# Fileset

[Effect of Ca addition on the oxidation resistance of Ni-Al alloy.pdf](https://mdr.nims.go.jp/filesets/91588bac-33d4-4dbc-9aa0-6dd2fe7a722d/download)

## Creator

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

## Rights

[In Copyright](http://rightsstatements.org/vocab/InC/1.0/)

## Other metadata

[Effect of Ca Addition on the Oxidation Resistance of Ni–Al Alloy](https://mdr.nims.go.jp/datasets/5cf296c8-b84d-4d6c-bbc4-196ebbbe7a53)

## Fulltext

1  Effect of Ca addition on the oxidation resistance of Ni-Al alloy  1  2 Chihiro Tabata1,2,3,*, Kyoko Kawagishi1,3, Tadaharu Yokokawa1, Jun Uzuhashi1, Tadakatsu Ohkubo1, 3 Hiroshi Harada1,4 and Shinsuke Suzuki2,3,5 4  5 1. National Institute for Materials Science (NIMS), 1-2-1 Sengen, Tsukuba, Ibaraki 305-0047, Japan  6 2. Department of Applied Mechanics and Aerospace Engineering, Waseda University, 3-4-1 Okubo, 7 Shinjuku-ku, Tokyo 169-8555, Japan 8 3. Department of Materials Science, Waseda University,3-4-1 Okubo, Shinjuku-ku, Tokyo 169-8555, 9 Japan 10 4. (Present address) Superalloy Design Laboratory, 3-18-33, Azuma, Tsukuba, Ibaraki, 305-0031, 11 Japan 12 5. Kagami Memorial Institute for Materials Science and Technology, Waseda University, Tokyo 169-13 0051, Japan 14  15 *Corresponding author 16 chihiro448@akane.waseda.jp (C. Tabata) 17  18   19 about:blank2  Abstract:  20  Ni-base superalloys are known for their high temperature oxidation resistance, however, impurity of 21 S at ppm order can significantly decrease the oxidation resistance of the alloy. Melting using a CaO 22 crucible is reported to improve the oxidation resistance, but Ca can negatively impact the oxidation 23 resistance of the alloys, depending on the amount and method of addition. The objective of this 24 research was to determine the effect of Ca addition using Al-Ca alloy, and to understand how Ca affects 25 the oxidation resistance of the alloy. Single crystal Ni-9.8 wt.% Al model alloys melted in an Al2O3 26 crucible were cast, with the addition of 20 ppm of S in all samples, and 0, 20, and 100 ppm of Ca using 27 Al-4.9wt.%Ca alloy. Ca addition improves the oxidation resistance of the samples, though it lacked in 28 efficiency in the improvement of this trait compared to when using a CaO crucible. Increasing the 29 amount of Ca addition led to the increase in the amount of CaS formation and desulfurization, though 30 the removal of slag is crucial. Although Ca was effective in terms of desulfurization, to limit the 31 negative effects of Ca itself, the formation of sulfides must be the dominating reaction.  32  33 (max 200 words)  34 Keyword: oxidation, sulfur segregation, Ca addition, CaS, single crystal superalloy 35   36 3  Ⅰ. Introduction 37  Ni-base superalloys are widely used in jet engines and other gas turbine systems, due to their excellent 38 mechanical properties and oxidation resistance [1]. Recent heightened awareness of establishing a 39 sustainable society requires the alloys with higher temperature capability to reduce CO2 emissions and 40 increase the thermal efficiency of turbines. For this reason, many rare metals are used in the alloys to 41 improve their mechanical properties. But due to the recent increase in the price of the rare metals, ways 42 to recycle and reuse materials are strongly required. Used turbine blades contain large amounts of S 43 contamination, which often comes from the fuels, inlet air, and the materials used to cast the alloys. 44 This degrades their properties, especially oxidation resistance, in high temperature environment [2]. S 45 is known to segregate at the oxide and substrate interface, decreasing the adhesion of oxides, resulting 46 in the spalling of oxides [3-7]. Even S at ppm level can have a detrimental effect on the alloy, therefore, 47 ways to remove and prevent the segregation of S are necessary [8,9]. 