Investigation of the Use of Laser Shock Peening for Enhancing Fatigue and Stress Corrosion Cracking Resistance of Nuclear Energy Materials

Project No. 10-682

Fuel Cycle/Reactor Concepts Mission Relevant Investigator Initiated Research

Vijay K. Vasudevan University of Cincinna

Sue Lesica, Federal POC Sebasen Teysseyre, Technical POC Final Report

Project Title: Investigation of the Use of Laser Shock Peening for Enhancing Fatigue and Stress Corrosion Cracking Resistance of Nuclear Energy Materials Technical Workscope: MR-IIR Covering Period: October 1, 2010 – June 30, 2016 Date of Report: March 7, 2017 Recipient: Name: University of Cincinnati Street: 2600 Clifton Ave. City: Cincinnati State: Ohio Zip: 45221

Contract Number: 102835 Project Number: 10-682 Principal Investigator: Vijay K. Vasudevan - (513) 556-3103 - [email protected] Abhishek Telang (PhD student); Chang Ye (Postdoctoral Fellow); S. R. Mannava and Dong Qian Collaborators: Sebastien Teysseyre, John Jackson (INL), B. Alexandreanu, Yiren Chen (ANL) Project Objective: The objective of this project, which includes close collaboration with scientists from INL and ANL, is to investigate and demonstrate the use of advanced mechanical surface treatments like laser shock peening (LSP) and ultrasonic nanocrystal surface modification (UNSM) and establish baseline parameters for enhancing the fatigue properties and SCC resistance of nuclear materials like nickel-based alloy 600 and 304 stainless steel. The research program includes the following key elements/tasks: 1) Procurement of Alloy 600 and 304 SS, heat treatment studies; 2) LSP and UNSM processing of base metal and welds/HAZ of alloys 600 and 304; (3) measurement and mapping of surface and sub-surface residual strains/stresses and microstructural changes as a function of process parameters using novel methods; (4) determination of thermal relaxation of residual stresses (macro and micro) and microstructure evolution with time at high temperatures typical of service conditions and modeling of the kinetics of relaxation; (5) evaluation of the effects of residual stress, near surface microstructure and temperature on SCC and fatigue resistance and associated microstructural mechanisms; and (6) studies of the effects of bulk and surface grain boundary engineering on improvements in the SCC resistance and associated microstructural and cracking mechanisms. TPOC: [email protected] Federal reviewers: [email protected] TABLE OF CONTENTS

Section Page #

Cover Page……………………………………………………………………………….. 1

1. Executive Summary ………………………………………………………………………. 4

2. Introduction ……………………………………………………………………………… 7

3. Effects of Laser Shock Peening on SCC Behavior of Alloy 600……………………………. 13

3.1 Introduction…………………………………………………………………………… 13 3.2 Experimental…………………………………………………………………………… 16 3.3 Results and Discussion…………………………………………………………………. 19 3.3 Conclusions…………………………………………………………………………….. 30

4. Effects of Laser Shock Peening on SCC Behavior of Alloy 600 in Tetrathionate Solution. 34 4.1 Introduction…………………..……………………………………………………….. 34 4.2 Experimental…………………………………………………………………………… 37 4.3 Results…………………………………………………………………………………. 43 4.4 Discussion………………………………………………………………………………. 56 4.5 Conclusions…………………………………………………………………. 60

5. Surface Grain Boundary Engineering of Alloy 600 for Improved Resistance to Stress Corrosion Cracking………………..……………………………………………………….. 63 5.1 Introduction…………………..……………………………………………………….. 64 5.2 Materials and Methods…………………………………………………………………. 65 5.3 Results…………………………………………………………………………………. 69 5.4 Discussion………………………………………………………………………………. 82 5.5 Conclusions…………………………………………………………………………….. 87

6. Iterative Thermomechanical Processing of Ally 600 for Improved Resistance to Corrosion and Stess Corrosion Cracking………………..…………………………………………….. 92 6.1 Introduction…………………..……………………………………………………….. 93 6.2 Materials and Methods…………………………………………………………………. 94 6.3 Results…………………………………………………………………………………. 99 6.4 Discussion……………………………………………………………………………. 113 6.5 Conclusions…………………………………………………………………………….. 118

7. Effect of Thermo-Mechanical Processing on Sensitization and Corrosion of Alloy 600 Stduied by SEM- and TEM-Based Diffraction and Orientation Imaging Techniques…….. 124

7.1 Introduction…………………..……………………………………………………….. 125 7.2 Materials and Methods………………………………………………………………. 126 7.3 Results and Discussion………………………………………………………………. 129 7.4 Conclusions……………………………………………………………………………. 146

8. Effect of Laser Shock Peening on Stress Corrion Cracking of Alloy 600 in Simulated Pressurized Water Rector Environment……………..………………………………….. 151 8.1 Introduction…………………..…………………………………………………….. 151 8.2 Experimental………………………………………………………………………… 152 8.3 Results and Discussion………………………………………………………………. 156 8.4 Conclusions……………………………………………………………………………. 162

