Dynamic steady-state tertiary creep in a Nickel-base single crystal superalloy at high temperatures

Abstract

Nickel-base single crystal (SX) blades, oriented along the [0 0 1] crystallographic direction, were introduced for gas turbine engines in 1980 [1, 2]. Creep resistance of SX materials is derived from the large volume fraction, about 70%, of ordered cubical γ ′ precipitates in a disordered γ matrix. Since the γ /γ ′ morphology changes during creep, considerable attention has been paid to this topic [1, 3]. Because SX materials exhibit high stress sensitivity and large elongation, the increase in stress, due to the reduction in the cross-sectional area during creep, must control the strain rate and the time to rupture for constant load tests. The objective of this letter is to present our study on this subject. Both experimental and analytical techniques presented here are applicable in general to other materials exhibiting high ductility. The “third generation” CMSX-10 with a nominal composition (wt.%) of 2Cr, 3Co, 0.4Mo, 6Re, 5W, 5.7Al, 0.2Ti, 8Ta, 0.1Nb, 0.03Hf, balance Ni [2] was used. Creep specimens, with a nominal diameter of 4.0 mm and a uniform gauge section of 25 mm, were prepared from fully heat-treated [0 0 1]-oriented bars. The γ ′ precipitates had average edge dimensions of about 0.5 μm and there were rows of spherical pores, up to 25 μm in diameter, aligned parallel to the growth direction. Six constant load tests, for an initial stress σ i of 700 MPa, were performed in air at 1073 to 1273 K (800– 1000◦C) ± 1 K using a single lever dead-load system, a three-zone furnace and an extensometer. Out of these six tests, two rupture tests were performed first at 1123 K and 1173 K to determine the fracture times and strains (19% and 18% at 1123 K and 1173 K respectively). Four strain relaxation (creep) and recovery tests (SRRT) [4] for relaxation times, trlx required to reach a total strain of 15% (close to fracture), were then performed at 1073 K, 1148 K, 1223 K and 1273 K. SRRT involves the application of the load as rapidly as possible, deforming a specimen for trlx, and then removing the load fully and rapidly, but keep monitoring the strain for a long time until no further strain recovery is noticed. SRRTs allow determination of elastic, delayed elastic and viscous components of strain as functions of trlx. Only the strain relaxation results of these tests will be presented and discussed here. Fig. 1 shows all the strain relaxation curves. The average value of the bulk Young’s modulus, E = 110 GPa, obtained from the initial elastic strain, ee, agreed with the expected range for the [0 0 1]-orientation at these temperatures [3, 5]. This allowed the estimation of creep strain, ec (= e − ee), where e is the total strain. The high stress of 700 MPa, equivalent to 6.4 × 10−3 E, was chosen to avoid complexities of γ ′ rafting [6, 7]. The fracture strains for the two tests bear strong similarities with those of the “second generation” alloy CMSX-4 for lower stresses, 130 to 400 MPa, at 1273 K [7]. Assuming incompressibility during creep, the actual stress, σA was estimated from σA = σ i (1 + ec). This gives, for example, the fracture stress, σAf of 833 MPa for the test at 1123 K. It would be appropriate to mention here that the specimen cross-sectional shape after the tests was found to be slightly elliptical. For example, the majorand the minor-axis of the specimen tested at 1148 K was 3.76 mm and 3.70 mm respectively near the middle of the specimen – giving a small but measurable eccentricity of 0.183. This does not necessarily mean asymmetry in the load axis and, in fact, is expected due to a number of active slip systems in cubic materials, as reported for SX alloy SRR99 in [8]. Fig. 2 shows the strain rate-time, e − t and σA − t curves for the test at 1223 K (950 ◦C). It shows that e changes with time in a complex manner similar to observations in CMSX-4 [9, 10]. However, the monotonically increasing σA suggests that the conventional approach of plotting eec curves for analytical purposes (assuming constant stress) violates the basic principles of experimental physics. The relationship between e and σA for the results in Fig. 2 and plotted in Fig. 3 clearly shows that the ‘tertiary’ or the ’time-wise accelerating’ stage represents a “stresswise dynamic steady state” (broken line in Fig. 3). Note the high regression coefficient, R2 of 0.997. Similar observations were also made at other temperatures (Fig. 4). The test at 1173 K shows some inconsistency due to some