48  One way to desulfurize the alloys is to use CaO. Harada et al. and Utada et al. developed a method 49 using CaO crucible to melt turbine components made of Ni-base superalloys, removing impurities such 50 as S before casting [10,11]. Utada et al. proved that this method desulfurized the Ni-base superalloy 51 samples and improved the oxidation resistance, all the while maintaining its mechanical properties 52 [12]. Sugiyama et al. also used this method and found that the samples melted in a CaO crucible had 53 higher oxidation resistance compared to the ones melted in an Al2O3 crucible, even though the S 54 content of the two samples were about the same [13]. According to previous research, it was revealed 55 that the S segregation for the sample melted in CaO crucible was lower than the one melted in Al2O3 56 crucible [14]. Researches indicate that the CaO reacted with the Al within the melt, forming calcium 57 aluminates and Ca, and the excess Ca reacted with S to form CaS, preventing S from segregating to 58 the interface of the Al2O3 oxide layer and the Ni-base substrate [12,14,15]. In order to simplify the 59 reaction, similar tests were conducted for Ni-Al binary alloys. The usage of CaO crucible on Ni-Al 60 alloy also proved to be effective, and the mechanism and location for the formation of CaS has been 61 clarified [16]. According to Yokokawa et al., CaO pellets were also effective in the removal of S from 62 the melt [17]. On the other hand, CaO is known to easily hydrate, making it difficult to use CaO 63 crucibles at an industrial level [18,19]. Therefore, ways to improve this method are needed.  64 Although CaO improved the oxidation resistance of the alloys, excessive Ca addition and CaO 65 deposits from the different types of coals and geographic locations the turbines were used in, are known 66 to degrade the oxidation resistance of the superalloys. Azakli et al. reported that excessive amounts of 67 Ca decreased the oxidation resistance of the alloys, due to the formation of Ca-rich complex oxides 68 [20]. DeBarbadillo also found that excess Ca can form intermetallic compounds with Ni that are brittle 69 and moisture-sensitive, which resulted in severe hot tearing of the Ni-base alloy [21]. Reports on CaO 70 deposits showed an increase in the oxidation rate and degradation rate of the samples [22,23]. Jung et 71 al. also observed significant degradation during cyclic oxidation at 950°C for both coated and uncoated 72 superalloys by CaO deposits [24]. Yet, the difference in the mechanism behind the usage of CaO and 73 the Ca addition is not clear. Also, in contrast with the effect of mass addition of Ca as described above, 74 4  the effect of limited addition of Ca on the oxidation resistance of the alloys has not been investigated.  75  The objective of this research is to determine whether it is possible to desulfurize the alloys by the 76 addition of Ca at ppm level using Al-Ca alloy, and to understand how Ca affects the oxidation 77 resistance of the Ni-Al binary single crystal alloy. In order to simplify the reactions between the melt 78 and the crucible, and to compare the results with our previous research, Ni-Al binary single crystal 79 alloys were used for this research. The oxidation resistance of the samples at 1100°C were investigated, 80 and the reaction between the melt and the crucible were observed.  81  82 Ⅱ. Experimental Procedures  83 Ni-9.8 wt.% Al alloys were melted in an Al2O3 crucible at 1600 °C with the pressure kept under 84 6 × 10-2 Pa, using a directional solidification furnace with high frequency vacuum induction 85 heating. S was added using NiS powder that was inserted within a Ni tube, and Ca was added using 86 Al-4.9 wt.% Ca alloy. 