Introduction

Nickel-based single-crystal (SX) superalloys are widely used materials in the production of turbine blades for the aerospace industry due to their excellent mechanical properties (fatigue, creep) at high temperatures [1]. Nevertheless, the service life of these components is often limited by the hot corrosion behaviour of uncoated gas turbine components owing to harmful conditions such as the presence of sodium sulphate (Na2SO4) [2,3,4]. Indeed, Na2SO4 is the most representative salt deposit of this degradation phenomenon and is linked to the reaction of sodium chloride (NaCl) from marine environments ingested by the engine with sulphur oxides (SO2, SO3) formed during the combustion of kerosene or the overflight of industrial areas. Different degradation mechanisms can then be identified, depending on the specific temperature range, such as pitting (Type II) or relatively homogeneous attacks (Type I) that can lead to the appearance of cracks and the reduction of the metal cross-section [5]. These two corrosive modes are kinetically characterized by an initiation stage (slow degradation) and a propagation stage (acceleration of corrosion) [6]. While Type I often occurs once Na2SO4 is molten [7], Type II hot corrosion is observed at temperatures below the melting temperature of Na2SO4 (Tm = 884 °C). As described by Meier [8] and Luthra [9, 10], the materials degrades by reaction between the salt deposited in solid form and the oxide layer formed and requires a sufficient amount of SO3 that Wang et al. reported to be as low as 100 ppm for the corrosion of NiAl by Na2SO4-K2SO4 at 700 °C [11]. The critical SO3 content may vary depending mostly on the substrate as recently reported by Malacarne et al. [12] provided that the SO2 + 1/2 O2 reaction leading to SO3 is efficiently catalysed by e.g., Pt or Fe catalysts or the active surface of the Na2SO4 solid itself or other elements of the alloy [13]. In Type II hot corrosion, the oxides like NiO or CoO react first to form their respective sulphates and then, eutectic phases by reaction with solid Na2SO4 [14]. This lowers the melting temperature to 671 °C and 578 °C for Na2SO4-NiSO4 and Na2SO4-CoSO4 mixtures, respectively [15, 16]. However, the influence of Na2SO4 without SO3 has almost never been discussed before, even though J. Stringer reported that Type I can be observed at temperatures as low as 750 °C, and perhaps lower [17] but the reasons for this to occur were not developed although it is known that SO2 (leading to SO3) is not permanently present in the hot gas coming out of the combustion chamber or in the purged cooling air of the compressor. Therefore, this study intends to elucidate whether hot corrosion occurs at such low temperatures by comparing with air oxidation on a 1st generation AM1 nickel-based single-crystal superalloy. Two different salt deposits are applied with the goal of determining the exclusive contribution that the Na2SO4 salt may exert. A comparison is also established between the As-Cast (AC) and fully homogenized (FHT) microstructures to highlight the chemical segregation effects.

Experimental Procedure

The AM1 superalloy, whose chemical composition is shown in Table 1, was supplied by Safran Aircraft Engines (SAE). This first-generation nickel-based monocrystalline superalloy was cast in the form of rods approximately 20 mm in diameter in the direction < 001 > [18, 19]. After solidification, some of the material in As-Cast (AC) AM1 underwent a full heat treatment (FHT). This additional step includes solution treatment at 1300 °C for 3 h, followed by ageing at 1100 °C for 5 h, and then ageing at 870 °C for 16 h, with air quenching between each processing step. The resulting materials have a specific γ/γˈ microstructure comprising dendrites, interdendritic spaces and eutectic pools. This characteristic is clearly observed in the case of the AC material with more irregular and coarser γˈ precipitates in the interdendritic zone. However, the structural morphology of the FHT material is more difficult to discern due to homogenization to reduce chemical segregation and eutectic pools to optimize mechanical properties [20,21,22].

