The Effects of Chromium on the High Temperature Corrosion of Ni-Cr Alloys Exposed to Calcium Sulfate

Calcium and sulfur rich deposits have been linked to failure of turbine components as a consequence of high temperature exposures (> 1000°C). There are only limited studies on the effects of these deposits on the degradation behavior of turbine alloys. To gain further understanding of this phenomenon, a systematic study was undertaken with model binary nickel-chromium alloys. Three alloys with different chromium contents – low, medium, and high – represented by Ni-5Cr, Ni-10Cr and Ni-18Cr, were exposed to CaSO 4 -deposit-induced corrosion in the 900–1100°C temperature range. At 1000 and 1100°C, the decomposition of CaSO 4 led to the formation of calcium chromates and chromium sul�des. At the lower temperature, 900°C, there was only limited decomposition of CaSO 4 , allowing the formation of a continuous Cr 2 O 3 scale was able to form.


Introduction
Alloys used in gas turbines usually contain Al and Cr to form protective Al 2 O 3 or Cr 2 O 3 scales.While these oxides are highly protective in purely oxidative environments, the presence of additional impurities in the environment can lead to accelerated degradation of these lms.Historically, the impurities present in gas turbines were rich in sodium and sulfur.The predominant sources of these impurities were: ocean water (sodium) and aircraft fuel (sulfur).These two impurities led to the formation of sodium sulfate, Na 2 SO 4 , which deposited on the turbine components as a liquid lm and resulted in a phenomenon called "hot corrosion" [1][2][3][4][5].This type of corrosion was further subdivided into Type I or Type II hot corrosion.Type I hot corrosion was caused by the presence of a pure Na 2 SO 4 lm during exposures in the 900-950°C range.The acidity or basicity of the salt lm is controlled by the partial pressure of SO 3 (p SO3 ) in the surrounding environment, e.g., high values of p SO3 leads to the formation of a more acidic lm.Hot corrosion occurs through an oxide uxing mechanism where the p O2 and p SO3 gradient in the lm creates an oxide solubility gradient [1][2][3].The oxides growing on the alloy will dissolve if their solubility is higher at the alloy/salt lm interface.This dissolution will be followed by re-precipitation closer to the salt lm/air interface where the solubility of the oxide(s) is lower.This dissolution-precipitation phenomenon occurs due a negative solubility gradient.While NiO is known to satisfy the negative solubility gradient criterion and is susceptible to hot corrosion, Al 2 O 3 and Cr 2 O 3 are considered protective in these environments as the negative solubility gradient is never satis ed [1][2][3].At lower temperatures (650-750°C), type II hot corrosion (also known as "low temperature hot corrosion") can occur [4,5].In this temperature range, and under high partial pressures of SO 3 , a low-melting eutectic forms between Na 2 SO 4 and NiSO 4 (or CoSO 4 ), leading to the formation of a liquid lm on the turbine components [4,5].
The oxide beneath this lm can dissolve and as the lm is continuously replenished by new salt deposits in the turbine environment, continuous dissolution of the oxide occurs as the lm never gets saturated.
Since both Type I and II hot corrosion require elevated partial pressures of SO 3 , efforts have been made to limit the amount of sulfur in aircraft fuels to eliminate this form of corrosion [6].The success of this approach paralleled the development of high engine operating temperatures.The latter development was driven by the need to improve engine e ciencies.Since these temperatures are well above the temperature ranges where hot corrosion has been reported to occur, it was expected that oxidation is the only mechanism that may be observed [4,5].However, even after these developments, analysis of corrosion products on gas turbines were still found to be rich in sulfur [6].In addition, other constituents such as calcium, magnesium, aluminum, or silicon (CMAS) have also been reported [6][7][8][9][10][11].The calciumrich species may have originated from ingested sand particles which can contain either CaO (lime) or CaSO 4 (gypsum) or from seawater.Apart from the fuel, sources of sulfur are likely to be atmospheric pollutants, seawater or other ingested sulfur-containing compounds [7,8].When operating temperatures are high, thermal barrier coatings (TBCs) are applied to turbine blades.CMAS may melt and penetrate the TBCs due to their low viscosity [12,13].The molten CMAS can then solidify deeper in the TBC.In addition, molten CMAS can dissolve yttrium from the CMAS, destabilizing the cubic ZrO 2 structure and cause a phase transformation to tetragonal ZrO 2 .Both of these mechanisms contribute to internal stresses in the TBC, causing fracture and spallation of the TBC.This exposes the underlying substrates to the CMAS and the surrounding environment [12,13].
Several studies on the effects of CMAS on alumina and chromia formers reveal that the corrosion behavior depends on the constituents in CMAS and in the substrate.Some of the constituents of CMAS, such as alumina and silica, do not negatively in uence the high temperature stability of alumina or chromia forming alloys.However, other constituents, such as calcium oxide, react with alumina and chromia to form calcium aluminates or calcium chromates [14].For alloys that are alumina formers, the reaction with calcium oxide results in the formation of calcium aluminates.The subsequent thinning of the alumina surface lm leads to faster consumption of aluminum from the substrate [15][16][17].For chromia formers, calcium oxide reacts with the chromia lm to form calcium chromates.This can lead to accelerated degradation of the alloy due to the formation of a liquid phase [15][16][17].The liquid phase forms because of a eutectic reaction between CaCrO 4 and Ca 5 Cr 3 O 13 in the CaO -Cr 2 O 3 system with a predicted eutectic temperature of 1069°C [18][19][20].
Studies on the effect of CaSO 4 on the corrosion of alumina and chromia formers are limited.Early reports of CaSO 4 deposit-induced degradation were from the power plant sector, where it was observed in ironbased alloys operating at low partial pressures of oxygen (p O2 ) in uidized bed coal combustor environments.This low p O2 -environment leads to the formation of wüstite, FeO, which then catalyzes the decomposition of CaSO 4 to CaO and SO 3 .Minimal interaction was observed between CaO and FeO; however, the high p SO3 resulting from the CaSO 4 decomposition led to internal sul dation [21][22].In the case of nickel-based alloys for aircraft turbine environments, the effect of CaSO 4 appears to be temperature dependent.At low temperatures (~ 900°C), CaSO 4 decomposes slowly, with minimal effect on degradation, thereby resulting in alloy microstructures similar to pure oxidation.At higher temperatures (> 1050°C), decomposition of CaSO 4 to CaO and SO 3 occurs rapidly, with potential negative outcomes in regard to alloy stability.For Ni-base superalloys that form alumina scales, CaSO 4 decomposition can produce sulfur-containing calcium aluminate-sulfates such as Ca 4 Al 6 O 16 S [23,24], with CaSO 4 exposures at 1150°C leading to faster degradation kinetics compared to a pure alumina scale.Localized failure of the oxide scale leads to NiO nodules forming at various locations.A thin Al 2 O 3 layer is observed at the bottom of the oxide scale with chromium sul des particles beneath.It is to be noted that these alloys, for 900°C exposures, formed a protective Al 2 O 3 scale [23].For chromia forming alloys, the effect of CaSO 4 on degradation phenomena is less understood and reported, especially in regard to aircraft turbine environments.In this study, a systematic study was undertaken to understand the effect of temperatures of exposure and chromium contents, in model binary Ni-Cr alloys.

