Creep properties and heat treatment of four experiment al third generation single crystal nickel based superalloys Tedi Marie Usher April 20 th 2010 Materials Science and Engineering Dr. Gerhard Fuchs Committee Chair Dr. Michele Manuel and Dr. Gerald Bourne Committee Members
1 Abstract The effectiveness of solution and aging heat treatments was confirmed for four variations of a third generation single crystal nickel based superalloy. The four variations include a baseline alloy a variation with molybden um substituted for tungsten, a variation with titanium substituted for tantalum, and a variation with a 1.6wt% ruthenium addition. Creep tests at 950C and 320 MPa were performed on each alloy variation. The two alloy variations with lower density substitutions showed s horte r creep lifetimes than the baseline and the alloy variation with added ruthenium showed an increased creep lifetime. This is in agreement with other work that demonstrates the ability of ruthenium to act a sol increase the creep lifetime. Ruthenium has also been shown to suppress the f ormation of topologically close packed (TCP) phases which are undesirable however no TCP phases were observed in this work. The mic rostructure of each alloy variation experienced rafting, or directional coarsening, perpendicular to the stress direction during creep. Alloy variations with longer lifetimes showed a higher degree of rafting, which is supported by evidence in the literat ure that at high temperature and low stress rafting increases creep lifetimes. Background Nickel based s uperalloys are used in many high temperature applications including industrial gas turbine engines and jet engines. Due the high mechanical and therm al demands on these alloys, they have become compositionally and microstructurally complex in order to have appropriate strength. These alloys can contain up to ten different elements, and it is still not entirely understood how the proportion of each ele ment affects the properties of the alloy. Superalloy development is often characterized by or general composition changes, in which the rhenium content has increased from 0 wt % to 3 wt %, to 6 wt %. The 6 wt % rhenium in third generat ion superalloys increases the high temperature performance but decrease s the microstructural stability of the alloy Topologically close packed (TCP) phases are known to form in third generation super alloys due to the high rhenium content and are detrime ntal to the mechanical properties . In order to suppress the formation of TCP phases, some experimental alloys have a small (1 3 wt %) percentage of ruthenium added to them. These alloys are sometimes referred to as fourth generation superalloys. Ruthe nium is thought to suppress the formation of unfavorable TCP phases while maintaining or increasing the strength associated with third generation alloys  High density is another issue for all superalloys, and in particular third generation superalloy s due to the higher rhenium content These alloys often have ve ry high densities, near 9.0 g/cm. Any composition changes that result in lower density and comparable strength would be beneficial to the weight and efficiency of the engines in which the su peralloys are use d Some new experimental alloys that use molybdenum as a solid solution strengthener instead of tungsten to reduce the density of the alloys have show n promising results . The four alloy variations investigated in this research includ e a baseline alloy, two alloys that are lower density due to the substitution of less dense elements and an alloy with added 1.6 wt % ruthenium
2 Table 1: Alloy variation compositions Changes from baseline noted in red. Ni Cr Co Mo W Ta Re Al Ti Hf Ru LMSX 1 57.70 4.1 12.2 0 5.85 8.6 5.9 5.55 0 0.1 0 LMSX 7 58.85 4.1 12.2 1.6 3.1 8.6 5.9 5.55 0 0.1 0 LMSX 14 60.20 4.1 12.2 0 5.85 6.0 5.9 5.65 0.8 0.1 0 LMSX 16 56.10 4.1 12.2 0 5.85 8.6 5.9 5.55 0 0.1 1.6 Technical Approach S olution heat treatme nts were performed to homogenize the alloy variations since a significant amount of segregation between the dendrites and inter dentritic regions exist ed in the as cast condition The solution heat treatments were performed using an Elatec high temperatur e vacuum furnace which is capable of temperatures of 1400 C and vacuums greater than 10 4 torr. The heating elements and