Table 1 Chemical composition of the AM1 single-crystal nickel-based superalloy (data provided by Safran Aircraft Engines)
Full size table

The rods were then cut perpendicular to the direction of solidification to obtain circular specimens approximately 1 mm thick and 10 mm in diameter. Before exposure, each sample was gently and gradually polished with SiC sandpaper up to grade P1200 in order to reduce the residual stresses induced during cutting. The specimens were then cleaned with acetone and ethanol in an ultrasonic bath before being weighed with a ± 10–5 g accurate balance. For the corrosion tests only, two separate deposits of sodium sulphate (1.0 ± 0.3 and 3.0 ± 0.3 mg/cm2) were evenly sprayed onto the surface of AM1 from a saturated aqueous solution. This deposit was renewed every 24 h according to the ISO 17224 standard [23]. The oxidation and corrosion tests were performed in a Carbolite XCWF 1300X muffle furnace and a Pyrox resistive furnace, respectively, at an isothermal temperature of 750 °C under air (1 atm.) at varying times (5, 30 min, 1, 10, 24, 48, 96, 144, and 240 h). To increase the reproducibility of the experiments, two samples were introduced for all tests. After exposure, the samples were examined by Leica DMRM optical microscope to analyse the surface and cross-sections at different magnifications. The composition and crystal structure of the oxidized phases were obtained by combining the use of Raman spectroscopy (Jobin–Yvon Labram HR Evolution, λ = 532 nm), X-ray diffraction, XRD (Bruker AXS D8, λ = KαCu) and fluorescence spectroscopy (Bruker Tornado M4, λ = KαRh). Further detailed analyses were performed by Energy Dispersive Spectrometry (EDS) with an EDAX detector coupled to a FEI Quanta 200-F Scanning Electron Microscope (SEM) operating systematically at 20 kV to establish comparisons between the potential different thicknesses of the corrosion products. The cross-sections were made by mounting the samples in a phenolic resin and then polishing them to a 1 μm finish using a diamond paste. Special care was taken to prepare the cross-sections with waterless polishing and cleaning with absolute ethanol to prevent the dissolution of corrosion products. To reveal the microstructure of the substrates, an oxalic acid H2C2O4 (10% vol.) etching was performed at 4 V for a few seconds. The positive and negative poles of the electrode are in contact with the metal and the oxalic acid solution, respectively.

Results and Discussion

Isothermal Oxidation

Figure 1-a shows the specific mass change (Δm/S) with its standard deviation as a function of time at 750 °C for AC AM1 and FHT AM1. Both curves increase rapidly in the early stages of oxidation before slowing down and stabilizing at 10 and 24 h, respectively. However, there is a difference with a slightly more significant increase in mass for AC AM1 after 240 h of oxidation, indicating a more pronounced oxide growth. To better detail these variations, the linear kl and parabolic kp oxidation constants were calculated using the complete law [24]. The values presented in the table embedded in Fig. 1-a consistently confirmed previous observations, revealing higher linear and parabolic rate constants for the AC material. The Arrhenius plot of the kp values vs. the reciprocal temperature (Fig. 1-b) clearly show that the parabolic rate constant of the AC and FHT AM1 superalloys tends to decrease with decreasing temperature and that the difference in kp between AC AM1 and FHT AM1 is more pronounced at 750 °C than at higher temperatures [22]. By comparing with the work of Brumm and Grabke for NiAl with and without Cr [25], our experimental kp values fall between those for the metastable growth of Al2O3 (γ-Al2O3) and those for the stable phase of Al2O3 (α-Al2O3). Since the kp value of AC AM1 is closer to that of the γ-Al2O3, one may speculate this is the major oxide. In contrast, a comparable share of the γ-Al2O3 and α-Al2O3 phases in FHT AM1 can be assumed because the kp value lies between those of corresponding oxides. Yet, it will be shown later (Fig. 4) that no γ-Al2O3 is formed. Instead, the metastable θ-Al2O3 and the stable α-Al2O3 polymorphs grow on different locations of the alloy substrates. After a 240-h exposure period at 750 °C in air the formation of transient alumina phases is expected to predominate according to Garriga-Majo et al. [26]. However, the formation of the stable α-Al2O3 can be explained by the presence of Cr in the superalloy, which acts as a third element to promote the formation of α-Al2O3 [22, 25, 27].