Materials and Methods
Three Ni-Cr alloys (Ni-5Cr, Ni-10Cr and Ni-18Cr) were cut to dimensions of around 7 x 7 x 3 mm.These samples were covered by a CaSO 4 -ethanol suspension (0.5 gram salt / 5 mL ethanol) and dried repeatedly until 35±3 mg/cm 2 of CaSO 4 was applied on the surface of the coupon.These coupons were placed onto a ceramic plate and the mass of the coupon, the plate and the combined coupon + plate mass were recorded.
Coupons were inserted into a furnace and held isothermally at 900, 1000 or 1100°C for 50 hours in owing air.The ramp-up stage consisted of a 2-hour heating period to 300°C, followed by a 5 minute hold that that temperature, then heating at 4.4°C/min to the test temperature.The cool-down stage was a furnace cool to room temperature.After isothermal exposure, the coupon, the ceramic plate and the combined mass of the coupon plus ceramic plate was recorded.
Sample surfaces were characterized using X-ray diffraction (XRD).The scan rate was 2.4 °/min (resolution of 0.02 ° and dwell time of 0.5 seconds) and the scan range (2θ) was from 10 to 90°.For crosssectional analysis, the coupons were mounted in epoxy and ground and polished to 0.05 µm in dry media to avoid any reaction of the CaSO 4 lms with water.Optical micrographs were obtained at ve different locations at a 500x magni cation and three different locations at 300x to measure the thickness of the oxide lms and to check for homogeneity of the lms.Characterization using scanning electron microscopy was conducted backscattered electron contrast and electron dispersive microscopy to obtain elemental maps.with a Ca/Cr ratio of 0.5 (through EDS point analysis), suggesting the formation of β-CaCr 2 O 4 (Figure S3 and Table S3).S4).This layer is likely to be NiO with dispersoids of NiCr 2 O 4 .