9. Effect of Mechanical Loading and Temperature on Relaxation of Residual Stresses Induced by Surface Treatments in Alloy 600……………..……………………………….. 163 9.1 Introduction…………………..……………………………………………………….. 163 9.2 Experimental…………………………………………………………………………… 165 9.3 Results and Discussion………………………………………………………………. 166 9.4 Conclusions…………………………………………………………………………….. 170

10. Gradient Nanostructure and Residual Stress Induced by Ultrasonic Nanocrystal Surface Modification in 304 Austenitic Stainless Steel for High Strength and High Ductility…… 172 10.1 Introduction…………………………………………………………………………… 173 10.2 Experimental Details………………………………………………………………… 174 10.3 Results and Analysis……………………………………………………………….. 178 10.4 Discussion…………………………………………………………………………… 194 10.4 Conclusions…………………………………………………………………………. 202

11. Effects of Ultrasonic Nanocrystal Surface Modification on the Residual Stress, Microstructure and Corrosion Resistance of SS304 Welds………………………………… 208 11.1 Introduction…………………………………………………………………………. 208 11.2 Experimental Methods…………………..…………………………………………… 209 11.3 Results and Discussion………………………………………………………………. 212 11.4 Conclusions………………………………………………………………………….. 218

12. Directions for Future Work………………………………………………..………………. 221

13. Publications ……………………………………………………………………………….. 222

14. Presentations ………………………………………………………………………………. 223

1.0 Executive Summary

Stress corrosion cracking (SCC) of Alloy 600 has been a major problem in commercial

light water reactor (LWR) nuclear power plants. Localized corrosion and intergranular SCC

(IGSCC) have been observed in Alloy 600 in the high temperature (288-340 ˚C) pure water

environment of LWRs. Additionally, IGSCC of Alloy 600 has been reported even at room

temperature under certain conditions in thiosulfate and tetrathionate solutions. In general, SCC

can be attributed to the presence of tensile stress, an aggressive environment and a susceptible

microstructure. Therefore, SCC mitigation techniques address these factors by modifying the

environment, metallurgical processing treatments and alleviating the tensile stresses by

mechanical surface treatments/stress relief.

This study investigated the application of laser shock peening (LSP) as a technique to

mitigate SCC in Alloy 600. LSP induced large compressive residual stresses (-550 MPa) that

decreases gradually through depth. The pressure pulse generated during the LSP treatment causes plastic deformation, resulting in high dislocation density, twins and formation of

misoriented sub-grains/crystallites that have sizes in the range of 50-300 nm in the near-surface

region. Slow strain rate tests (SSRTs) and constant load tests performed in tetrathionate solution

at room temperature were used to evaluate the effect of LSP on the SCC behavior. LSP treated

samples had a significantly longer time to failure and reduced susceptibility to SCC as compared with untreated sensitized Alloy 600. These improvements were attributed to LSP induced compressive residual stresses, increased yield strength (YS) and hardening caused by near-

surface microstructural changes. SSRTs in simulated PWR environment also show similar results

with higher YS, tensile strength and strain to failure. Additionally, the gage section shows fewer

cracks and smaller crack lengths in the LSP treated samples as compared with the untreated

samples.

The other approach involved using mechanical surface treatments/cold work followed by annealing to engineer the Alloy 600 microstructure for increased resistance to corrosion and

IGSCC. We demonstrated a novel method of surface grain boundary engineering (SGBE) in

Alloy 600 using iterative cycles of ultrasonic nanocrystalline surface modification (UNSM) treatment and strain annealing. Three cycles of UNSM and strain annealing at 900-1000 ˚C were used to modify the microstructure to a depth of 250 µm from the surface. This surface treatment

based method increased the fraction of low coincident site lattice (CSL) grain boundaries whilst decreasing the fraction and connectivity of random high angle boundaries (HABs) in the near surface region. Similar results were achieved using thermo-mechanical processing (TMP) with iterative cycles of 10% cold work and strain annealing in Alloy 600. A disrupted random HAB network and large fraction (70%) of CSL boundaries (Σ3-Σ27) reduced the propensity to sensitization. SSRTs in tetrathionate solutions at room temperature show that SGBE and TMP lowered the susceptibility to intergranular SCC. Detailed analysis using EBSD showed cracks arrested at J1 (1-CSL) and J2 (2-CSL) type of triple junctions. The probability of crack arrest, calculated using percolative models, was higher after SGBE and TMP in Alloy 600 and explains the improved IGSCC resistance.

The effects of UNSM on residual stresses, microstructure changes and mechanical properties of austenitic stainless steel 304 were investigated. The dynamic impacts induced by

UNSM leads to surface nanocrystallization, martensite formation, and the generation of high magnitude of surface compressive residual stresses (-1400 MPa) and hardening. Highly dense deformation twins were generated in material subsurface to a depth of 100 µm. These deformation twins significantly improve material work-hardening capacity by acting both as dislocation blockers and dislocation emission sources. Furthermore, the gradually changing

5 martensite volume fraction ensures strong interfacial strength between the ductile interior and the two nanocrystalline surface layers and thus prevents early necking. The microstructure with two strong surface layers and a compliant interior embedded with dense nanoscale deformation twins and dislocations leads to both high strength and high ductility. The work-hardened surface layers

(3.5 times the original hardness) and high magnitude of compressive residual stresses lead to significant improvement in fatigue performance; the fatigue endurance limit was increased by

100 MPa. The results have demonstrated that UNSM is a powerful surface engineering technique that can improve component mechanical properties and performance.

