US Patent, patent number 10689741 (2020.06.23). 429 [8] R.T. Holt and W. Wallace: Int. Met. Rev., 1976, vol. 21, pp. 1–24. 430 [9] G.B. Thomas and T.B. Gibbons: Met. Technol., 1979, vol. 6, pp. 95–101. 431 [10] G.W. Meetham: Met. Technol., 1984, vol. 11, pp. 414–8. 432 [11] Y. Takata, T. Sugiyama, M. Yuyama, J. Uzuhashi, T. Okubo, H. Harada, S. Suzuki, and K. 433 Kawagishi: J. Japan Inst. Met. Mater., vol. 87, pp. 88-94 (in Japanese). 434 [12] Y. Nakamura, T. Mukai, M. Kuwabara, and K. Arihara: Tetsu-to-Hagané, 1977, Vol. 63, pp. 435 19  2246-53 (in Japanese). 436 [13] H. Wang, T. Zhang, H. Zhu, G. Li, Y. Yan, and J. Wang: ISIJ Int., 2011, Vol. 51, pp. 702–6. 437 [14] T. Sakai, T. Nakamura, F. Noguchi, and Y. Ueda: J. Min. Metall. Inst. Jpn., 1987, Vol. 103, pp. 438 587-92 (in Japanese). 439 [15] Y. Kishimoto, S. Utada, T. Iguchi, Y. Mori, M. Osawa, T. Yokokawa, T. Kobayashi, K. 440 Kawagizhi, S. Suzuki, and H. Harada: Metall. Mater. Trans. B, 2020, vol. 51, pp. 293-305. 441 [16] Y. Kishimoto, T. Kono, T. Horie, T. Yokokawa, M. Osawa, K. Kawagishi, S. Suzuki, and H. 442 Harada: Metall. Mater. Trans. B, 2021, vol. 52, pp. 1450-62. 443 [17] C. Tabata, K. Kawagishi, J. Uzuhashi, T. Ohkubo, K. Hono, T. Yokokawa, H. Harada, and S. 444 Suzuki: Scr. Mater., 2021, vol.194, pp. 113616. 445 [18] C. Tabata, K. Kawagishi, J. Uzuhashi, T. Ohkubo, T. Yokokawa, H. Harada, and S. Suzuki: 446 Metall. Mater. Trans. A, 2022, vol. 53, pp. 2452-8. 447 [19] B. Gault, M.P. Moody, J.M. Cairney, and S.P. Ringer: Atom Probe Microscopy, vol. 160, 448 Springer New York, New York, NY, 2012, pp. 3-7. 449 [20] R. Izquierdo, E. Sacher, and A, Yelon: Appl. Surf. Sci., 1989, vol. 40, pp. 175-7. 450 [21] W. K. Liu, W. T. Yuen, and R. A. Stradling: J. Vac. Sci. Technol., B: Microelectron. Nanometer 451 Struct., 1995, vol. 13, pp. 1539. 452 [22] R.J. Fruehan: Metall. Trans. B., 1978, vol. 9, pp. 287–92. 453 [23] T. Degawa and T. Ototani: Tetsu-to-Hagané, 1987, vol. 73, pp. 1684–90 (in Japanese). 454  455  456 APPENDIX 457 Interference corrections between Ca and Sb for EPMA on the CaO rods were carried out. In Figure 458 7, the elemental maps for Sb are the results after interference correction. As shown in Figure A1(a), 459 Sb was detected in the CaO rods, but because the detection wavelengths of Sb (3.4354 Å) and Ca 460 (3.3524 Å) are close, the peak intensity at the detection wavelength of Sb before correction is most 461 likely includes the Ca from the CaO rod. Therefore, the removal of Ca from the peak intensity at the 462 detection wavelength of Sb is needed to know the actual point of Sb existing. In this study, an 463 interference correction method, which subtracts the increment due to the peak intensity of Ca from 464 the peak intensity at the detection wavelength of Sb, was applied. The correction factor R, which is 465 the ratio of the average peak intensity at the detection wavelength of Sb, ICa(3.4354)0 , to the average 466 peak intensity at the detection wavelength of Ca, ICa(3.3524)0 , was obtained within the range shown in 467 Figure A1(b) for the as-received CaO rod. 468 R = ICa(3.4354)0 /ICa(3.3524)0 #[A1]  Multiplying the correction factor R by the peak intensity at the detection wavelength of Ca for each 469 sample, ICa(3.3524), the increment due to the peak intensity of Ca at the detection wavelength of Sb, 470 20  ICa(3.4354), was obtained. 471 ICa(3.4354) = R × ICa(3.3524)#[A2]  Subtracting the increment due to the peak intensity of Ca at the detection wavelength of Sb, 472 ICa(3.4354), from the peak intensity at the detection wavelength of Sb, ICa+Sb(3.4354), the peak intensity 473 of Sb at the detection wavelength of Sb, ISb(3.4354), was obtained. 474 ISb(3.4354) = ICa+Sb(3.4354) - ICa(3.4354)#[A3]  As shown in Figure 7, the Sb peak disappeared in the as-received CaO rod by the interference 475 correction. Therefore, the increment due to the peak intensity of Ca at the detection wavelength of 476 Sb was removed. 477  478  479 Fig. A1—The interference correction for EPMA of the CaO rods. (a) The distribution of Sb before 480 the interference correction. The unit of the color bars is wt pct. (b) The range used for the 481 interference correction. 482  Cover Letter Response Letter Article File Article File with Track Changes Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure A1