Increase of Austenitic Ductile Iron type D5S durability by high temperature pre-treatment

1. Abstract In the present work, a performance of ASTM A439 Austenitic Ductile Iron type D5S at high temperature in the oxidizing environment was investigated. The obtained results revealed that exposure at temperatures 800°C, 850°C and 900°C resulted in relatively high mass gain and an extensive oxide scale spallation from the samples’ surfaces during cooling. On the contrary, the material exposed at 950°C revealed a better oxidation resistance and no oxide scale spallation. The material exposed at 1000°C showed the best oxidation resistance among the studied samples. The surfaces and cross-sectional investigation revealed that the material exposed at 950°C formed mostly Ni/Cr/Mn-mixed protective oxide scale and local formation of Fe-rich nodules. In comparison with the sample exposed at 1000°C, a smaller amount of Fe-rich nodules per area unit was observed and most of the surface was covered by Ni/Cr/Mnmixed protective scale. The latter was explained by the change in the calculated diffusion coefficients in the alloy for Ni and Fe, namely up to 900°C the diffusion coefficient for Fe was much higher than for Ni, while above 900°C the diffusion coefficient for Ni becomes higher than for Fe. This phenomenon was correlated with a phase transformation from α-Fe into γ-Fe resulting in the diffusion coefficient change.


Introduction
The automotive turbine housings are components used for collecting exhaust gases generated from individual cylinders, directing them to the turbine wheel, and transfer further into the exhaust system. The turbine housing is subjected to the highest loads, partly due to increased mass flow of exhaust gases and partly to additional loads from the weight of the turbocharger. Raw turbine housing oxidation resistance is relevant to the quality of the protective oxide scale, created on its inner surface when corrosive exhaust gases pass through the component. During thermal cycling caused by daily operation of the car engine, stresses arise in the scale, increasing the risk of oxide spallation. Continuous oxidization of material results microscopically in depletion of oxide-forming elements and macroscopically consumption of the metal, leading to degradation of component mechanical properties. It is therefore important that the formed oxide scale is protective, adherent, and homogeneous. Then, the formation of an Al2O3 or Cr2O3 based oxide scale is the most desirable. Surface corrosion also negatively influences the component fatigue life, since simultaneous crack growth and oxidation is known to be detrimental [1,2,3,4]. Therefore, the turbine housing high durability at high-temperature arises.
Depending on the exhaust gas temperature (usually 700 -1000°C), the materials available for turbine housings include ferritic and austenitic ductile cast iron alloys and cast stainless steel [5]. Ferritic alloys have the benefit of higher thermal conductivity and a lower coefficient of thermal expansion compared with austenitic alloys. However, the austenitic alloys, on the other hand, are characterized by higher strength at high temperatures, making them a better choice for high-temperature applications due to precipitation hardening [6,7].
The Ni-containing ductile irons belong to a group of austenitic ductile cast irons. The alloy commonly used in turbine housings is the austenitic ductile iron D5S, also referred to EN-GJSA-XNiSiCr35-5-2 or 5.3505, containing the highest Ni content, responsible for austenitic structure in ambient temperature and improving high-temperature strength, toughness and creep properties [8]. Relatively high silicon and chromium contents provide high mechanical properties and erosion resistance by the formation of carbides and self-protective oxide scale [9]. D5S turbine housings operate close to their fatigue and oxidation resistance limits. However, even with superior overall properties, the application of this material is still limited by its rather high cost compared with the other possible alternative materials [9,10].
Ductile iron was already investigated in terms of its mechanical and fatigue properties at temperature range 20 to 1000°C [11,12]. Also, several attempts at descriptions of oxidation resistance of family of ductile iron, including D5S alloy, at high temperature was already performed [13,14,15,16]. However, all the above-mentioned studies for the oxidation resistance were performed at a temperature up to 800°C. Celik et al. [13] found the best oxidation resistance against air oxidation at 650°C for AISI 409 stainless steel. However, the cast iron SiMo showed better performance at high temperatures in comparison with GGG40. Scheidhauer et al. [14] observed a similar oxidation behavior during air exposure at 820°C for ferritic SiMoAl in comparison with D5S alloy. Ferritic SiMoAl was proposed as a cheaper alternative for D5S alloy. Brady et al. [15] observed a better oxidation resistance for D5S than SiMo alloy during exposure at 650°C in air + 10% H2O. Moreover, the extensive oxide scale spallation was found during exposure of D5S at 700°C, while for SiMo formed oxide scale remained adherent. Ekstrӧm et al. [16] investigated the effect of Cr and Ni addition to SiMo51 alloy. They found that the addition of Cr in the concentration ranging between 0.5 -1 wt.% improves the oxidation resistance during exposure at temperature 700 -800°C. On the other hand, a negative effect of Ni addition on the oxidation resistance of alloy was observed.
It is known that the increase the operating temperature results in an increase the efficiency of the turbines [17]. Therefore, the study aims is to investigate the oxidation resistance of material typically used for turbine casing, namely ductile iron D5S for the first time at a temperature range 800-1000°C.