UNSM of welded 304 stainless steel 304 samples was carried out. On the top surface,

UNSM effectively eliminates the tensile stress generated by the welding process and imparts beneficial compressive residual stress. In addition, UNSM can effectively refine the grains and increase hardness in the near surface region. SCC testing in boiling MgCl2 solution shows that

UNSM can significantly improve resistance to SCC due to the compressive residual stresses and changes in the near surface microstructure.

2.0 Introduction

Alloy 600 is a solid solution-strengthened, austenitic alloy containing about 75% Ni, 15%

Cr and 7% Fe as major alloying elements (Figure 1.1). It exhibits an excellent combination of

high strength and good hot/cold workability, good welding and thermal expansion

characteristics. In addition, it has good corrosion resistance against high temperature water

environments resulting in low levels of corrosion products. Alloy 600 is used as steam generator tubing, spray nozzles, heat transfer tubing and other components in light water reactors (LWRs)

(Figure 2.2). In spite of statistically good performance of Alloy 600, failures have been reported in service sometimes after an incubation time of 11-12 years [1]. Some of the failure modes have

been summarized in Figure 2.3 and stress corrosion cracking is clearly the most dominant cause

of failure in mill annealed Alloy 600. As a consequence, Alloy 690 (60Ni-30Cr-10Fe) has now

replaced Alloy 600 as steam generator tubing material in some power plants. Alloy 690 has

superior resistance to SCC in pure water, chlorides and alkaline environments as compared with

Alloy 600. However, it is believed that the superior resistance to SCC might just be due to a

longer incubation time and that Alloy 690 may crack eventually[1].

SCC has been reported in U.S. and international plants have been reported from Control

rod drive mechanism (CRDM) nozzles and bottom mounted instrumentation (BMI) nozzles,

reactor pressure vessel top head penetration nozzles (RPVHPNs), reactor vessel bottom-mounted

nozzles (BMNs), and dissimilar metal welds (DMWs) of primary system piping as well as on

other Alloy 600/82/182 pressure boundary components [2,3]. More importantly, the aging fleet

of reactors means there is greater emphasis on safe operation for extended lifetimes.

Figure 2.1. Ternary equilibrium diagram of Fe-Cr-Ni system at 400˚C

Figure 2.2. Application of Alloy 600 throughout the primary system of a modern PWR[42] The causes of primary water SCC (PWSCC) can be broadly categorized into susceptible material, tensile stresses and environment. Alloy 600 in mill annealed condition is susceptible to

PWSCC and neither microstructure modification (thermal treatment to precipitate grain boundary carbides) nor chemistry control (environment) has successfully mitigated PWSCC.

Additionally, Alloy 600 has also been observed to crack in tetrathionate and thiosulfate solutions

[4–6]. As a consequence of welding or post-weld stress relief treatment, carbides may be precipitated along the grain boundaries. This results in a Cr depleted zone in the area adjacent to the grain boundary. If Cr is severely depleted, Alloy 600 has been observed to be susceptible to intergranular stress corrosion cracking[4,7,8].

Figure 2.3. Failure modes of mill-annealed Alloy 600 steam generator tubes in US PWRs over a 38-year period [1]. Mechanical surface treatment techniques like shot peening (SP), water jet peening (WJP), ultrasonic peening, deep rolling, low plasticity burnishing (LPB) and laser shock peening (LSP) introduce compressive residual stresses. Each surface treatment has different characteristics and unique advantages/disadvantages affecting materials differently [9]. A number of these treatments have been demonstrated to improve the fatigue life (including high temperature fatigue) [10–14], increase wear resistance [15,16] and increase resistance to stress corrosion cracking [17–19]. Some of the treatments are even applied in service to mitigate fatigue, stress corrosion cracking and wear.

Laser peening along with water jet peening (WJP) are being considered as surface residual stress improvement methods to mitigate PWSCC. LSP introduces deep compressive residual stresses as well as changes in the near surface microstructure. Though LSP has been shown to improve resistance to SCC, the mechanism of improvement in SCC resistance is not well understood. Surface treatments like LSP may be essential to increase the life of components in LWRs and preventive measures against SCC related failures in service. In this thesis, we have investigated the changes in the microstructure and residual stresses after LSP. Further, we have shown that LSP increased the resistance to SCC.

Thermo-mechanical processing (TMP) has also been shown to improve the SCC resistance in austenitic stainless steels and nickel alloys[20–24]. TMP routes involve single or multi-step cold work/deformation step followed by strain annealing to increase the fraction of low coincident site lattice (CSL) boundaries. Concurrently, the fraction of random high angle boundaries also decreases. It was theorized that the improvement observed in corrosion resistance after TMP in austenitic SS and nickel alloys was due to increased fraction of low CSL boundaries, mainly Σ3 twin boundaries and its variants. Subsequently, it was reported that analysis of microstructures using triple junction (TJ) characteristics provided a more direct understanding of the influence of microstructural changes on corrosion and SCC properties[25,26]. In recent years, percolation theory approach has been used to rank and design microstructures for improved corrosion and SCC resistance[27].

