Application of thermal barrier coatings to hot section components in industrial gas turbines has been of interest for many years due to the significant fuel consumption benefits that can be achieved through reduced cooling air requirements or higher firing temperatures. A second benefit can be through improved corrosion resistance, particularly for use with vanadium contaminated fuels which aggressively corrode metallic coatings when not infiltrated by fuel additions. With the objective of making more fuels available worldwide for its fleet, the U.S. Navy is sponsoring programs to study the effects of lower quality fuels on engine performance and life and to develop ceramic coatings for protection of marine gas turbine components against vanadium-assisted hot corrosion.

Previous work at GE and elsewhere has shown that the most common ceramic material for thermal barrier coatings, yttria stabilized z.irconia, while capable of providing significantly greater corrosion resistance than the best metallic attacked by vanadium assisted hot corrosion. Ceria is an alternative stabilizer that may offer improved hot corrosion resistance. Ceramic coating microstructure has also been shown to be very,important to hot corrosion resistance. Porosity in the ceramic may allow salts to penetrate to the metallic bond coat where rapid corrosion may occur. Another failure mechanism which has been observed is spallation on thermal cycling due to the freezing of the salt in the pores of the ceramic. On the other hand, denser ceramic coatings increase the probability of cracking or spallation due to thermal expansion mismatch between the ceramic and the metal substrate.

A David Taylor Naval Ship RAD Center (DTNSRDC) sponsored program for development of ceramic coatings for marine gas turbine first stage hot-section blades and vanes as used in engines of greater than 5000 horsepower has been initiated at GE. The fuels of interest in this program are diesel fuels with higher impurity content than currently allowed in the Navy fuel specification. Anticipated contaminant levels are;

Studies for the U.S. Navy have shown that diesel fuels with these levels of impurities are much more widely available on a worldwide basis than the higher quality fuels currently specified. LM2500 stage 1 high pressure turbine (HPT) blades from two different engines aboard the GTS Admiral William Callaghan are shown in Figure 1. These blades are Rene '80 alloy coated with BC23, a multi-step coating consisting of electron beam physical vapor deposition (PVD) CoCrAl followed by «deposition of aluminum and hafnium in a pack cementation process, and finally an electroplated platinum layer and diffusion heat treatment. The twin shank design blade from engine 901/7 has a nominally 5 mil thick coating and operated for 9225 hours while the blade from engine 807/12 had a nominally 10

mil thick coating and operated for 13,157 hours. The inlet air filtration systems were identical and the engine power profiles were similar. The blade from engine 901/7 had a maximum corrosion rate of 0.25 mils per 1000 hours compared to 1.1 mi's per 1000 hours for the blade from engine 807/12. Although qualitative fuel analyses were not obtained, lower quality fuels (including possibly low levels of vanadium contamination) are believed to be the cause of the increased corrosion rates on engine 807/12. This experience demonstrates the strong dependence of corrosion rates on fuel quality and the relatively severe corrosion that can occur when metallic materials are exposed to vanadium-asssisted hot corrosion. In this paper, two topics relevant to the development of ceramic coatings for hot corrosion protection will be discussed; 1) at sea engine test experience of ceramic coated LM2500 blades in the GTS Call'aghan and 2) initial results of hot corrosion studies of ceramic coatings being developed as part of the DTNSRDC-sponsored program.

Ceramic Coating evaluation of LM2500 Stage 1 HPT Blades

In order to obtain an early assessment of ceramic coatings in an actual gas turbine engine operating in a marine environment, twin shank design LM2500

(a) (b)

Figure 1. BC23 coated LM2500 stage 1 HPT blade from; a) engine 901/7 after 9225 hours service (twin shank design) and b) engine 807/12 after 13,157 hours service (single shank design).

blades were prepared with the standard BC21 (PVD Co-22.5Cr-10.5Al-0.3Y) coating, over which a 1-2 mil PVD yttria stabilized zirconia (YSZ) coating (with 20% yttria) was applied (Figure 2).

The possible limitations of YSZ due to yttria leaching by vanadiunucmpounds or yttria sulfation in high SO environments were recognized; however, YSZ was used since it was available. In this application, vanadium contamination was not expected and fuel sulfur levels were expected to be low enough so that sulfation of yttria would not occur. The PVD process produces a ceramic coating with a columnar structure (Figure 3).

The individual columns are dense and essentially single crystals; however, porosity betwen the columns may permit salts to penetrate to the bond coat resulting in hot corrosion attack and subsequent spallation of the ceramic.

The objectives of testing the YSZ coating were to evaluate its spallation resistance in a marine environment and to determine if salts would penetrate the porosity in the ceramic and lead to hot corrosion of the bond coat. Blade pairs with this coating as well as BC21 and several developmental coatings were tested back-to-back in LM2500 engine 806/10 on the GTS Callaghan for 1077 hours behind a single-stage moisture separator which results in an average sea salt ingestion level of 0.006 ppm in this installation. This is a higher salt level than is typically found on U.S. Navy destroyers (0.002 ppm sea salt); however, this high inlet salt level has been shown to reproduce the corrosion patterns and morphologies found on destroyer engines but at a faster rate. Diesel fuel was used with an average sulfur level (shown to be a very significant factor in the Type 2 corrosion rate) of about 0.4%. An average of 8700 horsepower, is produced by this power profile and results in HPT blade temperatures that are in the low temperature (Type 2) hot corrosion range (1200-1400°F). Most of the remainder of the time the engine was operated at higher power resulting in an overall average of 11,500 horsepower.