Experimental
In the present work high-temperature oxidation behavior of ductile iron D5S, with the composition given in table 1, was investigated. The chemical composition measurement performed using arc spark optical emission spectrometer SPECTROMAX revealed, that the element contents are in the range of nominal compositions. However, one should notice that the OES measurement revealed the presence of 0.19 wt% of Mg in the alloy. Rectangular shaped specimens with the dimensions of 12 x 20 x 2 mm samples were cut and subsequently ground using SiC sand-papers till 1200 grit. After grinding, all samples were ultrasonically cleaned in acetone. Then the samples were oxidized in thermogravimetrical furnace Xerion XTUBE for 48 hours in air at five different temperatures, namely 800, 850, 900, 950 and 1000°C. The samples after exposure were investigated by laser profilometer Sensofar S-Neox Non-contact 3D Optical Profiler with a vertical resolution of 1 nm and scanning electron microscopy (SEM) ZEISS Field-Emission Gemini SEM 300 equipped and correlated with Bruker Quantax EDS detector was used for SEM/EDX analyses. Images were taken with the use of backscattered electrons (BSE) detector under different magnifications and 20kV acceleration voltage. EDS detector enabled chemical composition analyses in micro-areas (chemical composition analysis, elements concentration mapping, line analyses). After surface investigation, samples were covered by a thin gold layer, subsequently electroplated with nickel and mounted in epoxy resin. Metallographic cross-sections were prepared by a series of grinding and polishing steps with the final step of polishing using the SiO2 suspension with 0.25 μm granulation. The cross-sections were investigated by light optical microscope (LOM) Nikon ECLIPSE MA200 Inverted Metallurgical Microscope and scanning electron microscope (SEM). The mass change curves were created by using Xerion X-tube thermogravimetrical furnace. The mass change curves were obtained "in-situ" during oxidation. The mass of the samples was measured and recorded continuously during exposure. After exposure the mass change plots were corrected considering the so-called Archimedes effect. Considering the character of isothermal thermogravimetrical oxidation test, only one sample was tested during one campaign (i.e. one sample was exposed during one run at certain temperature). However, to check and ensure repeatability of obtained results, a second run of oxidation test at the same temperature was performed. This means that per each temperature two samples were investigated. The obtained results showed no discrepancies, therefore only one result per each exposure temperature was shown in the present work. The information about specimens' surface topography (textures and profiles) was obtained using Sensofar S neox optical 3D profiling microscope by interferometry measurement technique. Nikon 10X DI objective lens were used. Image post processing was carried out using Mountains® surface texture metrology software.