20 ppm of S was added in all 3 samples, and 0, 20, and 100 ppm of Ca were 87 also added to each sample, which will be referred to as Ni-Al (0 ppm Ca), Ni-Al (20 ppm Ca), and 88 Ni-Al (100 ppm Ca). After the addition of S and Ca, the melt was kept for 15 min and was poured 89 from the crucible into a ZrO2-base mold kept at 1500 °C, and unidirectionally solidified into single 90 crystal round bars of 11 mm in diameter and 135 mm in length. The mold was withdrawn from the 91 heating chamber to the cooling chamber at 200 mm/h. “Starters” are placed at the bottom of the 92 mold, which solidifies the melt into columnar grained, polycrystalline alloys. The “selectors” exist 93 between the starters and the cylindrical sample to help solidify the melt into a single crystal alloy. 94 Solution-heat-treatment was performed at 1300 °C for 5 h, and aging heat-treatment was performed 95 at 870 °C for 20 h in vacuum using a high frequency induction heat treatment furnace [10]. 96 The chemical composition was analyzed using the same methods as our previous research [16]. For 97 Ni and Al, an inductivity coupled plasma optical emission spectrometer (ICP-OES, Aligent, ICP-OES 98 720-ES) was used, while glow discharge mass spectrometry (GD-MS, Nu Instruments LTC. 99 ASTRUM) was used for S and Ca. Oxidation testing samples were also prepared following the same 100 procedures as our previous research [16], where the samples were cut from the cylinders to be 5 mm 101 in height and 9 mm in diameter, with the surfaces finished using U.S. Grid #600 SiC abrasive paper.  102 The remaining Ni-Al alloy on the bottom sides of Al2O3 crucibles after pouring were analyzed in 103 order to understand what kinds of reactions occurred during the melting process. The samples were 104 embedded in resin and observed using electron probe micro analyzer with a field emission electron 105 gun (FE-EPMA, SHIMADZU, EPMA-8050G). Inclusions found within the starters were observed 106 using scanning electron microscopy (SEM) with energy dispersive X-ray spectroscopy (EDS) (ZEISS, 107 GeminiSEM300). Cyclic oxidation tests were performed at 1100°C for 100 cycles, with each cycle 108 containing 1 h of heating in air and 1 h of cooling in air, and the samples were weighed after several 109 cycles. Samples oxidized at 1100°C for 1 h were also prepared, and the oxide scales were observed 110 using SEM (CrossBeam 550, ZEISS) and aberration corrected scanning transmission electron 111 microscope (STEM, FEI Titan G2 80-200) with energy dispersive X-ray spectroscopy (EDS) at the 112 5  acceleration voltage of 200 kV. The inclusion found within the sample were also lifted out and prepared 113 using focused ion beam (FIB)-SEM dual beam system (FEI Helios G4UX). High-angle annular dark 114 field (HAADF)-STEM images and EDS elemental maps were taken by STEM, also selected area 115 electron diffraction (SAED) images were taken from TEM mode of STEM.   116  117 Ⅲ. Results 118  Fig. 1 shows the backscattered electron image and elemental maps taken using FE-EPMA for Ni-Al 119 (20 ppm Ca), close to the bottom side of the Al2O3 crucible. Areas that appeared black in the BSE 120 images were observed. According to the EPMA mapping and line scan results, Al and O were 121 detected in similar locations, which are most likely Al2O3. Ca was also observed in some of the same 122 parts as Al and O, indicating that Ca-Al-O slags were observed inside of the Ni-Al melt. S was 123 detected both within and around the slag, resulting in the removal of S from the alloy. Although 124 several regions of the alloy had S signal without associated Ca signal, it is uncertain whether 125 aluminum sulfides had formed or if Ca had existed previously. Fig. 2 also shows the backscattered 126 electron image and elemental maps taken using FE-EPMA, this time for Ni-Al (100 ppm Ca). 127 Compared to Fig. 1, large amount of Ca was observed in the same place as Al and O. S was also 128 detected, but more frequently and at a larger amount. This indicates that increasing the Ca content 129 resulted in the increase in the amount of desulfurization. 