Numerous studies have evaluated the effect of TMP routes on the relative fractions of twin boundaries and their effect on corrosion properties[27–30]. The underlying mechanism of grain boundary engineering (GBE) using TMP routes has also been discussed in detail [23,31–

34]. It has been reported that the increased twin boundary fraction lowers the susceptibility to

10 sensitization in SS [20,27,35]. Other studies reported that Σ3 boundaries and its variants (Σ9,

Σ27) had higher resistance to SCC [36–40]. Percolation theory approaches have been proposed to design more resistant microstructures to improve the general corrosion and SCC resistance in

SS and nickel alloys[25,41].

In this work, we have used iterative mechanical surface treatments and high temperature annealing cycles to modify the grain boundary character distribution (GBCD) in Alloy 600.

Another approach using cold work and strain annealing has also been studied. Further, we have investigated the effect of this novel surface grain boundary engineering method on the corrosion properties and SCC behavior of Alloy 600 in tetrathionate solutions.

References 1. S.J. Zinkle, G.S. Was, Acta Mater. 61 (2013) 735–758. 2. J. Gorman, S. Hunt, P. Riccadella, Companion Guid. Ot ASME Boil. Press. Vessel Code Chapter 44 (2009). 3. Mater. Reliab. Progr. Tech. Basis Prim. Water Stress Corros. Crack. Mitig. by Surf. Stress Improv. (MRP-267, Revis. 1), EPRI, Palo Alto, CA 1025839. (2012). 4. V.B. Rajan, G.S. Was, Corrosion 43 (1987) 305–308. 5. R. Bandy, R. Roberge, R. Newman, Corrosion (1983). 6. R.C. Newman, R. Roberge, R. Bandy, Corrosion 39 (1983) 386–390. 7. J. Kai, C. Tsai, T. Huang, M. Liu, Metall. Trans. A 20 (1989) 1077–1088. 8. J.J. Kai, C.H. Tsai, G.P. Yu, Nucl. Eng. Des. 144 (1993) 449–457. 9. A. Gill, A. Telang, S.R. Mannava, D. Qian, Y.-S. Pyoun, H. Soyama, V.K. Vasudevan, Mater. Sci. Eng. A (2013). 10. C. Ye, A. Telang, A.S. Gill, S. Suslov, Y. Idell, K. Zweiacker, J.M.K. Wiezorek, Z. Zhou, D. Qian, S. Ramaiah Mannava, V.K. Vasudevan, Mater. Sci. Eng. A 613 (2014) 274–288. 11. X.J. Cao, Y.S. Pyoun, R. Murakami, Appl. Surf. Sci. 256 (2010) 6297–6303. 12. I. Nikitin, I. Altenberger, Mater. Sci. Eng. A 465 (2007) 176–182. 13. R.K. Nalla, I. Altenberger, U. Noster, G.Y. Liu, B. Scholtes, R.O. Ritchie, Mater. Sci. Eng. A 355 (2003) 216–230. 14. C. Ye, S. Suslov, B.J. Kim, E. a. Stach, G.J. Cheng, Acta Mater. 59 (2011) 1014–1025. 15. A. Amanov, I.S. Cho, Y.S. Pyoun, C.S. Lee, I.G. Park, Wear 286-287 (2012) 136–144.