Material in the as-received condition
The microstructure of the studied alloy is shown in Figure 1. The microphotograph revealed the presence of spheroids of the graphite within the austenitic matrix. The graphite nodules did not exhibit perfectly round shapes but with sharp edges with diameters about 20 µm. Additionally, the interdendritic mixture of chromium carbides and silicide particles was observed (Figure 2 a). Moreover, the eutectics were observed in the metallic matrix. In the eutectic co-enrichment of Ni, Cr, Si, Mg and Mn elements were observed in the form of precipitates in the Fe-rich matrix (Figure 2b).   Figure 3 shows the mass changes obtained for the alloy investigated at various temperatures. The mass change obtained at the end of exposure at 800°C (48 h) was equal to 8 mg·cm -2 . The increase of temperature from 800 to 850°C resulted in an increase mass change to 10 mg·cm -2 and further temperature increase to 900°C resulted in an increase mass change up to 12.5 mg·cm -2 . Despite the relatively high mass change, also extensive spallation of the oxide scale during cooling outside the furnace was observed for all samples exposed at the abovementioned temperature. Surprisingly, increasing exposure temperature up to 950°C resulted in a much lower mass change of studied material as compared to exposure at a lower temperature. For the alloy exposed at 950°C obtained mass change after 48 hours was equal to 3 mg·cm -2 . Further increasing of exposure temperature to 1000°C resulted in a further decrease in obtained mass change at the end of the test, and was equal to 1 mg·cm -2 . Moreover, no oxide scale spallation during cooling of studied material after exposure at 950 and 1000°C was observed, in both cases oxide scale remained adherent. Since the extensive oxide scale spallation at 800°C, 850°C and 900°C was observed, only specimens exposed at 900°C, 950°C and 1000°C were further investigated in detail. Figure 3 Mass change curves obtained during air oxidation of studied alloy for 48 hours at various temperatures. For the samples exposed at temperature marked with stars an intensive oxide scale spallation during cooling was observed.

Surface investigation
The SEM/SE images of the surfaces of studied material after exposure at 900°C (Figure 4a), 950°C ( Figure 4b) and 1000°C (Figure 4c) are shown. The SEM/SE image of the surface of the sample exposed at 900°C revealed the presence of light areas identified as bare metal was observed. Moreover, there were visible crushed oxides remnants in many places. This was the result of an oxide scale spallation during the cooling stage. The SEM/SE images of the surfaces of samples exposed at 950°C and 1000°C showed that the material tested at both temperatures formed similar oxide scale morphology, namely flat areas and oxide islands were observed. The SEM/EDS analysis showed that the flat surfaces consist of Ni/Cr/Mn-rich oxide, whereas the oxide islands were identified as Fe-oxides. Such elemental distribution was confirmed by SEM/EDS elemental maps performed for material exposed at 950°C ( Figure 5) and 1000°C ( Figure 6). However, it is important to point out that the amount of Ferich oxide nodules was far less on the surface of material exposed at 1000°C (Figure 4c) compared to the sample exposed at 950°C (Figure 4b). The higher amount of Fe-rich nodules per unit area for sample exposed at 950°C than 1000°C was additionally confirmed by the three-dimensional reproduction by the laser profilometer shown in Figure 7.     Figure 8 shows SEM/SE images of the studied alloy surfaces after air exposure for 48 hours at 900°C (a, b), 950°C (c, d) and 1000°C(e, f) taken at low (a, c, e) and high ((b, d, f) magnification. As shown in Figure 8 a, and b, only the internal oxidation zone is visible. Differently, there are regions where a thin oxide scale formed and regions where relatively thick nodules were present on the cross-sections of samples exposed at 950°C and 1000°C (Figure 4 c-f). However, similarly to observed on surfaces, also a smaller number of Fe-rich oxide nodules were present in the cross-sections in the material exposed at 1000°C (Figure 8 e) in comparison with material exposed at 950°C (Figure 8 c). The SEM/EDS elemental map ( Figure 9) revealed that the region of internal oxidation of material exposed at 900°C is enriched in Ni and Si. It could be assumed that Si and Ni were internally oxidized. The SEM elemental maps obtained for samples exposed at 950°C ( Figure 10) and 1000°C (Figure 11) showed the formation of the oxide scale with a similar microstructure. Namely, in both cases, regions with a thin continuous oxide scale rich in Ni, Cr, Mn, and regions with Fe-rich nodules were observed. However, contrary to observations on the sample exposed at 900°C, the formation of continuous sub-scale enriched in Ni, Cr and Mn at the bottom of the nodule were found. Additionally, in the region of internal oxidation, an enrichment in Ni and Si were present as well. The formation of Ni/Cr/Mn-rich sub-layer was responsible for better oxidation resistance at 950°C and 1000°C, which was indicated by lower mass changes, better oxide scale adhesion and a smaller number of Fe-rich nodules.