Fig. 1
figure 1

a Mass gain per unit area of AM1 oxidized at 750 °C in air for 240 h as a function of time with a table gathering the different linear and parabolic domains (represented on the graph by the coloured rectangles) and b the graphical representation of Arrhenius' law integrating the parabolic kinetic constant kp of AM1 oxidized in air at 750 °C for 240 h and various relative data from the literature, i.e., the isothermal oxidation of AC AM1 and FHT AM1 [22], the polymorphs Al2O3 of NiAl [25] and the Cr2O3 of Ni-30Cr [28]

Full size image

The microstructural impact on oxidation can be readily shown in Fig. 2. Indeed, the surface images in Fig. 2-a show that the heterogeneous oxide growth occurs mostly during the early stages of oxidation (between 5 min and 1 h) and essentially at the dendrite level for AC AM1. This microstructural inhomogeneity is naturally not observed at the surface of the FHT AM1. In all cases, the scales are believed to be very thin because the polishing lines and the dendritic and interdendritic areas are readily shown. Figure 2-b gathers the EDS surface analyses on the different zones of the two materials. Except for the predominant contribution of oxygen (excluded in Fig. 2-b to account only for the metal contribution), the scales are rich in Ni, Cr and Al rather than with Co, Ti and Ta although the signal of the underlying substrate shall not be neglected given the very thin scales formed. Yet, the corresponding oxides of the main metallic elements were also observed by XRD (NiO) (not shown) and by Raman spectroscopy (NiO and (Cr,Al)2O3) (shown later in Fig. 5). However, since Cr has been shown to segregate further in the dendrites than in the interdendritic areas of the AC AM1 superalloy [22], it appears unclear at this stage why the oxides would be richer in Cr in the interdendritic zones. It can be hypothesized that a thicker layer of Ni oxide hinders Cr detection at the dendrite level. Alternatively, the oxide scale is thin on the interdendritic area, thus favouring detection of the γˈ-depleted zone and/or the substrate. Cross-sections were thus performed to elucidate the oxidation mechanism of AM1 materials at this very low temperature of 750 °C (Fig. 3).

Fig. 2
figure 2

a Optical microscopy images of the surface of the AC AM1 and FHT AM1 after various oxidation times and b comparative metallic contribution (in at. %) obtained by EDS performed at 20 kV on different areas of the surface after 240 h of oxidation in air at 750 °C

Full size image
Fig. 3
figure 3

Cross-sections of the AC (dendrite, interdendritic zone and eutectic pool) and of the FHT AM1 after oxidation in air for 240 h at 750 °C

Full size image

The three characteristic zones (dendrites, interdendritic zones and eutectic pools) of AC AM1 are clearly identified by slightly different oxidation attacks, although they all show a γˈ-depleted zone below the oxide scale. Oxidation appears to be limited to certain areas of the eutectic pools, as shown in Fig. 3. However, except for these characteristic areas, the oxidation appears to be relatively similar involving internal oxidation in all cases. The complementary Raman analyses in different locations of the cross-section reveal two internal alumina phases (Fig. 4). For the AC AM1 substrate, a metastable phase of θ-Al2O3 was mainly observed at the interdendritic zones, while in areas with a low volume fraction of γˈ (dendrite), the stable α-Al2O3 phase is almost the only one to be formed despite the very high temperature or very long oxidation times required for its development [26]. Therefore, the formation of the α-Al2O3 phase in the dendritic areas is believed to occur because of the greater Cr content that favours Cr2O3 from which α-Al2O3 grows (third element effect) [22, 25]. The intermediate zones present both phases simultaneously, as does the entire surface of the FHT AM1. Complementary EDS analyses highlighted the presence of Al2O3 in regions with a large volume fraction of γˈ with Ni and Cr oxides above (Fig. 3) in line with the observations of Perez et al. on the same FHT AM1 superalloy [27]. Although there are still some uncertainties about the chromium segregation in the interdendritic areas, there appears to be a preferential trend for chromia formation in areas with a reduced volume fraction of the γˈ phase such as dendrites discussed earlier.