hearth plate are made of graphite and the samples are place d on alumina trays to separate them. The temperature wa s controlled by tw o Type C (W + 3%Re / W + 25%Re) thermocouples which were held near the sampl es. After the heat treatment wa s complete, the samples were quenched with high purity helium circulated with a water cooled copper heat exchanger that can cool at rates of 250 C /min. Below is the heating schedule used for the solution heat treatment. Table 2 : Solution heat treatment schedule Segment Type Start temp (C) Time (mins) or ramp rate (C/min) 1 soak 23 1 min 2 ramp 23 15 C/min 3 soak 1265 60 min 4 ramp 1265 10 C/min 5 soak 1280 120 min 6 ramp 1280 10 C/min 7 soak 1285 120 min 8 ramp 1285 10 C/min 9 soak 1290 120 min 10 ramp 1290 10 C/min 11 soak 1295 120 min 12 ramp 1295 10 C/min 13 soak 1300 120 min 14 ramp 1300 10 C/min 15 soak 1305 120 min 16 ramp 1305 10 C/min 17 soak 1310 600 min 18 ramp 1310 10 C/min 19 soak 1325 600 min 20 ramp 1325 10 C/min 21 soak 1335 600 min 22 ramp 1335 100 C/min (cooling) 23 soak 100 1 min Total time 44.5 hours
3 The aging heat treatment wa s conducted t o develop the to the desired size It was performed in Carbolite box furnaces with maximum temperatures of 1300 C in ambient atmosphere. The temperature was controlled with two Type K thermocouples in direct contact with the samples. After the heat t reatment, the samples were air cooled on alumina racks outside of the furnace. Cooling rates were estimated to be greater than 100 C /min. Table 3 : Aging heat treatment schedule Creep tests were performed to asses s the high temperature performance of the alloys under constant stress, which is similar to conditions experience d in service. The creep tests were performed i n Satec M 3 type creep frames at 950 C and 320 MPa. The frames have 16 1 lev er arm ratios and the tests were performed at constant load. The threads on the sample and the frame were lubricated with boron nitride high temperature lubricant. The elongation wa s measured using extensometers with set screws in the shoulders of the sa mples. The samples were surrounded by a clamshell furnace which is capable of temperatures from 1200 C to 1300 C The temperature was controlled by three K type thermocouples that were attached to the gauge section of the sample with 24 gauge ( 80Ni 2 0Cr ) wire. After the temperature ramp ed up to the test temperature, the sample wa s allowed to soak for one hour to ensure temperature uniformity It wa s then loaded in multiple steps so that the hot modulus could be determined. The frame, furnace, and d ata collection wa s controlled by Nuvision Mentor Creep software. The samples were prepared for metallography according to the ASM Metals Handbook. The samples were sectioned by an Allied slow cut diamond saw. After mounting, the samples were leveled w using alumina media. To observe the microstructure, the samples were etched with a etchant developed by Pratt and Whitney which consists of 100mL HCl, 100mL HNO 3 10g MoO 2 and 100mL H 2 O The etchant was appli ed with a cotton tipped applicator and swabbed until the sample wa s hazy. A JEOL 6400 scanning electron microscope (SEM) wa s used to characterize the samples after heat treatments and to examine longitudinal sections and fracture surfaces from the creep t ests. Results and Discussion Creep Data Figure 1 shows the creep curves of the four alloy variations. Each curve exhibits the same general shape, showing a very small amount of primary creep, a region of secondary or steady state creep, and finally tertiary creep where the creep rate accelerate s to failure. The steady state creep rate has a clear effect on the creep lifetime of the alloy variation, as the alloy variations with lower steady state creep rates have longer lifetimes as shown in figure 2. Time (hours) Temperature (C) 4 1150 24 870 30 760
4 Figure 1 : Creep behavior of all four alloy variations. Figure 2: Creep rates of the four alloy variations. The alloy variation with added ruthenium (LMSX 16) performed the best at these conditions. Ruthenium has been shown to increase t he creep performance of Ni based superalloys [2, 8]. Two mechanisms have been proposed to explain this effect. Ruthenium is known to suppress the formation of topologically closed packed (TCP) phases, which are deleterious to the creep strength of supera lloys [1, 2, 8]. It also acts as a sol id solution this work, no TCP phases were observed in any of the alloy variations, including the alloy to suppress TCP formation cannot be commented on at this time. Long term elevated temperature exposures would have to be LMSX 1 LMSX 1 6 LMSX 7 LMSX 1 4 LMSX 1 LMSX 1 6 LMSX 7 LMSX 1 4