130  131 6  Fig. 1 (a) Backscattered electron image and (b) FE-EPMA elemental maps for S, Ca, O, and Al taken 132 from the remaining Ni-Al (20 ppm S, Ca) near the bottom of the Al2O3 crucible used to melt the 133 sample. The red rectangle in the backscattered electron image (a) indicates where the elemental maps 134 were taken. (c) Line profiles were taken for S, Ca, O, and Al, from the area shown in blue.  135  136  137 Fig. 2 (a) Backscattered electron image and (b) FE-EPMA elemental maps for S, Ca, O, and Al taken 138 from the remaining Ni-Al (100 ppm Ca) near the bottom of the Al2O3 crucible used to melt the 139 sample. The red rectangle in the backscattered electron image indicates where the elemental maps 140 were taken. (c) Line profiles were taken for S, Ca, O, and Al, from the area shown in blue. 141  142  In order to understand whether Ca addition could form CaS inclusions to suppress S segregation at 143 the interface between Al2O3 oxide layer and the substrate, as when using a CaO crucible, the starters 144 were observed using SEM-EDS in fig. 3 and fig. 4. What appears to be inclusions formed within the 145 voids, or inclusions promoting the formation of voids were found near the grain boundary and the 146 eutectic γ′ phase, most likely in the inter-dendritic regions, for both Ni-Al (20 ppm Ca) and Ni-Al 147 (100 ppm Ca) samples. The peaks of Ca and S were observed, indicating that CaS had formed, 148 similar to the results for melting using a CaO crucible [16]. For each sample, the width and the 149 height of 10 inclusions were measured, and the average size of the inclusions were calculated. For 150 7  Ni-Al (20ppm Ca), the inclusions were 379 nm in width and 338 nm in height average. For Ni-Al 151 (100 ppm Ca), the inclusions were 747 nm in width and 726 nm in height. The number of inclusions 152 found within an area of a certain size have not changed drastically. These results indicate that the 153 increase in the Ca content led to the increase in the size of the inclusions.  154  155  156 Fig. 3 (a) Angle selective BSE image and (b) SE image of the as-cast starter of Ni-Al (20 ppm Ca), 157 (c) the peak profile of the inclusion, taken using EDS, and (d) schematic diagram of the cast sample 158 to show the starter. 159  160  161 Fig. 4 (a) Angle selective BSE image and (b) SE image of the as-cast starter of Ni-Al (100 ppm Ca) 162 and (c) the peak profile of the inclusion, taken using EDS. 163  164  In order to determine whether the inclusions were actually CaS, the inclusion found within the 165 starter of Ni-Al (100 ppm Ca) was lifted out using FIB-SEM system and observed by STEM. Fig. 166 5(a) shows the cross sectional HAADF-STEM image and the EDS elemental maps of the inclusion. 167 Ca-S inclusion of about 300 nm in size was detected, with a formation of a void above. SAED 168 8  pattern was also taken from the inclusion and shown in Fig. 5(b). Judging from the SAED patterns 169 and the crystal orientations of the substrate fcc-Ni3Al, mismatched fcc-CaS crystals have formed, 170 similar to the ones detected when using a CaO crucible [16]. The inclusions likely form either at the 171 voids or promote the formation of voids.  172  173  174 Fig. 5 (a) Cross sectional HAADF-STEM image and EDS elemental maps of the inclusion found 175 within the starter of Ni-Al (100 ppm Ca). (b) Selected area electron diffraction (SAED) pattern taken 176 from the inclusion presented in (a). The electron beam was incident along the [110] direction for 177 CaS, and [22�1�] direction for Ni-Al.  178  179 To determine whether Ca addition had any effect on the oxidation resistance of the samples, cyclic 180 oxidation tests were conducted at 1100 °C. Fig. 6 shows the cyclic oxidation results of Ni-Al (0 ppm 181 Ca) and Ni-Al (20 ppm Ca). Note that Ni-Al (100 ppm Ca) was removed from the results, and the 182 reason for this will be explained in the discussion section. Results for Ni-Al without any S additions, 183 taken in our previous research, are also included as a reference [16]. Ca addition did result in the 184 increase of oxidation resistance, compared to the sample with the addition of S only. However, it did 185 not have as much effect on the cyclic oxidation resistance as the sample melted using a CaO crucible. 