Results
Discrete particles of Cr 2 O 3 (internal oxidation) were observed beneath the outer oxide layers.Cr-rich sul des (likely Cr 3 S 2 based on the Cr/S ratio) extend deeper into the alloy from the oxide/alloy interface.
Ni-10Cr formed an outer NiO scale, a middle calcium chromate layer and an inner Cr 2 O 3 layer.From EDS analysis (Figure S5 and Table S5), the calcium chromate is likely CaCrO 4 due to the Ca/Cr ratio being close to 1:1.Cr-rich sul des (likely CrS from EDS analysis) are found beneath the alloy/oxide interface and extending into the alloy; however, the extent of penetration of the sul des into the substrate is less for Ni-10Cr compared to Ni-5Cr.Ni-18Cr also formed an outer NiO scale, a middle calcium chromate layer and an inner Cr 2 O 3 layer.The calcium chromate had some regions with a Ca/Cr ratio of 1:1, suggesting the formation of CaCrO 4 , while other regions had a Ca/Cr ratio of 0.5, suggesting β-CaCr 2 O 4 as shown in Figure S6 and Table S6.Some Cr-rich sul des are present, but to a lesser extent compared to Ni-10Cr.These sul des were too small to be analyzed through EDS point analysis.Figure 5 shows backscattered electron images of the cross-section of Ni-5Cr, Ni-10Cr and Ni-18Cr exposed to CaSO 4 for 50 hours at 1100°C.Ni-5Cr formed a thick uneven NiO scale with embedded CaSO 4 particles located closer to the oxide/salt interface.Calcium chromate particles are also present in the NiO scale, located towards the bottom of the scale, i.e., near the oxide/alloy interface.From EDS analysis (Figure S7 and Table S7), the Ca/Cr ratio of the calcium chromate particles is around 0.  S9 and Table S9).Ni-18Cr also formed Cr-rich sul des in the substrate beneath the scale, similar to the other alloys.