16. H. Lim, P. Kim, H. Jeong, S. Jeong, J. Mater. Process. Technol. 212 (2012) 1347–1354. 17. W. Tsai, C. Chang, J. Lee, Corrosion 50 (1994) 98–105. 18. Y. Sano, M. Obata, T. Kubo, N. Mukai, M. Yoda, K. Masaki, Y. Ochi, Mater. Sci. Eng. A 417 (2006) 334–340. 19. J.Z. Lu, K.Y. Luo, D.K. Yang, X.N. Cheng, J.L. Hu, F.Z. Dai, H. Qi, L. Zhang, J.S. Zhong, Q.W. Wang, Y.K. Zhang, Corros. Sci. 60 (2012) 145–152. 20. M. Shimada, H. Kokawa, Z.. Wang, Y.. Sato, I. Karibe, Acta Mater. 50 (2002) 2331–2341. 21. G. Palumbo, P.J. King, K.T. Aust, U. Erb, P.C. Lichtenberger, Scr. Metall. Mater. 25 (1991) 1775–1780. 22. C. Cheung, U. Erb, G. Palumbo, Mater. Sci. Eng. A 185 (1994) 39–43. 23. T. Watanabe, J. Mater. Sci. 46 (2011) 4095–4115. 24. V. Randle, Mater. Sci. Technol. 26 (2010) 253–261. 25. M. Kumar, W.E. King, A.J. Schwartz, Acta Mater. 48 (2000) 2081–2091. 26. C.A. Schuh, M. Kumar, W.E. King, Acta Mater. 51 (2003) 687–700. 27. S. Tsurekawa, S. Nakamichi, T. Watanabe, Acta Mater. 54 (2006) 3617–3626. 28. V. Randle, M. Coleman, M. Waterton, Metall. Mater. Trans. A 42 (2010) 582–586. 29. V. Randle, R. Jones, Mater. Sci. Eng. A 524 (2009) 134–142. 30. L. Tan, K. Sridharan, T.R. Allen, R.K. Nanstad, D. a. McClintock, J. Nucl. Mater. 374 (2008) 270–280. 31. M. Kumar, A.J. Schwartz, W.E. King, Acta Mater. 50 (2002) 2599–2612. 32. A.J. Schwartz, W.E. King, M. Kumar, Scr. Mater. 54 (2006) 963–968. 33. V. Randle, G. Owen, Acta Mater. 54 (2006) 1777–1783. 34. V. Randle, Acta Mater. 47 (1999) 4187–4196. 35. R. Jones, V. Randle, Mater. Sci. Eng. A 527 (2010) 4275–4280. 36. V.Y. Gertsman, S.M. Bruemmer, Acta Mater. 49 (2001) 1589–1598. 37. B. Alexandreanu, B. Capell, G.S. Was, Mater. Sci. Eng. A 300 (2001) 94–104. 38. B. Alexandreanu, B.H. Sencer, V. Thaveeprungsriporn, G.S. Was, Acta Mater. 51 (2003) 3831–3848. 39. A. King, G. Johnson, D. Engelberg, W. Ludwig, J. Marrow, Science (80-. ). 321 (2008) 382–386. 40. E.A. West, G.S. Was, J. Nucl. Mater. 392 (2009) 264–271. 41. T.J. Marrow, L. Babout, A.P. Jivkov, P. Wood, D. Engelberg, N. Stevens, P.J. Withers, R.C. Newman, J. Nucl. Mater. 352 (2006) 62–74. 42. X. Ru, R.W. Staehle, Corrosion 69 (2013) 423–447.

3.0 Effects of Laser Shock Peening on SCC Behavior of Alloy 600

Abhishek Telang1, Chang Ye1, Amrinder Gill1, Sebastien Teysseyre2, S.R.Mannava1, Dong Qian3, Vijay K. Vasudevan1 1Department of Mechanical and Materials Engineering, University of Cincinnati, 501B ERC, Cincinnati OH 45221-0072 2Idaho National Laboratory, Materials Properties and Performance, Idaho Falls, ID 83415-2218 3Department of Mechanical Engineering, University of Texas, Richardson, TX 75080-3021

Published in the Proceedings of the 16th International Conference on Environmental Degradation of Materials in Nuclear Power Systems- Water Reactors, Asheville, NC, 2013.

ABSTRACT

In this study, the effects of laser shock peening (LSP) on stress corrosion cracking (SCC) behavior of Alloy 600 in tetrathionate solution were investigated. The degree of sensitization was quantified using double loop electrochemical potentiokinetic reactivation (DLEPR) tests. The sensitized Alloy 600 was demonstrated to be susceptible to intergranular SCC in tetrathionate solution. Following LSP, residual stresses and the amount of plastic strain introduced in Alloy 600 were characterized. The effects of LSP on SCC susceptibility of Alloy 600 in tetrathionate solution were evaluated by slow strain rate tests and constant load tests. Results indicate a significantly increased resistance to crack initiation and decreased susceptibility to SCC after LSP. Keywords: Stress corrosion cracking, Alloy 600, laser shock peening, sensitization, residual stress. INTRODUCTION

Intergranular stress corrosion cracking (IGSCC) continues to be an issue of materials degradation and maintenance concern for many nuclear components in light water nuclear reactors (LWRs). Failures due to IGSCC have been observed and studied extensively. Steam generator tubing materials like austenitic stainless steels, Ni-based Alloy 600 and their weldments have been observed to be susceptible in sulfur bearing solutions at room temperature

[1–7].

SCC of Alloy 600 in sulfur bearing solutions at room temperature can be attributed to a complex interplay between a susceptible microstructure that is usually a result of processing,

welding/stress relief treatment, an aggressive environment (thiosulfate and tetrathionate) and

tensile stresses (applied or residual stress). Bandy et al.[8] reported susceptibility of sensitized

Alloy 600 to SCC in thiosulfate solution at room temperature. They also investigated the effects

of sensitization, pH and solution concentration on the SCC behavior of Alloy 600 in thiosulfate

and tetrathionate solutions at room temperature. Was and Rajan [2] performed detailed studies to

2- demonstrate that S4O6 ions were primarily responsible for intergranular cracking and intergranular cracking occurred by stress assisted intergranular attack. Kai and co-workers [9,10] investigated the effect of Cr concentration at the grain boundaries on IGSCC behavior of Alloy

600 and 690 in thiosulfate solutions using constant load tests. They concluded that the Cr concentration at the grain boundaries of less than 8 wt% would result in increased susceptibility to IGSCC in Alloy 600 while Alloy 690 was found to be immune to IGSCC in thiosulfate (due to its higher Cr content) [9]. It is well established that Alloy 600 was susceptible to IGSCC under certain conditions viz. sensitized metallurgical condition, aggressive environment like thiosulfate or tetrathionate, pH and tensile residual stresses.