5 performed, and would be an appropriate direction for future work. However, this work is in agreement with other results that show ru third generation superalloys. LMSX 16, the alloy variation with added 1.6wt% Ru, shows an 87 hour or almost 20% creep lifetime advantage over the baseline alloy. Figure 3 ime to failure for each alloy. The alloy vari ation with a small amount of molybdenum substituted for tungsten performed the worst at this condition. Recent work by MacKay et al (2010) suggests that lower density high performance superalloys can be design ed with molybdenum instead of tungsten only 3wt% rhenium low chromium levels, about 6wt% tantalum and aluminum and no titanium. These alloys use 7 10 wt % molybdenum instead of about 5 6 wt% tungsten as a solid solution strengthener as in CMSX 4, a com mon second generation superalloy [3 7 ]. In LMSX 7 the 5.85wt% tungsten in the baseline, LMSX 1, is replaced with 1.6% molybdenum and 3.1wt% tungsten. LMSX 7 had a significantly lower creep lifetime than the baseline as shown in figure 3 This indicat es that small amounts of molybdenum cannot directly substitute for tungsten. In the work by MacKay et al they used molybdenum instead of tungsten and their alloys showed sufficient strength  This contradiction can perhaps can be explained by their us e of only molybdenum and no tungsten, and higher weigh t percentages of molybdenum (9wt%) than previous levels of tungsten (5 6wt%). In LMSX 14, 8.6wt% tantalum was replaced with 6wt% tantalum and 0.8wt% titanium This caused a decrease in the creep life time from the baseline of 439 hours to 361 hours, which is a decrease of 78 hours or 18%. T itanium and tantalum both partition to the phase. The relative amounts of each likely impact phase boundary (APB) energy. However, t hese effects are outside the scope of this project.
6 Metallography results The as cast structure of all four alloy variations sho wed segregation between the dendritic and inter dendritic regions, as shown in figure 4 The inter dendritic regions showed large areas with lower melting points The inter The solution heat treatment was successful in homogenizing the microstructure of all four alloys, as shown in figure 5 No eute ctic areas or incipient melting was seen in any of the alloy as indicated in figure 5 . The aging heat treatment also achi eved the desired microstructure, as s hown in figure 6. more cuboidal and appeared to be more aligned with each other after the aging heat treatment. Figure 4 : SEM images of LMSX 1 (a), LMSX 7 (b), LMSX 14 (c) and LMSX 16 (d), in the as cast condition. Original magnification: 1000x
7 Fig ure 5 : SEM images of LMSX 1 (a), LMSX 7 (b), LMSX 14 (c) a nd LMSX 16 (d) in the Original magnification: 10,000x Figure 6 : SEM images of LMSX 1 (a), LMSX 7 (b), LMSX 14 (c) and LMSX 16 (d) in the aged heat treated condition. Original magnification: 10,000x 1 2
8 The fracture surfaces of all four alloy variations ess entially showed the same morphology, a shown in figure 9 with a pore or creep cavity in the center, which indicates that pores or creep cavities acted as the main top view of a crack that began at a pore or creep cavity While difficult to quantify, the facets of the alloy variations with short er creep lifetimes tended to be larger, as shown in figure 7 A possible explanation is that the cracks grew faster in the alloy variations with shorter creep lifetimes, while in the alloy variations with longer lifetimes, the cracks grew slower, so more were able to nucleate before failure. Longitudinal sections of the failed samples were cut and polished to be exa mined in the SEM. These images clearly show that pores or creep cavities acted as nucleation sites for cracks, which appeared as facets on the fracture surfaces. Almost all cracks appeared to begin at a pore or a creep cavity. Figure 8 shows an excellen t example of a group of pore s acting as a nucleation point for a crack. The longitudinal sections also show varying amounts of rafting, which is