186 Table I shows the results of the chemical composition of both the 20 ppm Ca added sample and the 187 non-Ca added sample, both of which are done before the oxidation tests, using ICP-OES for Al and 188 9  GD-MS for S and Ca. The results indicate that the improvement in the cyclic oxidation was due to 189 both the decrease in the S content with the addition of Ca, and the suppression of the oxide spallation 190 by the formation of CaS. However, most of the Ca has not been detected within the sample and its 191 content was significantly lower than intended, since most of the Ca added had reacted with the Al2O3 192 crucible and did not remain within the alloy. There was also an increase in the Al content for both 193 samples, which likely came from the Al2O3 crucible. The Al2O3 crucible itself does not have the 194 ability to remove S from the alloy, therefore, the decrease in the S content is due to the addition of 195 Ca. The reaction between the Al2O3 crucible and the added Ca likely led to the formation of the 196 calcium aluminates slag, which is known to remove S from the alloy [12-16]. Although there is the 197 possibility of the formation of CaS directly after the addition, we believe most of the added Ca will 198 likely react with the Al2O3 crucible first, before forming CaS within the melt. This is likely due to the 199 order of the reactions that had occurred. As it will be shown in the discussion section by the Gibbs 200 free energy calculations, when S and Ca are added within the melt, oxides will form before the 201 sulfides. The decrease in the S content is most likely due to this reaction, leading to the increase in 202 the oxidation resistance observed in the cyclic oxidation test result in Fig. 6.  203  204  205 Fig. 6 Cyclic oxidation results for Ni-Al (0 ppm Ca) and Ni-Al (20 ppm Ca) at 1100 °C for 100 206 cycles. Results for Ni-Al melted in an Al2O3 crucible and a CaO crucible are also included as a 207 reference [16].  208  209 Table I Analyzed chemical compositions of the single crystal alloys used in this study (Ni bal.) 210 Sample Al (wt.%) S (ppm) Ca (ppm) 10  Ni-Al (Al2O3 crucible) [16] 9.80 2 - Ni-Al (CaO crucible) [16] 9.76 3 6 Ni- Al (0 ppm Ca) 10.0 23 - Ni-Al (20 ppm Ca) 10.1 13 0.2  211  Samples of Ni-Al (0 ppm Ca) and Ni-Al (20 ppm S, Ca) that had been oxidized at 1100 °C for 1 h 212 were lifted out using FIB, and the oxide scales were observed using in-lens secondary electron (IL-213 SE) SEM and STEM-EDS, as shown in Fig. 7. Figure 7(a) shows the results for Ni-Al (0 ppm Ca). 214 Although Al rich regions were detected, the Al2O3 scales are not continuous, and the scale width for 215 each layer of oxides was not even. On the other hand, the width of the oxide scales found in Ni-Al 216 (20 ppm Ca) was even, and continuous Al2O3 layers had formed, as shown in Fig. 7(b). Ca could not 217 be detected for both samples. The results suggest that Ca addition led to the removal of S from the 218 alloy, and the reduced S content led to the formation of continuous oxides. However, as it can be seen 219 in Table I, amount of S removed was not enough to fully prevent the oxide spallation. 220  221  222 Fig. 7 In-Lens secondary electron (IL-SE) SEM images, HAADF-STEM images, and EDS elemental 223 maps of the oxide scales of (a) Ni-Al (0 ppm Ca) and (b) Ni-Al (20 ppm Ca), which were oxidized at 224 1100 °C for 1 h.  225  226 IV. Discussion 227 11   Judging from the results in Fig. 1 to Fig. 5, larger amounts of Ca addition led to the increase in the 228 desulfurization of the melt and formation of CaS. It can be assumed that large amount of Ca addition 229 could lead to the increase in oxidation resistance of the sample. However, for this experiment, several 230 inclusions, which are most likely parts of the slag formed during the reaction between the Al2O3 231 crucible and the melt, had ended up within the metal. This resulted in the formation of cracks in the 232 starters, as shown in fig. 8. Presumably, part of the slag that had formed during the reaction between 233 the Al2O3 crucible and added Ca had entered the melt, and this may negatively affect the oxidation 234 resistance of Ni-Al (100 ppm Ca) sample. For this reason, the cyclic oxidation results for Ni-Al (100 235 ppm Ca) had been removed from Fig. 6. For this experiment, filters were not used when pouring the 236 melt into the cast, in order to prevent the Ni-Al melt from clogging up the filter. Filters can most 237 likely remove any sorts of slags that could get involved by the melt, preventing the Ca-Al-O-S slags 238 from getting in the metal, preventing the decrease in the oxidation resistance of the sample.  239  240 Fig. 8 SEM image of the cracks found within the as-cast starter of Ni-Al (100 ppm Ca).  241  242  Next, we will discuss the differences in the reactions for the usage of CaO crucibles for melting and 243 Ca addition. Figure 9 presents the schematic diagram of the reactions that had likely occurred in the 244 two processes. When considering the efficiency of melting using a CaO crucible or Ca addition, it is 245 crucial to consider the two kinds of reaction that had occurred. The first reaction is the 246 desulfurization by the reaction between the melt and the crucible. The second reaction is the trapping 247 of the S by the formation of CaS. One of the major differences between the effect of Ca from the Ca 248 addition and the usage of CaO crucible is where Ca is located initially. For the usage of the CaO 249 crucible (Fig. 9 (a)), Ca is located within the crucible. The amount of Ca within the crucible is 250 significantly larger than the amount of S within the melt. When using a CaO crucible, large amounts 251 of slag will form, which results in the highly efficient desulfurization of the melt. Enough Ca will 252 continue to be supplied within the melt by the reduction process of CaO, resulting in large amounts 253 of CaS formation. On the other hand, for Ca addition (Fig. 9 (b)), Ca is located within the melt, and 254 its amount is similar to or slightly larger than S. The amount of slag that can be formed is limited, 255 12  lowering the efficiency of the desulfurization process. Because the Ca within the melt reacts with the 256 crucible, the amount of Ca that can react with the remaining S within the melt will decrease 257 significantly, lowering the efficiency of CaS formation. Therefore, the efficiency of melting using a 258 CaO crucible is notably higher than Ca addition.  259  260 Fig. 9 Schematic diagram of the reaction between the melt and the crucible for (a) the usage of CaO 261 crucible (adapted from [16]) and (b) addition of Ca to the melt.  262  263  Another factor that needs to be considered is the order of the reaction that had occurred. When 264 melting using a CaO crucible, as mentioned in our previous research, the order of the reactions are as 265 follows [12,15,16,25,26]: 266 3CaO(within the crucible)+2Al(within the melt)=3Ca+Al2O3 (1) 267 CaO(from the crucible)+Al2O3(from previous reaction)=calcium aluminates (2) 268 Ca�from equation(1)�+S(within the melt)=CaS (3) 269 On the other hand, the reactions for Ca addition are most likely as follows: 270 Ca(within the melt)+O(within the melt or crucible)+Al2O3(within the crucible)=calcium aluminates (4) 271 Ca(remainders within the melt)+S(within the melt)=CaS (5) 272 The dominant reaction of Ca was with the Al2O3 crucible, since Al2O3 crucible on its own cannot 273 desulfurize the alloy, as it can be seen in table I, and the slight increase in the amount of Al is likely 274 the result of this reaction. With equation (4), the Ca within the melt reacts with the crucible to form 275 slags, which helps with the desulfurization process. But the remaining amount of Ca for equation (3) 276 is higher than equation (5). This is because for Ca addition, the oxygen found within the