Fig. 4
figure 4

Raman spectra in the fluorescence domain (with normalized intensity) of AC AM1 and FHT AM1 oxidized in air at 750 °C for 240 h

Full size image

The oxidation processes of the γ/γˈ structure of AM1-type superalloys can first be explained thermodynamically using the Ellingham diagram which takes into account the partial pressure of oxygen and temperature [29]. Aluminium requires the lowest oxygen partial pressure to form Al2O3 compared to the other superalloy elements. However, the kinetics and volume fraction of the γˈ phase also influence the oxidation mechanisms. It is indeed observed the same type of successive layers on the different areas of the substrate, but with variable oxide levels. In particular, the internal oxidation (alumina formation) is observed to be greater in areas with a high fraction of γˈ (interdendritic and eutectic zones) than in the dendritic zones (area with a lower fraction of γˈ). Yet, the outer layer of NiO appears in the dendritic and interdendritic areas due to its relatively rapid diffusion at low temperatures [29, 30] but barely grows over the eutectic areas because of the coverage with Al2O3. The initial thickening of the NiO layer reduces the partial pressure of oxygen at the interface between the oxides and the substrate, favouring the external formation of chromium oxides (Cr2O3) and of aluminium oxides (Al2O3) in contact with the substrate. Such minor heterogeneities of the oxides growth at 750 °C have been ascribed to the different partitioning of the elements like chromium in other superalloy systems oxidized at higher temperatures [22, 30, 31].

Hot Corrosion by Na2SO4 Deposits

The addition of sodium sulphate to the surface of the two materials was then studied under the same experimental conditions as for oxidation at 750 °C under air for 240 h. The presence of salt on the surface of each sample limits the value of kinetics and surface analyses. The only relevant observations concern the change in the structure of Na2SO4 (from orthorhombic thenardite to trigonal aphthitalite) and the formation of traces of Na2CrO4 during the corrosion tests as observed by XRD (not given here) and Raman spectroscopy (Fig. 5) in line with the Raman results of other works [32, 33]. Then, similar peaks to those found under pure oxidation are observed in XRD and Raman, i.e. NiO and (Cr,Al)2O3. However, the oxide layer formed is very thin and the large number of elements in the AM1 superalloy make the phase analysis difficult. It is possible that undetected phases correspond to other spinel or rutile phases associated with refractory elements (Ta, W, and Mo).

Fig. 5
figure 5

Raman spectra (with normalized intensity) of AC AM1 and FHT AM1 oxidized (salt-free, 0 mg/cm2) and corroded (with 1 and 3 mg/cm2 of Na2SO4) in air at 750 °C for 240 h

Full size image

The cross-sections of the AC and FHT AM1 after the corrosion under deposits of 1 and 3 mg/cm2 are shown in Fig. 6. It can be shown in Fig. 6a that a thicker corrosion layer grows with the smaller deposit of Na2SO4 irrespective the metallurgical area of the AC AM1. Such growth of the corrosion layer seems to be associated with a relatively thicker γˈ-depleted zone (Fig. 6). Furthermore, the depletion is more significant in the dendritic areas than in the interdendritic ones. Therefore, there is a clear effect of the chemical segregation on the corrosion of the AC AM1 material (eutectic vs. dendrite and interdendrites, see Fig. 7) as opposed to the FHT condition of the AM1 superalloy, which in turn grows a much thicker corrosion layer (about 10 µm) than the AC version (about 2 µm), which is indicative of a propagation step and detaches off the surface in different areas.

Fig. 6
figure 6

a Cross-sections of AC AM1 and FHT AM1 corroded in air at 750 °C for 240 h with 1 and 3 mg/cm2 including magnifications of representative areas and b graph showing the average thickness with standard deviation of the γˈ-depleted zones, as a function of salt amount and specific metallurgical areas of the material

Full size image
Fig. 7
figure 7

High magnifications of Fig. 6-a showing the cross-sections of AC AM1 and FHT AM1 corroded in air at 750 °C for 240 h with 1 and 3 mg/cm2 Na2SO4

Full size image

The higher magnification of Fig. 7 highlights that the dissolution of the outer γˈphase appears more marked with the decrease in the salt content in the AC condition but similar between the interdendritic AC and FHT given the greater initial γˈ distribution. In addition, it is noteworthy that the formation of round-shaped sulphides is significantly higher with the 1 mg/cm2 than with the 3 mg/cm2 deposits and that they appear to be mainly concentrated in the low γˈ volume fraction areas (dendritic areas in AC and in FHT). The sulphidation mechanism appears therefore to be the most harmful process in these tests despite the observation of internal oxidation at some locations. This suggests that in the absence of additional SO2/SO3 to the gas flow, the formation of the sulphides derives from the reactions, whereby sulphates decompose due to the reduction of PO2 upon the formation of NiO as follows: SO42− → SO3 + O2−, then SO3 → SO2 + 1/2 O2 and finally, SO2 → 1/2 S2 + O2 [34,35,36]. Then, the reaction of (S2 + M) gives different stoichiometries depending on the metal considered. Looking at the different effects (salt amount and microstructure), the attack turns out to be more pronounced for the material subjected to a complete heat treatment with a deposition of 1 mg/cm2 including a marked propagation step.

