The effects of water vapour on the hot corrosion of gas turbine blade materials at 700 °C

ABSTRACT Future developments in power generation are likely to require gas turbines to operate in novel combustion environments. The level of water vapour in the turbine’s gas stream is one variable that may change as a consequence. This paper explores the effects of water vapour on hot corrosion. The ‘deposit recoat’ technique was used to evaluate the hot corrosion performance of superalloys PWA 1483 and MarM 509 in atmospheres containing between 0 and 20 vol.% water vapour. Exposures were carried out at 700 °C in atmospheres containing 300 ppm SO2 for up to 500 hours. The deposit was an 80% Na2SO4, 20% K2SO4 solution applied with a deposition flux of 1.5 µg/cm2/h. The findings are compared to similar exposures with 3.6 ppm SO2. Increasing levels of water vapour were observed to reduce sound metal loss in atmospheres containing 300 ppm SO2 while increasing sound metal loss in atmospheres containing 3.6 ppm SO2.


Introduction
The majority of the driving force behind the development of future industrial gas turbine technologies aims to achieve two things: increased efficiency of power generation and a reduction in CO 2 gas emissions [1]. To meet these two objectives, gas turbines are likely to be required to operate under novel conditions. These could include higher firing temperatures to achieve greater thermal efficiencies and the combustion of alternative fuels. Hydrogen rich syngas derived from the gasification of solid fuels is a potential form the new fuels may take [2].
The level of water vapour in the gas stream passing through the power turbine section of an industrial gas turbine is one variable influenced by the type of fuel combusted [3]. For example, power turbine environments resulting from the combustion of natural gas have been calculated to contain approximately 10 vol.% water vapour, whereas power turbine environments resulting from the combustion of hydrogen-rich syngas are expected to contain approximately 20 vol.% water vapour [4].
The effect water vapour has on the oxidation of materials typically used in high temperature applications has been investigated by a number of authors [5][6][7][8]. Most authors observe increased rates of oxidation for Cr 2 O 3 forming alloys in environments containing water vapour. One mechanism explaining this increased rate of oxidation is through the formation of volatile CrO 2 (OH) 3 and Cr(OH) 3 species that may be subsequently vaporised [6,7]. J. Zurek et al. attributed increased oxidation rates to an observed decrease in the chromia grain size, however these observations were made in a low pO 2 atmosphere [9].
Hot corrosion is an accelerated form of degradation through the amphoteric fluxing of salt species rather than having condensed onto component surfaces [10,11]. Hot corrosion is recognised to occur in two stages: an incubation stage where little metal loss occurs followed by a propagation stage where catastrophic metal loss occurs [12]. The length of the incubation stage is thought to correspond to the breakdown of the protective oxide scale of the afflicted alloy. It is therefore plausible that if water vapour can degrade the protective chromia scale of a chromia forming alloy (either through vaporisation or another method), then it can be postulated that water vapour will increase hot corrosion rates through the shortening of the incubation period.
Previous research has compared hot corrosion in atmospheres generated through the combustion of natural gas and hydrogen-rich syngas [13]. These environments had water vapour contents of 10 and 20 vol.%, respectively, and an SO x content of 3.6 ppm. The deposit used was 80% Na 2 SO 4 20% K 2 SO 4 (molar ratio) applied at a flux of 1.5 µg/cm 2 /h. Under these conditions, corrosion rates appeared more rapid in the atmosphere containing 20 vol.% with exposures in the 10 vol.% water vapour atmosphere remaining in incubation over the 500 hour test period. The median probability of change in sound metal exceedance plots for these exposures is given in Figure 1.