melt, the 277 crucible, or the furnace itself, reacts to form CaO or calcium aluminates before the formation of CaS. 278 The equations below demonstrate the Gibbs free energy (J/mol) of CaO, CaAl2O4, and CaS, taken 279 from the thermochemical database at around 1600 °C (1800K for CaAl2O4) [27-29]:  280 2∆𝐺𝐺𝐶𝐶𝐶𝐶𝐶𝐶0 = −852 × 103  J/mol (6) 281 ∆𝐺𝐺𝐶𝐶𝐶𝐶𝐴𝐴𝐴𝐴2𝑂𝑂40 = −168 × 104  J/mol (7) 282 2∆𝐺𝐺𝐶𝐶𝐶𝐶𝐶𝐶0 = −684 × 103  J/mol (8) 283 The formation energy is higher for oxides than for sulfides, thus explaining how the reaction to form 284 13  CaS is inefficient, unless specific measures are taken to ensure its formation. For the usage of the 285 CaO crucible, the reaction starts off with Ca at a stable state of CaO. The Al within the melt starts off 286 the reduction process, creating Ca. Although some of the Ca may react with the few ppm of O within 287 the melt, most of it will react with S. The initial structure and the reduction process likely led to the 288 efficient formation of CaS. On the other hand, when the initial structure is Ca, as demonstrated in 289 equations (6-8), oxides will form before sulfides. Therefore, only parts of S get desulfurized or form 290 CaS.  291 Increasing the amount of Ca may help with the efficiency of the reaction, but it will also increase 292 the formation of other oxides such as CaO or CaAl2O4. The reason other Ca addition and CaO 293 deposits reduced the oxidation resistance for the alloys was due to the formation of these oxides 294 [20,22,23]. Therefore, to use Ca addition for the desulfurization process or the prevention of S 295 segregation at the interface of oxides and substrate, it is important to find the balance between the 296 amount of Ca and S within the melt and understand the order of the reaction that is likely to occur. 297 On the other hand, using stable oxides such as CaO in the beginning will lead to the reduction 298 process since, according to the calculations in equations (6-8), it is thermodynamically beneficial to 299 form calcium aluminates. This likely requires the formation of Al2O3 beforehand, which can be 300 achieved by the reduction of CaO [12-16]. The formation of Ca from this process will most likely 301 initiate the formation of sulfides. This process is presumably the same for Ni-base superalloys as 302 well.  303  304 Ⅴ. Conclusion 305  To conclude, the following has been made clear about the effect of Ca addition on the oxidation 306 resistance of Ni-Al single crystal alloy. 307 1. Ca addition succeeded in the desulfurizing of the alloy and the forming of CaS. By increasing the 308 amount of Ca, the amount of S that has been removed and the CaS that had formed increased as 309 well. But with the increase in Ca addition, parts of the slags that had formed were involved within 310 the melt, negatively affecting the oxidation resistance of the sample. The usage of filters may help 311 solve this problem.  312 2. Ca addition of 20 ppm improved the oxidation resistance of the alloy, compared to Ni-Al (0 ppm 313 Ca). But it has not improved as much as the sample melted in the CaO crucible. This is likely due 314 to the amount and the order of the reaction of Ca that had reacted to desulfurize and prevent the 315 segregation of S by the formation of CaS. Increasing the amount of Ca addition increased the 316 amount of desulfurization and formation of CaS, but slag formation had also increased as well.  317 3. The dominating reaction for Ca addition was the formation of oxides. In order to efficiently 318 desulfurize, form CaS, and prevent the negative effect of Ca on the oxidation resistance, the 319 formation of sulfides has to be the dominating reaction. Using CaO will help by initiating the 320 reduction of CaO to produce Ca, which will react with the S within the melt.   321  322 14  Acknowledgments  323  This research was financially supported by the Council for Science, Technology and Innovation 324 (CSTI), Cross-ministerial Strategic Innovation Program (SIP), “Materials Integration for revolutionary 325 design system of structural materials” (Funding agency: JST). The authors would like to express our 326 gratitude to Dr. Makoto Osawa for the helpful discussions. We also would like to thank Ms. Kyoko 327 Suzuki for the support of microstructural investigation, and Mr. Yuji Takata for the preparation of the 328 single crystal alloys.  