Mitigation of SCC in stainless steels (SS304, SS316) and nickel based Alloy 600 has been studied primarily from the perspective of alleviating the susceptible microstructure by a suitable heat treatment or by changing the environment [8,11] but not necessarily by modifying the nature of the stresses. Tsai et al. [12] studied the effects of shot peening on SCC behavior of

Alloy 600 and observed decrease in percentage of weight loss after Huey’s test in shot peened specimens. More importantly, they reported improvement in resistance of Alloy 600 after shot peening to SCC using U-bend tests in 0.01 M thiosulfate solution at 95°C. Sano et al. [13] reported improved resistance to SCC in SS304 after LSP treatment using creviced bent beam tests in high temperature water (288°C) [13]. Recently, Hurrell et al. [14] reviewed residual

14 stress mitigation methods (including LSP, water jet peening and ultrasonic peening) that were proposed and/or employed for applications in nuclear power plants. They particularly emphasized the applicability of these techniques in key/critical weld locations in service and the need to understand their long term benefits [14].

LSP is an advanced mechanical surface treatment technique that has been widely used in the aerospace industry to improve component fatigue performance. LSP typically uses a Q- switched Nd:Glass or Nd:YAG (λ= 1064nm) laser of high energy (1-8 J) and short pulse duration (6-20 ns) that passes through a transparent confining medium (water or glass) to ablate a sacrificial thin coating (tape or opaque medium) on the material surface (Figure 3.1). As the laser beam passes through the water or glass, it is absorbed by the opaque overlay and only a thin layer (few microns) of this sacrificial layer is ablated. Thermal effects on the material surface are suppressed [15]. As the generated plasma continues to absorb the rest of the laser energy, it is readily heated and expands between the material and the confining medium, thus generating a shock wave that propagates through the material. The volume affected by the shock wave is plastically deformed during its propagation to a depth beyond which the peak pressure does not exceed the Hugoniot Elastic limit of the material. The surrounding material in the sub-surface region is opposed to lateral straining resulting in biaxial compressive stresses near the surface.

Physics of LSP technique have been reported elsewhere in literature [15,16].

A few studies have reported effects of residual stress mitigation techniques but the mechanism of improvement in SCC resistance is not well understood [12,13,17]. To the best of our knowledge there are no studies reported in literature focusing on the effects of LSP on SCC behavior of Alloy 600 in sulfur bearing solutions at room temperature. The present work was undertaken to systematically evaluate the effects of laser shock peening on SCC resistance of

Alloy 600. Double loop Electrochemical Potentiokinetic Reactivation (DLEPR) tests were used to quantify the degree of sensitization and susceptibility to IGSCC. Residual stresses through depth were measured after LSP treatment on Alloy 600. Effect of LSP on SCC behaviour was evaluated using slow strain rate tests (SSRT) and constant load tests.

Figure 3.1. Schematic of laser shock peening process.

EXPERIMENTAL

Material and Heat Treatment

Alloy 600 (chemical composition as given in Table 1) specimens were sectioned from a billet using a wire electric discharge machine (EDM), solution annealed (SA) at 1100 ºC for 30 minutes followed by different sensitization (SEN) treatments, as follows:

• Solution Annealed (SA), water quenched (WQ)

• SA,WQ + Sensitization (SEN) at 621 ºC, 18 hours, WQ

• SA, WQ + Sensitization (SEN) at 700 ºC, 1 hour, WQ

• SA,WQ + Sensitization (SEN) at 700 ºC, 3 hours, WQ

Table 1. Chemical composition (wt.%) of Alloy 600 [22]

C Mn Fe S P Si Cu Ni Cr Ti Nb Co 0.07 0.22 7.39 0.002 0.006 0.012 0.05 76.0 15.55 0.24 0.07 0.058

Specimen were polished progressively to 1200 grit followed by fine polishing with 0.05 µm

colloidal silica to achieve a mirror finish and etched with 10% oxalic acid solution (3V, 15s).

Microstructures from the following sensitization treatments were characterized using a digital

optical microscope. Transmission Electron Microscopy (TEM) thin foils from sensitized

specimen were prepared by a twin-jet polisher using a 90:10 vol% CH3OH:HClO4 solution. To

obtain foils from close to the LSP treated surface, a thin section was sliced off parallel to peened

surface and thinned (dimpled and ion milled) from one side to a thickness of 20 µm, leaving the

peened surface intact. The foil was then ion milled from only one side, at low angle (12°) to

ensure that there is no damage induced by ion beam. This enables one to obtain thin areas which

are very close to surface (within a micron). The near surface specimens are important as they

help in understanding the fundamental deformation changes introduced by these surface

treatments. TEM thin foils were analyzed with a Phillips CM-20 TEM.

Double Loop Electrochemical Potentiokinetic Reactivation (DLEPR) Test

DLEPR tests were performed in accordance with ASTM G108-94 in a solution composed of 0.01 M H2SO4 + 20 ppm KSCN using a Gamry Potentiostat (Reference 600) [18]. The scan rate was set at 0.5 mV/s for activation and reactivation loop and the sample size was 1 cm2. Each

heat treated coupon was mechanically polished, wet polished with 3 µm, 1 µm diamond

suspension and finished with 0.05 µm colloidal silica suspension. Freshly prepared solution was

de-aerated with high purity N2 gas before and during each test. All tests were performed at room

temperature. Coupons were kept immersed in the test solution for 1 hour at open circuit potential

before the start of each test.