Additional SEM–EDS analyses were performed for more accurate identification of degraded areas, and they were annotated directly on Fig.7. Chromium sulphide (CrS) is detected in all areas of the γˈ-depleted zones. In addition, an Al depleted layer is identified at the interdendritic zones and eutectic pools suggesting two distinct regions separated by a front of tiny CrS precipitates. This boundary shows an upper zone depleted in aluminium and enriched in sulphur. This observation suggests that Na2SO4 decompose following the reactions given above and that the O2 released induces oxidation in a first stage to develop an oxide layer through which sulphur is transported. Therefore, oxidation and sulphidation of the substrates occur in addition to basic fluxing assumed to occur, because Na2CrO4 has been detected (Fig. 5), without any contribution from SO2/SO3 flux. Progress of internal sulphidation has been reported to occur by the oxidation of Cr that promotes the release of S atoms that diffuse into the substrate [7]. The γˈ-cubes embedded in the corrosion layers of the interdendritic and eutectic areas with 1 mg/cm2 Na2SO4 (Fig. 7) are solely ascribed to the polishing process or electrochemical etching [37] but not to their dissolution since the Na2SO4 salt remained unmolten.

The analysis of the corrosion products of FHT AM1 indicates the formation of an outer layer of NiO, followed by a layer of oxides rich in Co and Cr, and an internal Al-rich layer, i.e., the same structure as with the pure oxidation, but the oxide layers are now far thicker. This is in line with the findings of Lortrakul et al. [38] when corroding CMSX-4 at 700 °C with Na2SO4 and O2/SO2/SO3, i.e. under typical Type II hot corrosion conditions, they reported that the oxide scales grew much faster than in our case following two stages. The first one was related to the active corrosion through the eutectic Na2SO4-NiSO4 although like in our case, they did not find any Ni and/or Co sulphate at the top. The second stage was governed by the SO2/SO3 mixture but they also claimed that the initial oxides had formed a barrier for the further attack of the S-containing gas and therefore, the gas had little influence. This also occurs with the increase in the amount of deposited salt from 1 to 3 mg/cm2, which decreases surface degradation according to D.A. Shores who indicated that a thick film and a diluted SO3 environment leads to basic conditions [39]. Despite the absence of the SO2/SO3 gaseous environment, this corrosion study indicates that the corrosion resistance of AC AM1 superalloy appears to be greater than that of FHT AM1, due to the protective Cr2O3 in areas characterized by a low volume fraction of γ ˈ that reacts under the present conditions to result in a Na2CrO4 (Fig. 5). The formation of Cr2O3 is assumed to occur very much like in the pure oxidation situation previously discussed.

Conclusions

The oxidation and hot corrosion of the A s- C ast (AC) and Fully Heat-Treated (FHT) AM1 single-crystal nickel-based superalloy were investigated at 750 °C in air (1 atm.) for 240 h. The main experimental observations lead to the following conclusions:

  1. 1.

    Under pure oxidation, a complex oxide layer is formed at 750 °C with mainly NiO, Cr2O3 and Al2O3. Depending on the volume fraction of γˈ, oxidation is more pronounced. The chemical segregation of the AC state enables more Cr2O3 to be formed in the dendrites (low fraction of γˈ) and more Al2O3 in the interdendritic and eutectic pools (high fraction of γˈ).

  2. 2.

    Hot corrosion at 750 °C can occur in single-crystal AM1 superalloy in the absence of SO3 (g) flow given the sulphidation, oxidation and basic fluxing observed. This likely derives from the decomposition of the sulphate and then of the SOx molecules producing S2 and O2. The increase in salt content decreases hot corrosive attack by forming a barrier layer to the external gas (air). Furthermore, the metallurgical segregations in the AC superalloy allow to extend the incubation period through the formation of a protective Cr2O3 layer like the one observed under pure oxidation in air.