also called directional coarsening, and is shown in figure 10 In this process te at 90 to the stress direction, which is vertically oriented in figure 10 The alloy variations with longer creep lifetimes, LMSX 1 and LMSX 16, show greater amounts of rafting than the alloy variations with shorter lifetimes, LMSX 7 and LMSX 14. There is not a consensus in the literature on whether or not rafting results in improved creep properties [5,6]. Kamaraj (2003) proposes a theory in which at low temperature (< 950C) and high stress (LTHS), the dis precipitates so rafting decreases the creep lifetime, but at high temperature ( > 1000C) and low stress (HTLS), rafting increases the path the dislocation must take a nd improves the creep life. The HTLS theory seems to best represent the results seen in this work. Figure 7 : Creep fracture surfaces of a) LMSX 14, which lasted 361h and b) LMSX 1 6, which lasted 526 h. Original magnification: 250x Figure 8 : Longitudinal section of LMSX 7 post creep test showing crack initiation at pores. Original magnification: 1000x
9 Figure 9 : SEM images of typical fracture surfaces. LMSX 1 (a), LMSX 7 (b), LMSX 14 (c) and LMSX 16 (d). Original magnification: 1,000x Figure 10 : SEM images of longitudinal sections near fracture surface showing rafting. LMSX 1 (a), LMSX 7 (b), LMSX 14 (c) and LMSX (d) Original magnification: 7,500x
10 Conclusions The solution and aging heat treatments were effective in solutionizing each of the alloy variations without any incipient melting or remaining eutectic areas and developing the Small ruthenium additions of 1.6wt% increase the creep lifetime of the baseline alloy by acting as a solid solution strengthener. TCP phases were not observed in any alloy so the ability of ruthenium to suppress their formation cannot be evaluated. Molybdenum cannot directly replace tungsten as a solid solution strengthener at low levels Replacing all of the tungsten with a higher weight percentage of molybdenum has shown promising results in the literature. Replacing 2.6wt% tantalum with 0.8% titanium decreases the creep lifetime of the alloy at the tested condition. Each alloy variation showed the formation of raft s perpendicular to the stress direction during creep. A lloy variations with higher degrees of rafting had higher creep lifetimes. Acknowledgements The author would like to first acknowledge Dr. Gerhard Fuchs for advising and supporting this project. The author would like to thank Andrew Wasson and the rest of the High Temperature Alloys group for their help and advice. The author would also like to acknowledge the Major Analytical Instrumentation Center, Materials Science and Engineering Department, University of Florida
11 References 1.) Topologically close packed phases in an experiment rhenium containing single crystal superalloy. (2000). C.M.F. Rae, M.S.A. Karunaratne, C.J. Small*,R.W. University of Cambridge / Rolls Royce University Technology Centre 2.) Microstructural stability and creep of Ru containing Ni base superalloys (2004). L.J. Rowland, Q. Feng, and T.M. Pollock University of Michigan, Dept. of Materials Science and Engineering 3.) MacKay, R.A., Gabb, T.P., Smialek, J.L., & Nathal, M.V. (2010). A n ew approach of designing superalloys for low density. JOM: Journal of the Minerals, Metals and Materials Society, 62(1), 48 54. 4.) precipi tation in cmsx 6 monocrystalline nickel base superalloy. Materials Characterization, 60, 1114 1119. 5.) Tinga, T., Brekelmans, W.A.M., & Geers, M.G.D. (2009). Directional coarsening in nickel base superalloys and its effect on the mechanical properties. C omputational Materials Science, 47, 471 481. 6.) Kamaraj, M. (2003). Rafting in single crystal nickel base superalloys an overview. Sadhana, 47, 115 128. 7.) Matan, N., Cox, D.C., Rae, C.M.F., & Reed, R.C. (1999). On the kinetics of rafting in cmsx 4 superalloy single crystals. Acta mater, 47(7), 2031 2045. 8.) Hobbs, R.A., Zhang, L., Rae, C.M.F., & Tin, S. (2007). The Effect of ruthenium on the intermediate to high temperature creep response of high refractory content single crystal nickel base super alloys. Materials Science and Engineering A, 489, 65 76.
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EZG99B127_I9R3BV INGEST_TIME 2017-06-01T19:33:36Z PACKAGE AA00057689_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC