Copyright ©JCPDS - International Centre for Diffraction Data 2005, Advances in X-ray Analysis, Volume 48.

FATIGUE-INDUCED PHASE FORMATION AND ITS DEFORMATION BEHAVIOR IN A COBALT-BASED SUPERALLOY M. L. Benson1, T. A. Saleh1, P. K. Liaw1, H. Choo1,2, D. W. Brown3, M. R. Daymond4, X.-L. Wang5, A. D. Stoica5, R. A. Buchanan1, and D. L. Klarstrom6 1. Department of Materials Science and Engineering, The University of Tennessee, Knoxville, TN 37996, USA 2. Metals and Ceramics Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA 3. Los Alamos Neutron Science Center, Los Alamos National Laboratory, Los Alamos, NM 87545, USA 4. Department of Mechanical and Materials Engineering, Queen’s University, Kingston, ON K7L3N6, Canada 5. Spallation Neutron Source, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA 6. Haynes International, Inc., Kokomo, IN 46904, USA ABSTRACT The low-cycle fatigue behavior of a cobalt-based superalloy was studied in-situ using neutrondiffraction experiments. The alloy exhibited a strain-induced formation of a hexagonal-closepacked (hcp) phase within its parent face-centered-cubic (fcc) phase at ambient temperature under strain-controlled fatigue conditions with a total strain range, ∆ε = 2.5%. The (101) hcp peak was first observed during the 12th fatigue cycle under the given conditions following an incubation period, during which no hcp phase was detected. Subsequently, the intensity of the hcp peaks increased as fatigue progressed. Furthermore, within a single fatigue cycle, the intensity of the (101) hcp peak decreased during the compression half-cycle and increased again when the specimen was subjected to a subsequent tensile strain. The result suggests that the fcc to hcp transformation is partially reversible within one fatigue cycle. INTRODUCTION The ULTIMET® alloy is a cobalt-based superalloy manufactured by Haynes International, Inc. The nominal composition of the alloy is given in Table 1 [1]. Typical applications of the material include agitators, blenders, spray nozzles, screw conveyors, and valve parts in corrosive and high-wear environments [1]. ®a Table 1. The composition of the ULTIMET lloy given in weight percent [1].

Co

Cr

Ni

Mo

Fe

W

Mn

Si

N

C

26 As balance

9

5

3

2

0.8

0.3

0.08

0.06

a

54

a

Figure 1 shows the binary-phase diagram of the Co-Cr system, which approximates the behavior the ULTIMET® alloy [2]. The alloy was solution-heat treated in the fcc α-Co region of the phase diagram and, then, quenched to room temperature. While the phase diagram suggests the presence of a two-phase field of the hcp ε-Co and the tetragonal σ-Co at 26 weight percent

123

This document was presented at the Denver X-ray Conference (DXC) on Applications of X-ray Analysis. Sponsored by the International Centre for Diffraction Data (ICDD).

This document is provided by ICDD in cooperation with the authors and presenters of the DXC for the express purpose of educating the scientific community. All copyrights for the document are retained by ICDD. Usage is restricted for the purposes of education and scientific research. DXC Website – www.dxcicdd.com

ICDD Website - www.icdd.com

Copyright ©JCPDS - International Centre for Diffraction Data 2005, Advances in X-ray Analysis, Volume 48.

chromium (dotted vertical line) below 900oC, only the fcc phase was observed in a metastable state upon quenching. No evidence of the σ phase has been reported in the solution-treated alloy in previous studies [3-8]. It is expected that the σ phase would form if the alloy were held for a sufficient length of time at elevated temperatures. On the other hand, the hcp phase may form via a strain-induced phase transformation at room temperature. Jiang et al. have shown through X-ray diffraction studies of fatigued specimens that the hexagonal phase will indeed form after low-cycle fatigue [3-7]. However, the influence of the developing hcp phase during mechanical deformation on the lifetime of the alloy has not been quantified to date. Therefore, in-situ neutron-diffraction studies were conducted during low-cycle fatigue in order to characterize the strain-induced phase transformation.

fcc

hcp Figure 1. Equilibrium phase diagram of the binary Co-Cr system [2]. EXPERIMENTAL DETAILS The in-situ low-cycle fatigue was performed using the ENGIN-X instrument at the ISIS facility (United Kingdom), which is a pulsed polychromatic neutron source. Cylindrical, threaded-end specimens were machined from plate stock with an 8 mm gage diameter and a 24 mm gage length. The stock material was processed by (a) hot-rolling at a temperature of 1,204oC, (b) annealing at 1,121oC in air, and (c) water quenching to room temperature. No initial texture was present in the specimen. The ENGIN-X load frame was used to perform a strain-controlled fatigue experiment at room temperature using R = -1, where R is the ratio of the minimum strain (εmin.) to the maximum strain (εmax.). A total strain range of ∆ε = εmax. - εmin. = 2.5 % was applied to the specimen. The tensile specimen was oriented 45o to the incident neutron beam with the scattering angle fixed at 2θ = ±90o for two detector banks [9]. The scattering geometry of ENGIN-X allows for the measurements of diffraction patterns from grains with the crystallographic plane-normal vectors oriented parallel to the loading direction of the specimen (axial data) and with the crystallographic plane-normal vectors oriented perpendicular to the loading direction of the specimen (transverse data). Neutron-diffraction patterns were measured