Discussion
As shown above, an increase of exposure temperature from 800°C to 850°C and further to 900°C increased oxidation kinetics (expressed by the mass change). The relatively high mass change resulted in oxide scale spallation during samples cooling. Considering relatively high mass changes obtained for the mentioned temperatures, it can be assumed that the thickness of the formed external oxide scale was higher than the critical oxide scale thickness [10]. The latter resulted in stress accumulation and at the level for which stress relaxation is done by the cracks nucleation and propagation due to crossing so called "critical oxide scale thickness" [18]. Since the difference in coefficients of thermal expansion (CTE) for the oxide scale (ceramic material) and the metallic substrate is relatively high, the stresses near the oxide scale/substrate interface occurs during cooling [19]. At some point, the stress level was so high that the system can relax only by the cracks initiation and propagation. The latter resulted in an extensive oxide scale spallation. However, during exposure conducted at 950°C and 1000°C, no oxide scale spallation was observed. Moreover, the amount of Fe-rich nodules decreased with increasing temperature from 950°C to 1000°C. The better oxidation resistance was correlated with the formation of Ni/Cr/Mn-mixed oxide scale on larger areas ( Figures 5, 6 and 7) accompanied with the formation of Cr/Mn-mixed subscale beneath the Fe-rich oxide scale (Figures 10, 11). The question is what happened in studied material between 900°C and 950°C that increased better oxidation resistance of the material. It is known that for pure Fe increase of temperature results in a phase transformation from α-Fe into γ-Fe (at 912°C) [20]. Taking into account the phase transformation and observed differences in the oxidation resistance of studied material, a calculation of diffusion coefficients in the alloy was performed. The diffusion coefficients were approximated based on the literature data from Arrchenius plots. The diffusion coefficients were estimated as: for Ni the D0(10 -4 m 2 s -1 ) = 1.27 and Q (kJ mole -1 ) = 279.7 [21] and Fe for temperature range 1067-1169K: D0 = 121 and Q = 281.6 [22] and for the temperature range 1223-1473K: D0 = 4,08 and Q=311,1 [23]. The calculations results revealed that up to 900°C the diffusion coefficient of Fe in the alloy was much higher than for Ni, then at this temperature range formation of Fe-oxide was more likely expected. However, above 900°C, the calculated diffusion coefficient for Fe in the alloy become lower than the calculated diffusion coefficient for Ni. Considering relatively high Nicontent in the studied alloy at the level of 35.5 wt% and higher diffusion coefficient for Ni than for Fe at a temperature above 900°C, then the formation of a more protective oxide scale, namely Ni-rich oxide was expected.

Summary and conclusions
In the present work, a ductile iron used as a material for the automotive turbine core was studied at elevated temperatures. Since the efficiency of the turbine depends on temperature, then the temperatures used in the present work were far higher than the application temperature. The obtained results allowed to formulate the following conclusions: -Exposure of the ductile iron at a temperature range 800-900°C resulted in relatively high oxidation kinetics and an extensive oxide scale spallation, -Further increase in exposure temperature to 950°C and 1000°C surprisingly resulted in lowering in the oxidation kinetics, -The reason of increased oxidation resistance and better oxide scale adherence was correlated with more prone formation of Ni/Cr/Mn-mixed oxide scale, lower amount of Fe-rich nodules and formation of Cr/Mn sub-scale, -As the reason of the increase in oxidation resistance of ductile iron D5S a phase transformation of Fe at 912°C resulting in change in diffusion coefficients in the alloy for Fe and Ni was concluded, namely the calculated diffusion coefficient in D5S alloy up to 900°C (in α-Fe) the diffusion coefficient for Fe was much higher than for Ni, while at 950°C and 1000°C (in γ-Fe) the opposite situation was found, i.e. diffusion coefficient for Ni is higher than for Fe. The latter is claimed to be the main reason of improvement in oxidation resistance of D5S during air exposure at a temperature above 900°C. -The results obtained in the present work are very promising and can be applied in the real automotive turbine cores as a pre-heat treatment to form a more protective oxide scale and the same suppress the sites for cracks nucleation. However, the latter needs to be investigated as planned in the future.

Acknowledgment
Project financed by the National Center for Research and Development (Narodowe Centrum Badań i Rozwoju) under the TECHMATSTRATEG2/406725/1/NCBR/2020 program.

Data Availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.