Laser Shock Peening

Laser shock peening was performed using a Q-switched Nd:YAG laser (Continuum

Surelite) with infrared wavelength (1064 nm) and a frequency of 8 Hz. Glass (BK7) was used as

confining medium and Al tape (80 µm) was used as the sacrificial ablative coating. The pulse

power density was set at 5 GW/cm2 with spot size of 1 mm and pulse width of 6 ns. A schematic of the laser shock peening process is shown in Figure 3.1.

Residual Stresses

Residual stresses were measured using sin2ψ technique with a Proto LXRD instrument,

MnKα radiation and (311) peak of the austenite phase To measure residual stress through depth,

coupons were electropolished using 87.5:12.5 vol.% CH3OH:H2SO4 solution to remove layers of

predetermined thickness.

SCC Tests

Flat coupons with dimensions as shown in Figure were machined using wire EDM from

sensitized sheets of Alloy 600, polished to 1200 grit, degreased with acetone, dried and

immersed in the test solution for 1 hour prior to straining. LSP treatment was applied on the gage

section on both opposing surfaces of the coupons. LSP was performed using Al tape as the

ablative layer and glass as the confining medium with the parameters described earlier. Slow

strain rate tests (SSRT) were performed with CORTEST SSRT equipment driven by a servo

motor and fitted with a custom built environmental chamber. Load and displacement values were

recorded periodically and coupons were tested to failure. Constant load tests were performed

18 loading the coupons (dimensions as shown in Figure 3.2) to the predetermined stress level and recording the load, time to failure.

Figure 3.2. Schematic of coupon used for constant extension rate tests and constant load tests. Some coupons were LSP treated on both surfaces of the gage section. All dimensions in mm.

RESULTS AND DISCUSSION

Microstructure

Figure 3.3(a-d) shows optical microstructures of the solution annealed and sensitized coupons. The grain size of the initial as received material was 50-60 microns and increased subsequently after solution annealing and sensitization treatments. Some evidence of etchant attack along the grain boundaries, presumably due to carbides, can be seen in the sensitized samples, particularly in Fig. 3.33(d). Cross section of LSP treated Alloy 600 (Figure 3.33e) shows no grain structure modification.

TEM observations of sensitized Alloy 600 (SA + SEN 621C, 18h) samples revealed fine intergranular carbides (~200 nm) precipitated along the grain boundaries as shown in Figure 3.4.

The inset selected area diffraction (SAD) pattern in Figure 3.4 establishes the presence of M23C6 carbides with a orientation relationship given by {100}γ//{100}M23C6. Precipitation of these Cr- rich (~90% in M23C6 type) carbides along the grain boundaries usually results in neighboring

19 regions becoming depleted of Cr, the extent of which depends on the sensitization treatment

[6,8]. Figure 3.5 shows the near surface microstructure of LSP treated Alloy 600 (as received).

The near surface microstructure shows twins and dislocation cells due to the intense shock wave propagating through the material.

Figure 3.3. Optical micrograph showing the microstructure of Alloy 600 after (a) Solution Annealing at 1100 ºC,0.5 h, WQ (b) SA,WQ + SEN at 700 ºC, 1hr, WQ (c) SA,WQ + SEN at 700 ºC,3 h,WQ (d) SA,WQ + SEN at 621 ºC,18 h,WQ (e) Cross section of LSP treated Alloy 600 (as received).

Figure 3.4. TEM micrograph showing carbides precipitated along grain boundaries in SA+SEN 621C18h Alloy 600.

Figure 3.5. a) and b) TEM micrograph from the top surface of LSP treated (5GW/cm2) Alloy 600 showing twins and dislocation cells. DLEPR Test

Electrochemical potentiokinetic reactivation (EPR) tests have been used previously to quantify degree of susceptibility to IGA/IGSCC in Ni-based Alloy 600 [18,19]. DLEPR test is a fast, reproducible and quantitative method to determine degree of sensitization (DOS). The test parameters reported by Lim et al. [18] were used as a reference for DLEPR tests in this study.

The DOS after heat treatments was quantified as,

DOS, % =Ir/Ia×100 (1) where Ir is the maximum current density in the reactivation loop and Ia is the maximum current density in the anodic loop. Figure 3.6 shows the comparison between DLEPR curves for solution annealed and sensitized specimens, while the summary of the results is displayed in Figure 3.7.

A higher degree of sensitization indicates a wider Cr depletion zone and thus greater the susceptibility to intergranular stress corrosion cracking. From Figure 6 it is evident that the

621°C, 8 h heat treated coupon has the highest degree of sensitization and is expected to have a grain-boundary Cr-depleted region and the highest susceptibility to IGSCC.

Figure 3.6. DLEPR test results for various sensitization conditions.