Typically, blades with BC21 coating will have a coating life of 3500-4000 hours on the GTS Callaghan with this filtration system; however, borescope inspections revealed significant corrosion attacks on some of the blades. During a scheduled removal of the engine, the blades were removed for careful visual examination. This examination revealed that significant attack had occurred and, therefore, one blade pair of each group was removed for metallurgical eavluation.

Photographs of the stage 1 HPT blade pairs removed for evaluation are shown in Figure 4. The BC21 coated blades had significant corrosion on the pressure side of the airfoils. Visually, the ceramic coated blades were in excellent condition although some spallation and other distress of the ceramic was noted.

The micrographs in Figure 5 show the thumbprint locations at 80% span which is generally the most severely attacked region of the blades. The corrosion morphology commonly referred to as Type 2 corrosion was present. This morphology is characterized by a lack of alloy depletion zone beneath the scale and the presence of a dense, often

Figure 2. BC21 + PVD ceramic coated LM2500 stage 1 HPT blade pair. The ceramic coating has a blue color.

layered, inner scale which is A1 and Cr-rich and a Co-rich outer scale. The corrosion mechanism which produces the Type 2 corrosion in the LM2500 has been shown to be due to the formation of a low melting eutectic CoSO.-Na-SO, salt deposit. The SCL content of the combustion gas is critical in the formation and stability of the CoSO,. The BC21 coating was nearly penetrated (4 mils attack) whereas the BC21 + PVD ceramic coating had no evidence of hot corrosion attack.

A very thin depletion zone, about 2, was present beneath the ceramic coating and the A1203 scale. This depletion zone thickness is comparable to that which was present initially (Figure 3). Careful examination for the presence of sulfides (Figure 6) showed none. Electron microprobe analysis was performed (Figure 7) and no evidence of sulfur was found in the depletion zone. This result suggests that salts have not penetrated through the ceramic even though it is porous. Some caution should be exercised, however, since this engine operated at a lower power most of the time which puts

Figure 3. Microstructure of the PVD BC21 + PVD ceramic (yttria stabilized zirconia) coating on an LM2500 stage 1 blade.

the blade surface temperature below the melting point of Na^SO-. By providing a physical barrier between the outer surface where Na2S0, condenses and the ceramic/metal interface where CoSO^ would form, the low melting point Na2S04-CoS04 eutectic (1150°F) is avoided.This engine test does certainly demonstrate that the PVD YSZ coating provides excellent resistance to Type 2 hot corrosion.

Erosion was observed on the convex side near the tip and, on some blades, has removed all of the ceramic (Figure 9). This is the location on the blade where aerodynamic calculations would predict erosion by large particles to be most severe. Spallation did not appear to be the cause of loss at this location since some blades showed some ceramic remaining here as compared to other locations of the blade where spalling removed all of the ceramic without leaving any evidence of a thin z.irconia layer. The need for some form of erosion enhancement, at least for thing coatings, is clearly suggested by these results.

Hot Corrosion studies of Ceramic Coati ngs

As described previously, a DTNSRDC sponsored program is in progress at GE to develop ceramic coatings for the hot corrosion protection of first stage HPT blades and vanes for use in engines burning diesel fuels contaminated with low levels of vanadium (i.e., up to 5 ppm). Ceria-stabilized zirconia (CSZ) has been proposed for use in this program as a more corrosion resistant material than YSZ based on prior work. Recently, however, R.L. Jones has shown that, at least under certain conditions, CSZ may be more severely attacked by vanadium compounds than YSZ. To explore these conflicting data further, corrosion tests have recently been performed on YSZ and CSZ coatings applied by several methods (Table 1). The as-coated microstructures are shown in Figures 10-14.

The hot corrosion testing was performed in a quartz, glass tube inside a tube furnace at 1650°F. Slowly flowing

02+0.2%S02 gas was passed, at temperature, over a platinum catalyst upstream of the specimens to obtain S0?-S03 equilibrium. The test specimens were 1 diameter by 1/8" thick buttons with the ceramic coating applied onto one side. The?buttons were precoated with 3.0 mg/cm of Na2S04-20%NAV03 on the ceramic coated side. Test durations are listed in Table 2.

The specimens were cooled to room temperature every 15 hours and salt was redeposited on the specimens. It was observed that some salt generally remained on the specimens for the entire 15 hour period. Salt wicked onto the uncoated back side of the buttons causing severe corrosion of the unprotected alloy, thus weight change data were not useful in quantifying the amount of reaction. The amount of reaction was determined qualitatively by visual and scanning electron microscopy (SEM) of the surface and by non-aqueous metallographic techniques. Extent of salt penetration through the ceramic was obtained by observation of ceramic spallation (salt penetration and freezing in porosity or hot corrosion attack of the bond coat are likely to lead to spallation) and by metallography to determine if bond coat hot corrosion had occurred. Results are presented in Table 2.

Visually (Figure 15) the plasma sprayed CSZ and PVD YSZ had very slight spalling at some of the edges; this is likely to be related to the settling of the molten salt (due to gravity) at the bottom edge. However, the PVD CSZ had almost entire spallation. Although surface reaction was apparent, it is not certain whether the spallation is due to hot corrosion or coating processing since the process for CSZ deposition by PVD is not fully developed. The PVD and plasma sprayed YSZ coated specimens had a slight discoloration of the surface. The other specimens had dark areas on the surface.

SEM showed the dark areas to be reaction products and all specimens had undergone some reaction (Figures 16-20). The amount of reaction products on the surface correlated well with the visual observation. Yttrium and vanadium rich

0 0

Post a comment