Discussion
As an overall discussion point, for all alloys, the external oxide thickness increased as the exposure temperature was increased in the 900-1100 °C range, for the same time of exposure (50 hours).In contrast, the oxide thickness decreased with increasing Cr content.These observations are supported by the results summarized in Fig. 7, a plot of oxide thicknesses Vs exposure temperature.These ndings are reasonable given that it increasing temperatures result in faster CaSO 4 decomposition (to be discussed further below), and that the increased presence of Cr offers additional protection to oxidation and sul dation.
To better understand the mechanisms of alloy response to calcium sulfate exposure, microstructurebased schematics of Ni-5 Cr, Ni-10 Cr and Ni-18 Cr exposed to CaSO 4 at 900, 1000 and 1100°C for 50 hours are shown in Fig. 8.The discussion below pertains to this gure.
At 900°C, all three alloys formed an external NiO scale and an internal Cr 2 O 3 scale.Ni-5 Cr formed internal chromium sul des, suggesting that the outer scale was unable to fully protect against sul dation.At 1000 and 1100°C, the decomposition of CaSO 4 led to the formation of CaO and SO 3 , causing both CaOdeposit induced corrosion and sul dation.At both of these temperatures, Ni-5 Cr experienced the most sul dation, as evidenced by the presence of Cr-sul des deepest (relative to the other two alloys) within the substrate.Ni-18 Cr experienced the least amount of sul dation (although at 1100°C, there were a larger number of smaller Cr sul de particles, but not extending deep into the substrate).A thick NiO scale formed that decreased in thickness with increasing Cr content.In addition, except for Ni-5 Cr at 1000°C, calcium chromates were observed between the NiO and Cr 2 O 3 scales.
At high temperatures, CaSO 4 decomposes to CaO and SO 3 [23].The SO 3 , can decompose to SO 2 and O 2 , as it equilibrates with the surrounding air.The relevant chemical reactions (1-3) are listed below.).Note that the activity of CaO (a CaO ) and CaSO 4 (a CaSO4 ) are assumed to be unity as they are pure species.In addition, the rate of decomposition is much faster at higher temperatures, following an Arrhenius-like behavior [25].Given that little to no CaO or calcium chromates were detected on the alloys exposed to CaSO 4 at 900°C, it is assumed that the low partial pressure of SO 3 and slow kinetics resulted in little to no decomposition of CaSO 4 [25].Since almost no CaO was present, the salt lm did not react with any of the Ni-Cr alloys [10].Thus, oxidation was the main corrosion mechanism at this temperature and a stable Cr with the SO 2 /SO 3 gas mixture (initiated by the decomposition of calcium sulfate) initially dissolves into the alloy, forms discrete oxide particles just beneath the surface, and when these particles reach a critical number and size, they will coalesce to form a continuous outer oxide scale.
The critical amount of Cr to form an oxide scale depends on the solubility of O 2 in the alloy and the diffusivity of oxygen and Cr in the alloy as shown in the inequality (Eq.9), where N Cr is the concentration of Cr in the alloy, is the critical volume fraction of internal oxides to form a continuous scale, is the stoichiometric ratio of O to Cr in Cr 2 O 3 ( ), is the solubility of oxygen in the alloy, D O is the diffusivity of oxygen in the alloy, D Cr is the diffusion coe cient of chromium in the alloy, V ox is the molar volume of the oxide and V m is the molar volume of the alloy [30,31].A possible reason for the formation of a Cr 2 O 3 scale even in the low Cr alloy (Ni-5Cr) can be attributed to the establishment of a low p O2 value based on the equilibrium between CaSO 4 and CaS.Considering equilibrium between the oxygen at the surface of the alloy and the dissolved oxygen inside the alloy, the decrease in p O2 will decrease the amount of dissolved oxygen in the alloy .Following Eq. 9, the decrease in will decrease and lower the critical amount of chromium required in the alloy to form an external scale.An increase in p O2 could increase the critical amount of chromium needed; however, is also going to be limited by the solubility limit of oxygen in the Ni-Cr alloy, thus placing an upper limit on .Studies on oxygen solubility in solid Ni-Cr alloys were not available.However, in studies on liquid Ni-Cr alloys in equilibrium with chromium oxide, oxygen solubility was found to increase with Cr content [32].Table 2 shows the partial pressure of O 2 under equilibrium of reaction 10, which is likely the p O2 at the interface between the alloy and the CaSO 4 lm.At all three test temperatures, the p O2 is lower than 1e-9.This lower partial pressure of O 2 may result in a low amount of dissolved oxygen in the alloy, thereby reducing the amount of Cr necessary for Cr 2 O 3 to form.In a related recent study [14], Ni-Cr alloys exposed to CaO above 1000°C were observed to undergo signi cant degradation due to the formation and growth of the liquid calcium chromate lm.Despite the predicted eutectic being above 1000°C, extensive degradation was still observed either due to potential inaccuracies of the eutectic temperature reported in the literature [19,20,35] and/or the addition of NiO depressing the melting point and/or an exothermic reaction for oxide formation that locally heats up the surface [14,35].However, this catastrophic degradation behavior was not observed when the Ni-Cr alloys were exposed to CaSO 4 .It is possible that the rate of CaSO 4 decomposition limited the supply of CaO interacting with the substrate and the oxide scale, which would limit the amount of liquid calcium chromates that was able to form [14,25].A proposed mechanism for this behavior as described above is shown in Fig. 10 showing the decomposition of CaSO

Figure 1
Figure 1 shows backscattered electron cross-sectional micrographs of Ni-5Cr, Ni-10Cr and Ni-18Cr exposed to CaSO 4 at 900°C for 50 hours.In all alloys, a continuous Cr 2 O 3 layer was observed.For the Ni-18Cr, the Cr 2 O 3 layer, with embedded islands of NiO nodules, constituted the outer scale.For the Ni-5Cr and Ni-10Cr alloys, the chromia layer was located beneath a continuous outer NiO scale.Calcium was present in the oxide scale for Ni-10Cr and Ni-18Cr indicating the possibility of a reaction with CaSO 4 .The remnant salt lm above the oxide scale that formed on the Ni-18Cr contained calcium chromate particles