124

Copyright ©JCPDS - International Centre for Diffraction Data 2005, Advances in X-ray Analysis, Volume 48.

at 10 different cycles throughout the fatigue experiment: 0 (i.e., as received), 1, 4, 8, 12, 30, 50, 75, 100, 250, and 500 cycles. For a given cycle, six diffraction patterns were measured around the hysteresis loop, as shown in Figure 2. These points coincide with the maximum tensile and compressive strain points (points 1 and 4), the two zero-strain points (points 3 and 6), and the two zero-stress points (points 2 and 5). Cobalt exhibits significant incoherent neutron scattering, so forty-minutes were required to obtain appropriate statistics for each diffraction pattern, with a sampling volume of about 120 mm3. 1

6

5

4

2

3

Figure 2. Macroscopic stress-strain curves measured during the low-cycle fatigue. In-situ diffraction patterns were measured at six different points around the hysteresis loop during fatigue cycles 1, 4, 8, 12, 30, 50, 75, 100, 250, and 500. RESULTS AND DISCUSSION HEXAGONAL PHASE DEVELOPMENT The result to be discussed first is the development of the hcp phase during low-cycle fatigue. Diffraction patterns taken at “point 1” on each cycle are overlayed in Figure 3 for the axial detector bank. The (101) hcp peak did not form immediately at the first fatigue cycle, but it was first observed during cycle 12. Subsequently, the intensity of the peak increased continuously as the low-cycle fatigue progressed. Figure 4 shows the integrated intensity of the (101) hcp peak as a function of fatigue cycles on a semi-log scale. It shows that the (101) diffraction peak did not develop appreciable intensity before cycle 12. Following cycle 12, however, the intensity of the hcp peak increased almost linearly on the semi-log scale. The initial fatigue cycles, during which no hcp phase was observed, are considered as an incubation period. Similar behavior has been documented in austenitic stainless steels, where the metastable fcc phase shows an incubation period before partially transforming into a martensitic phase during low-cycle fatigue [10]. Based on the deformation behavior at room temperature, the stacking fault energy (γSFE) of the ULTIMET® alloy is likely less than γSFE = 20 mJ/m2 [11]. Therefore, perfect dislocations dissociate readily into Shockley partial dislocations in the fcc matrix, introducing a band of stacking faults [12]. Previous work on Co-based alloys has shown that the strain-induced phase transformation is actually the result of the coalescence of deformation stacking faults [13]. The observed incubation period could be the result of the nucleation of stacking fault clusters, as a precursor to the coalescence process.

125

Copyright ©JCPDS - International Centre for Diffraction Data 2005, Advances in X-ray Analysis, Volume 48.

The previous discussion was concerned with the evolution of the hcp phase as a function of fatigue cycles. Within a given fatigue cycle, however, another interesting observation regarding the hcp phase evolution can be made [see Figure 5(a)]. The intensity of the (101) hcp peak continuously decreases in intensity from “point 1” (maximum tension) to “point 4” (maximum compression) in the hysteresis loop (see Figure 2). The integrated intensity of the peak then increases again, moving from “point 4” to “point 6”. When the intensity is plotted as a function of the applied macroscopic strain of cycle 30, Figure 5(b), a linear correlation is suggested. Figure 5 displays data from both the axial and transverse detector banks. The fact that similar trends in the intensity evolution are noted in both detector banks suggests that the intensity changes are associated with the development of the volume fraction of the hcp phase during a single cycle, as opposed to the development of crystallographic texture within the hcp phase. If the effect was due to the grain rotation (i.e., texture or twinning), then the transverse data should show the opposite trend as compared to the axial data. Therefore, based on the present results, it is hypothesized that the transformation is partially reversible within one fatigue cycle. 111

311

- hcp 103

220

- fcc 102

200 101 Cycles

500 250 100 75 50 30 12 8 4 1 as-received

Figure 3. The overlay of diffraction patterns (measured at “point 1”) from the axial detector bank as a function of fatigue cycles, showing the development of the hcp phase.

500

Figure 4. Intensity evolution of the (101) hcp peak measured at “point 1” as a function of fatigue cycles.

126

Copyright ©JCPDS - International Centre for Diffraction Data 2005, Advances in X-ray Analysis, Volume 48.

127

CUBIC PHASE RESPONSE The diffraction data indicated that the development of the hcp phase had a significant effect on the parent fcc phase. Figure 6 shows the relative intensity changes of the fcc peaks at “point 1” (axial data) as a function of fatigue cycles, analogous to Figure 4.