329  330 Conflict of Interest 331 On behalf of all authors, the corresponding author states that there is no conflict of interest. 332  333 References  334 [1] K. Harris, G.L. Erickson, S.L. Sikkenga, W.D. Brentnall, J.M. Aurrecoechea, and K.G. Kubarych: 335 Superalloys 1992, 1992, pp. 297-306. 336 [2] A.W. Fundenbusch, J.G. Smeggil, and N.S. Bornstein: Metall. Mater. Trans. A, 1985, vol. 16A, pp. 337 1164-66. 338 [3] Y. Ikeda, M. Tosa, K. Yoshihara, and K. Nii: ISIJ International, 1989, vol. 29, pp. 966-72.  339 [4] M. A. Smith, W. E. Frazier, and B. A. Pregger: Mat. Sci. Eng. A, 1995, vol. 203, pp. 388-98. 340 [5] J. L. Smialek, and B. A. Pint: Mat. Sci. Forum, 2001, vols. 369-372, pp. 459-66. 341 [6] N. Birks, G.H. Meier, and F.S. Pettit: Introduction to the High-Temperature Oxidation of Metals, 342 2nd ed., Cambridge University Press, New York, 2006, pp.144-48. 343 [7] E. Fedorova, D. Monceau, and D. Oquab: Corr. Sci., 2010, vol. 52, pp. 3932-42. 344 [8] J. Smialek: Crystal, 2021, vol. 11, pp. 60-71. 345 [9] J. L. Smialek, Oxid. Met., 2022, vol. 97, pp. 1-50.  346 [10] H. Harada, K. Kawagishi, T. Kobayashi, T. Yokokawa, M. Osawa, M. Yuyama, S. Suzuki, Y. Joh, 347 and S. Utada: US Patent, patent number 10689741 (2020.06.23) 348 [11] S. Utada, Y. Joh, M. Osawa, T. Yokokawa, T. Kobayashi, K. Kawagishi, S. Suzuki, and H. Harada: 349 Superalloys 2016, 2016, pp. 591-9. 350 [12] S. Utada, Y. Joh, M. Osawa, T. Yokokawa, T. Sugiyama, T. Kobayashi, K. Kawagishi, S. Suzuki, 351 and H. Harada: Met. Mat. Trans. A., 2018, vol. 49A, pp. 4029-41. 352 [13] T. Sugiyama, S. Utada, T. Yokokawa, M. Osawa, K. Kawagishi, S. Suzuki, and H. Harada: Met. 353 Mat. Trans. A., 2019, vol. 50, pp. 3903-11. 354 [14] C. Tabata, K. Kawagishi, J. Uzuhashi, T. Ohkubo, K. Hono, T. Yokokawa, H. Harada, and S. 355 Suzuki: Scr. Mater., 2021, vol. 194, 113616. 356 [15] Y. Kishimoto, T. Kono, T. Horie, T. Yokokawa, M. Osawa, K. Kawagishi, S. Suzuki, and H. 357 Harada: Met. Mat. Trans. B., 2021, vol. 52B, pp. 1450-62. 358 [16] C. Tabata, K. Kawagishi, J. Uzuhashi, T. Ohkubo, T. Yokokawa, H. Harada, S. Suzuki: Met. Mat. 359 Trans. A., 2022, vol. 53, pp. 2452-8. 360 15  [17] T. Yokokawa, H. Harada, K. Kawagishi, M. Sakamoto, M. Osawa, Y. Takata, M. Yuyama, T. 361 Kobayashi, T. Sugiyama, and S. Suzuki: Superalloys 2020, TMS, Seven Springs, Champion, PA, 2020, 362 pp. 400-7. 363 [18] T. Xu, Y. Su, T. Shi, and X. Zhang: Ceram. Int., 2021, vol. 47, pp. 2165-71. 364 [19] M. Chen, N. Wang, J. Yu, and A. Yamaguchi: J. Eur. Ceram. Soc., 2007, vol. 27, pp. 1953-9. 365 [20] Y. Azakli, K. O. Gunduz, S. Cengiz, Y. Gencer, and M. Tarakci: Oxid. Met., 2021, vol. 95, pp. 366 135-56. 367 [21] J. J. deBarbadillo: Superalloys 1976, Seven Springs, Champion, PA, 1976, pp. 95-107. 368 [22] T. Gheno, G. H. Meier, and B. Gleeson: Oxid. Met., 2015, vol. 84, pp. 185-209. 369 [23] K. T. Chiang, G. H. Meier, and R. A. Perkins: J. Mater. Energ. Syst., 1984, vol. 6, pp. 71-86. 370 [24] K. Jung, F. S. Pettit, and G. H. Meier, Mater. Sci. Forum, 2008, vol. 595-598, pp. 805-12. 371 [25] R. J. Fruehan: Metall. Trans. B, 1978, vol. 9B, pp. 287-92. 372 [26] T. Degawa and T. Ototani: Tetsu-to-Hagané, 1987, vol. 73, pp. 1684-90. 373 [27] J. F. Elliot and M. Gleiser: Thermochemistry for Steelmaking, vol. 1, Addison-Wesley, 1960. 374 [28] S. Nagasaki: Metals Databook 3rd edition, Japan Institute of Metals and Materials, Japan, 1993, 375 pp. 96-7.  376 [29] I. Barin: Thermochemical Data of Pure Substances, 3rd ed., VCH Verlagsgesellschaft mbH, 377 Weinheim and VCH Publishers, Inc., New York, NY, 1995. 378  379