Figure 2
Figure 2 shows x-ray diffractograms of Ni-5Cr, Ni-10Cr and Ni-18Cr exposed to CaSO 4 at 900°C for 50 hours.In all three cases, CaSO 4 was detected.For Ni-5Cr and Ni-18Cr, NiO and Cr 2 O 3 were also detected.It is likely that the outer Cr 2 O 3 scale on Ni-18Cr was too thin to be discerned by x-ray diffraction.

Figure 3
Figure 3 shows backscattered electron images of the cross section of Ni-5Cr, Ni-10Cr and Ni-18Cr exposed to CaSO 4 at 1000°C for 50 hours.Ni-5Cr formed a thick outer NiO scale with the surface of the scale containing embedded CaSO 4 particles.Beneath the NiO scale was a layer rich in Ni and Cr.EDS data suggests that this could be NiCr 2 O 4 .Beneath the NiCr 2 O 4 scale is another Ni-rich layer that also contains Cr (around 10 at% Cr, Figure S4 and TableS4).This layer is likely to be NiO with dispersoids of NiCr 2 O 4 .

Figure 4
Figure 4 shows x-ray diffractograms of Ni-5Cr, Ni-10Cr and Ni-18Cr exposed to CaSO 4 at 1000°C for 50 hours.For Ni-5Cr and Ni-10Cr, only CaSO 4 and NiO were detected.The salt lm and the NiO scale were likely thick enough that any phases below these two layers were not detected through x-ray diffraction.In the Ni-18Cr x-ray pattern, CaCrO 4 was detected in addition to NiO and Cr 2 O 3 .

Figure 6
Figure 6 shows x-ray diffractograms of Ni-5Cr, Ni-10Cr and Ni-18Cr exposed to CaSO 4 at 1100°C.Both Ni-5Cr and Ni-10Cr showed the presence of CaSO 4 and CaO suggesting signi cant decomposition of CaSO 4 .The presence of Cr 2 O 3 and calcium chromates such as Ca 3 Cr 2 O 8 , CaCrO 4 and β-CaCr 2 O 4 were also detected on Ni-5Cr.On Ni-10Cr, the only calcium chromate that was detected was Ca 3 Cr 2 O 8 .Ni-18Cr only showed the presence of NiO and Cr 2 O 3 .

1 2 3
To provide a reference to the magnitude of the CaSO 4 decomposition, the equilibrium p SO3 was calculated based on the CaSO 4 decomposition as per chemical reaction (1) and the SO 3 -SO 2 -O 2 equilibrium as shown in chemical reaction(3).For CaSO 4 decomposition, Eq. 4 {based on the chemical reaction (1)} was used as the starting point and p SO3 was determined using equations 5 and 6 (where at equilibrium

4 5 6Figure 9
Figure 9 compares the equilibrium partial pressure of SO 3 due to the decomposition of CaSO 4 or equilibration of SO 2 and O 2 and Table1shows calculated ppm of SO 3 at each test temperature.As

RT
2 O 3 lm was able to form on all three Ni-Cr alloys.Past observations on the oxidation of Ni-Cr alloys in air suggests that the transition between internal Cr 2 O 3 to an external Cr 2 O 3 scale occurs around 10-11 wt% Cr [28, 29].Despite this, a Cr 2 O 3 scale was observed at 900°C even on the Ni-5 Cr alloy.The oxygen in equilibrium 4 into CaO and SO 3 leading to diffusion of its ionic constituents through the NiO lm.Interaction between the dissolve CaO and the inner Cr 2 O 3 scale forms β-CaCr 2 O 4 and the dissolved sulfur species diffuses further to form internal chromium sul des.5.ConclusionsNi-5Cr, Ni-10Cr and Ni-18Cr were exposed to a CaSO 4 salt lm at 900, 1000 and 1100°C for 50 hours in owing air.At 900°C, the decomposition of CaSO 4 was limited, and minimal interaction between the salt and the alloy was observed.An external Cr 2 O 3 scale was observed to form on all three alloys.At 1000 and 1100°C, the decomposition of CaSO 4 into CaO and SO 3 were found to result in the formation of unprotective calcium chromates and internal chromium sul des.The stability of the Cr 2 O 3 scale increased with increasing Cr content, where higher Cr contents led to the formation of a thinner oxide