Point 5 Point 4

Points 3,6

Point 1 Point 2

(a) (b) Figure 5. Intensity evolution of the (101) hcp peak during fatigue cycle 30: (a) as a function of diffraction points and (b) as a function of macroscopic strain. The diffraction points of (a) are labeled on (b). 200

311 111 220

500

Figure 6. Relative intensity changes of the fcc peaks from “point 1” diffraction patterns measured from the axial detector bank. The data in Figure 6 are normalized with respect to the intensity measured during cycle 1 for each respective fcc peak and, therefore, represent changes in the intensities as a function of fatigue cycles. The intensities of the (200), (311), and (111) fcc Bragg reflections do not change dramatically with increasing number of fatigue cycles. The intensity of the (111) peak, which demonstrated the most variation of the three, only decreased to 85 % of its original value by cycle 500. However, the (220) diffraction peak intensity decreased to 55 % of its original value by cycle 500. Such a large decrease in the intensity suggests that the fcc grains with the (220) plane-normal vectors oriented parallel to the axial direction of the specimen are preferentially

Copyright ©JCPDS - International Centre for Diffraction Data 2005, Advances in X-ray Analysis, Volume 48.

transforming into the new phase during the fatigue. Similar to the reasoning used here, Brown et al. used relative intensity changes in U-Nb alloys in order to identify the specifically-oriented grains that twinned during monotonic loading measurements [14]. CONCLUSIONS Neutron-diffraction studies during in-situ cyclic loading with ∆ε = 2.5% showed the straininduced hcp phase formation in a fcc cobalt-based superalloy. The new phase was not immediately observed until cycle 12 under the given low-cycle fatigue conditions. Following the incubation period, the hcp peak intensity increased continuously with fatigue cycles. Within a single fatigue cycle, the transformation was apparently partially reversible. However, further work involving microscopy techniques is required to verify this phenomenon. The intensity of the (220) fcc peak decreased substantially during low-cycle fatigue. This indicates that the grains containing the (220) plane-normal vectors oriented parallel to the loading direction of the specimen were preferentially transforming into the new hcp phase. ACKNOWLEDGEMENT The author acknowledges the financial support from the National Science Foundation (NSF), the Combined Research-Curriculum Development (CRCD) Programs, under EEC-9527527 and EEC-0203415, and the Integrative Graduate Education and Research Training (IGERT) Program, under DGE-9987548, and the International Materials Institutes (IMI), under DMR0231320, to the University of Tennessee (UT), Knoxville, with Ms. M. Poats, Dr. W. Jennings, Dr. L. Goldberg, Dr. L. Clesceri and Dr. C. Huber as program directors, respectively. Also acknowledged with gratitude are the additional funds that are provided by the Tennessee Advanced Materials Laboratory (TAML), which is under the direction of Professor E. W. Plummer. In addition, the assistance of Mr. D. Feilden of UT was invaluable for the completion of this project. This work benefited from the use of the ISIS facility, which is located at the Rutherford-Appleton Laboratory in the United Kingdom. REFERENCES [1] [2] [3] [4] [5] [6] [7]

www.haynesintl.com Binary Alloy Phase Diagrams, ASM, Metals Park, OH, 1986, vol. 1, p. 759. L. Jiang, C. R. Brooks, P. K. Liaw, H. Wang, C. J. Rawn, and D. L. Klarstrom, Mat Sci Eng A, 314, 2001, pp. 162-175. L. Jiang, C. R. Brooks, P. K. Liaw, D. L. Klarstrom, C. J. Rawn, and B. Muenchen, Mat Sci Eng A, 316, 2001, pp. 66-79. L. Jiang, H. Wang, P. K. Liaw, C. R. Brooks, and D. L. Klarstrom, Mechanics of Materials, 36, 2004, pp. 73-84. L. Jiang, H. Wang, P. K. Liaw, C. R. Brooks, and D. L. Klarstrom, Met Trans A, 32, Sept. 2001, pp. 2279-2296. L. Jiang, C. R. Brooks, P. K. Liaw, J. Dunlap, C. J. Rawn, R. A. Peasoe, and D. L. Klarstrom, Met Trans A, 35, March 2004, pp. 785-796.

128

Copyright ©JCPDS - International Centre for Diffraction Data 2005, Advances in X-ray Analysis, Volume 48.

[8] [9] [10] [11] [12] [13] [14]

M. Benson, T. A. Saleh, P. K. Liaw, H. Choo, D. W. Brown, M. R. Daymond, X.-L. Wang, A. D. Stoica, R. A. Buchanan, and D. L. Klarstrom, presentation at the American Conference on Neutron Science 2004, June 6-10, College Park, MD. www.isis.rl.ac.uk/engineering/ ASM Handbook, Vol. 19. ASM International, 1997, p. 86. H. Farhangi, R. W. Armstrong, and W. F. Regnault, Mat Sci Eng, Vol. 114, 1989, pp. 35-38. G. E. Dieter, Mechanical Metallurgy, 3rd ed., McGraw-Hill, 1986, pp. 135-136. H. M. Tawancy, V. R. Ishwar, and B. E. Lewis, J. Mat Sci Let, 5, 1986, pp. 337-341. D. Brown, M. Bourke, P. Dunn, R. Field, M. Stout, and D. Thoma, Met Trans A, 32, Sept. 2001, pp. 2219-2228.

129