Methodology
The effects of increasing levels of water vapour on hot corrosion have been investigated using the 'deposit recoat' method. This involves the cyclic high temperature exposure of samples with deposits applied at the start and between each cycle [14].
Exposures were carried out in a vertically aligned tube furnace fitted with a ceramic corrosion vessel ( Figure 2), with samples located within the furnace hot zone (±5 °C of target temperature) using a ceramic carousel. The test gas (300 ppm SO 2 in air) was passed through the centre of the carousel to enter the corrosion vessel from the bottom and exit from the top of the furnace. This gas pathway allows time for the SO 2 to reach an equilibrium with SO 3 . Water was injected into the gas steam as it passes through the carousel using an independently calibrated peristaltic pump. Three different concentrations of water vapour in the test atmosphere were tested; 0, 10 and 20 vol.%. A deposit of 80% Na 2 SO 4 20% K 2 SO 4 applied to the samples, heated to approximately 100°C, at a flux of 1.5 µg/cm 2 /h using a compressed air fed spray gun. Samples were removed after 50, 100, 200 and 500 hours of exposure.    The materials selection for this study included both a nickel-based superalloy, PWA 1483, and a cobalt-based superalloy, MarM 509. Both these alloys form a protective chromia scale and feature solid solution strengthening refractory metal carbide phases within the alloy matrix. The nominal compositions of these materials are given in Table 1. The area percentage of the alloy matrix that is made up of refractory metal carbides has been determined using Image J software [15], this was 0.9% for PWA 1483, and 6.2% for MarM 509. Samples were cylindrical in geometry with a chamfer machined at the top and bottom of the sample. The diameter and height were approximately 10 mm. Pre-test metrology of the sample dimensions was carried out using a digital micrometre with a resolution of ±1 µm.  ost test samples were mounted in a mixture of low shrinkage resin and ballotini (glass beads with a diameter of 0.04 to 0.07 mm. Samples were cross-sectioned, then ground and polished to a 1 µm finish using oil-based lubricants to avoid dissolving soluble species at the surface of the sample. Sample cross sections were measured at 30 points using an optical microscope with a calibrated x and y stage. Comparison with the pre-test metrology provided quantitative corrosion data. Qualitative assessment of corrosion morphology was carried out using an environmental scanning electron microscope (ESEM) outfitted with an energy dispersive X-ray (EDX) analysis system.
The partial pressure of species present in the different test atmospheres at 700 °C were determined using MT Data (NPL, UK) thermodynamic modelling software and are presented in Table 2. Of these species, SO 3 is particularly significant since it is known to stabilise the molten salt deposit [11]. In the 300 ppm SO 2 test atmospheres, the lower SO 3 partial pressures in the atmospheres containing H 2 O can mostly be attributed to the dilution of the test gas with H 2 O.