Figure 4 X
Figure 4

Figure 6 X
Figure 6

Figure 10
Figure 10 5, suggesting that these particles are β-CaCr 2 O 4 .Internal oxide particles, identi ed as Cr 2 O 3 , are observed within the alloy in the vicinity of the oxide/alloy interface.Further into the alloy, beneath the zone where the Cr 2 O 3 particles are NiO layer to the NiO/chromium oxide interface): (a) an almost continuous calcium chromate (dark gray contrast) layer with a Ca/Cr ratio of 1.5-1.6,suggesting the presence of a mixture of CaCrO 4 , Ca 5 Cr 3 O 13 and Ca 3 Cr 2 O 8 (Figure S8 and TableS8), and (b) discrete particles (medium gray contrast) with a Ca/Cr ratio of 0.5, indicating the presence of β-CaCr 2 O 4 .Beneath this complex oxide scale, Cr-rich sul des are observed inside the alloy.Ni-18Cr formed a three-layer oxide: an outer NiO scale, a middle calcium chromate scale and inner Cr 2 O 3 scale.The Ca/Cr ratio in the calcium chromate layer was around 0.5, suggesting it was β-CaCr 2 O 4 (Figure present, particles of Cr-rich sul des are observed.Ni-10Cr formed a complex, multi-layered oxide scale nominally consisting of (from the outermost to the innermost): an outer calcium chromate (dark gray contrast), a thick nickel oxide (light gray contrast), and internal chromium oxide (medium gray contrast) layers.There are other constituents present within the NiO layer.These consist of (from the middle of the Table 1 shows calculated ppm of SO 3 at each test temperature.As shown, the partial pressure of SO 3 due to the decomposition of CaSO 4 increases rapidly with temperature (note that the y-axis of the plot is in log scale) while the partial pressure due to SO 2 /O 2 equilibrium decreases with increasing temperature.At 900°C, the p SO3 due to CaSO 4 decomposition is below of what would be produced by equilibrium with 0.1 ppm SO 2 and 0.21 atm O 2 .At 1000°C, the p SO3 from CaSO 4 is close to that produced by 1 ppm SO 2 and at 1100°C, the p SO3 is above what would be produced by 10 ppm SO 2 .

Table 1
Equilibrium partial pressure of SO 3 (in ppm) due to CaSO 4 decomposition or equilibrium in air + SO 2 .

10
[15]34] and 1100°C, the CaSO 4 lm was able to decompose enough to form CaO and react with the alloy.For Ni-5 Cr, an external NiO scale was followed by the formation of internal Cr 2 O 3 oxides; for Ni-10 Cr and Ni-18 Cr, a three-layer scale composed of NiO, calcium chromates and Cr 2 O 3 formed.The NiO scale is thick compared to exposures to CaO.This is likely caused by the sulfur impurities in the NiO scale as sulfur can take an ionic charge from − 2 to + 6[33,34], which can drastically increase the amount of cation vacancies in NiO[33,34].As CaO is partially soluble in NiO, diffusion of Ca 2+ ions can occur through the NiO external scale and react with the inner Cr 2 O 3 scale forming calcium chromates[18].Another possibility is the initial formation of Cr 2 O 3 reacting with the formed CaO to create a calcium chromate scale.This would break down the Cr 2 O 3 scale, allowing for NiO to form and ux outwards given the low solubility of NiO in liquid calcium chromates to form an external scale[15].Therefore, because the NiO acts as a diffusion barrier between the alloy and the salt lm, the amount of Ca ions reaching the substrate is limited and Cr 2 O 3 is able to form as a continuous scale.Beneath this scale, chromium sul des were able to form at 1000 and 1100°C.As the chromium content increased, the depth at which the chromium sul des were present in the substrate appeared to decrease.The higher stability of the Cr 2 O 3 scale may have mitigated the rate at which sulfur was able to enter the substrate.