Results and discussion
Cumulative probability of change in sound metal exceedance plots are shown in Figure 3 for exposures containing 0, 10 and 20% H 2 O respectively. The change in sound metal is defined as the metal lost from the metal surface plus any internal damage.
Both PWA 1483 and MarM 509 generally show a lesser change in sound metal as the concentration of H 2 O in the atmosphere increases from 0% to 20%; this trend is more clearly illustrated in Figure 4. This finding runs counter to previous research (summarised in Figure 1) that observed an increasing change in sound metal with increasing water vapour. The mechanism behind this causing these contrasting trends is not clear; however, the key difference between these two studies lies in the SO 2 content in the test atmosphere (3.6 ppm in the previous research compared to 300 ppm) and any explanation for the difference in data trend is therefore likely based on atmospheric SO x content. It is possible that where SO x levels are in the order of 3.6 ppm, then oxidation becomes more important than fluxing mechanisms. When SO x levels are in the order of 300 ppm, fluxing becomes the more dominant mechanism of degradation. Water vapour may exacerbate oxidation (as supported by the majority of the literature) but retard hot corrosion. This could be through interactions with the gas phase or with the molten deposit phase.
Compared to MarM 509, PWA 1483 suffers a greater degree of sound metal loss in exposures with the three tested levels of water vapour 0, 10 and 20%. This finding is in contrast to the previous research that observed MarM 509 suffering greater sound metal loss than PWA 1483 (Figure 1). SO x in the gas phase is thought to shift the amphoteric behaviour of molten Na 2 SO 4 towards being more acidic. One possible explanation for the contrasting data trends is that in atmospheres containing lower SO x levels, alloy induced fluxing mechanisms brought about by refractory metal carbide precipitates in the alloy matrix become more important. MarM 509 has a greater proportion of such precipitates (6.2%) than PWA 1483 (0.9%) and therefore exhibits greater sound metal loss at a lower level of SO x . At higher levels of SO x other factors such as chromium content of the material may become more important, of which MarM 509 has 27.1 at.% compared to 13.7 at.% for PWA 1483, to form a better protective scale.
If the logic holds true that at high temperatures, increasing levels of water vapour causes increasing levels of degradation of the chromia scale through volatisation, and that the better the chromia scale, the better an alloys type II hot corrosion resistance, then for increasing levels of water vapour to not increase the degree of type II hot corrosion attack, an additional mechanism must take place. Such a mechanism could involve the removal of corrosive species from the system thereby changing the acidity/basicity of the melt.
Backscattered electron (BSE) images both representative of the majority of the sample and corrosion features for PWA 1483 and MarM 509 after 500 hours of exposure at all tested levels of H 2 O are shown in Figure 5. Most corrosion features of both PWA 1483 and MarM 509 exhibit internal damage which are shown to be species rich in both chromium and sulphur (Figures 6 and 7). The formation of chromium sulphide beneath the alloy surface is associated with type I hot corrosion basic fluxing which is more commonly observed at temperatures between 850°C and 950°C [16]. The corrosion features in Figure 5 show that damage is not evenly distributed and has formed pits, while the corrosion product has formed an outer layer shown in Figure 5 to be rich in nickel and cobalt and an inner layer rich in chromium. These features are commonly associated with type II and the temperature range of 605°C to 750°C [16]. Features associated with type I and II hot corrosion occurring together can be termed mixed mode corrosion and support Rapp's fluxing mechanism [11] (later expanded upon by Gleeson [17]) whereby the acidity and basicity of the corrosion melt changes over time.
Mixed mode corrosion features are seen to occur under conditions both with and without water vapour, therefore the presence of water vapour may not be connected to this phenomenon.
Micrographs representative of the majority of the samples' surfaces all exhibit at least some evidence of the formation of sulphide species beneath the alloy surface. These species are more numerous and have penetrated deeper in PWA 1483 compared to MarM 509.
For both PWA 1483 and MarM 509, corrosion pits have often, but not exclusively, formed around the oxidised remains of refractory metal carbide precipitates at the surface of the alloy. Such precipitates may exacerbate the hot corrosion process either by locally disrupting the formation of a continuous oxide scale or by allowing an alloy induced fluxing mechanism to occur [18,19].
Further research to thermodynamically model the phases that form under type II hot corrosion conditions that contain water vapour may shed light on the reason why water vapour appears to reduce the severity of type II hot corrosion attack and why only if the SO x levels are sufficiently high. Additionally, further empirical data could reveal if the same phenomenon occurs under type I hot corrosion conditions (typically around 900 °C).

Conclusions
A series of laboratory type II hot corrosion tests investigated the effects of increasing atmospheric water vapour levels. Tests were carried out at 700 °C in an atmosphere of 300 ppm SO 2 in air with 0, 10 and 20% water vapour contents, and a 80% Na 2 SO 4 20% K 2 SO 4 deposit with a flux of 1.5 µg/ cm 2 /h. Lower metal loss values were observed with increasing levels of water vapour in the test atmosphere. After 500 hours of exposure at water vapour levels of 0, 10 and 20% vol. respectively, the median metal loss values of PWA 1483 were as follows: 39.9 µm, 12.5 µm and 13.8 µm. For MarM 509 the median metal loss values were as follows: 12.2 µm, 12.6 µm and 5.3 µm.
The hot corrosion performance of MarM 509 is better than PWA 1483 in the 300 ppm SO 2 test atmosphere at all three tested levels of water vapour. This is the opposite of the observations made in the previous research that had a test atmosphere containing 3.6 ppm SO 2 , and 20% vol. H 2 O). It is proposed that at lower levels of SO 2 in the environment, alloy induced fluxing mechanisms of refractory metal phases (of which MarM 509 has more than PWA 1483) become more important than gas-phase fluxing mechanisms.
Mixed mode hot corrosion features (features typical of both type I and II hot corrosion) were observed under all test conditions. Mixed mode hot corrosion features are thought to come about by changes in the deposit melt chemistry that can alter which fluxing mechanism is dominant.