24.00 SA 20.19 20.00 SA+SEN700C1hr 17.54 SA+SEN700C3hr 16.00 SA+SEN621C18hr 11.29 12.00

DOS,% 8.00

4.00 0.11 0.00

Figure 3.7. Summary of results of Degree of Sensitization (DOS) for different sensitization conditions

Residual Stress

In depth residual stresses were measured using the d vs sin2ψ method. Gradient and depth corrections were applied and residual stresses after LSP are as shown in Figure 3.8. The residual

stresses were of the order of about -400 MPa at the surface and the depth of the compressive

layer was about 0.6 mm (in a 2 mm thick specimen). The surface residual stresses have been

significantly modified from tension (~100 MPa) before LSP to compression (-650 MPa) after

LSP. Magnitude and depth of compressive residual stresses can be controlled by varying the LSP parameters like pulse energy, spot size, overlap ratio. Full widths at half maximum (FWHM) values are a qualitative indication of the level of plastic strain introduced in a material. In Figure

3.8, FWHM values have been normalized (with respect to FWHM of unaffected material) for easier comparison. In LSP treated Alloy 600, the level of plastic strain is higher at the surface but drops rapidly and is relatively stable thereafter.

Figure 3.8. Surface and through depth residual stresses and FWHM in LSP treated Alloy 600.

Effect of Laser Shock Peening on SCC

Slow Strain Rate Tests were used to evaluate the susceptibility of sensitized Alloy 600 to

IGSCC in tetrathionate solution at room temperature and particularly to understand the effects of

LSP. Figure 3.9 shows the effect of sensitization treatments on the SCC susceptibility in tetrathionate solution. These tests were used to compare the effects of degree of sensitization, strain rate, and LSP treatment on the SCC behavior of Alloy 600.

Figure 3.9. SSRT results showing effects of sensitization treatment and 0.001 M tetrathionate solution on SCC behavior of Alloy 600.

In the test conditions investigated, Solution Annealed (SA) Alloy 600 was immune to

IGSCC whereas the sensitization treatment made Alloy 600 susceptible to corrosion in 0.001 M

tetrathionate solution. As its strain to failure was much lower, the sensitization treatment of 621

°C, 18 h after SA was more severe than 700 °C, 3 h. This corresponds well with DOS values

from DLEPR tests which indicate a higher degree of sensitization in coupons with the former

sensitization treatment. To understand the effect of the environment, a sensitized coupon was

tested in air (neutral environment). It was observed from the SSRT results that tetrathionate

solution was aggressive and resulted in premature failure. The strain to failure of sensitized

Alloy 600 was ~64 % in air as compared to 7 % in 0.001 M tetrathionate solution at the same

strain rate (10-6 s-1). Fracture surfaces were examined by SEM to analyze the SSRT failure

modes (Figure 3.10). Ductile cup and cone type of fracture surface was observed in the solution

annealed coupon (3.10a, 3.10b) while the sensitized coupon showed characteristic features of

IGSCC (Figure 3.10c, 3.10d).

The effect of LSP treatment on the SCC behavior of Alloy 600 in tetrathionate solution

was compared using SSR tests in 0.001 M solution at room temperature at the same strain rates

(10-6 s-1 and 5*10-6 s-1) as shown in Figure 3.11. At very high strain rates (10-5 s-1), the mode of failure is not fully SCC since the reactions involved in the growth of environment sensitive cracks proceed at a rate that is slower than that at which ductile cracking may propagate [20].

The LSP treated coupons have higher yield strength and ultimate tensile strength as compared to the untreated coupons (Figure 3.11). This could be attributed to the effect of deep compressive residual stresses as well as near-surface microstructural changes introduced in the material as a result of LSP. The relative contributions of these two LSP-induced changes need to be ascertained. At strain rate of 10-5 s-1, the LSP treated coupon shows slightly lower strain to

failure as compared with the untreated coupon, but the stress at which the transition to IGSCC

failure is much higher compared with the untreated coupon because of the hardening caused by

LSP. Failure analysis of coupons tested at higher strain rates (5*10-6 s-1 and 10-5 s-1) showed mixed mode of failure that included both intergranular and ductile cracking as shown in Figure

10(e,f).

a) b)

c) d)

e) f)

Intergranular cracking

ductile cracking

Figure 3.10. a) Fracture surface of Solution Annealed Alloy 600. b) High magnification image of Solution Annealed Alloy 600 showing ductile fracture surface. c) Fracture surface of sensitized (621 °C, 18 h) Alloy 600 subjected to SSRT. d) High magnification image of boxed in area in c) showing characteristic IGSCC type failure. e) Low magnification of fracture surface of sensitized (SA + SEN 621 °C, 18 h) Alloy 600 after SSRT in 0.001 M tetrathionate (strain rate = 5*10-6 /s). f) High magnification micrograph of the area in e) indicating mixed mode of failure.

Figure 3.11. Effect of strain rate on SCC behavior of sensitized Alloy 600 in 0.001 M tetrathionate solution at room temperature.

In this study, both sensitization treatments led to SCC susceptibility in tetrathionate

solution. Abe et al. [21] proposed the following equation to calculate the SCC index (ISCC),

⎡ ⎤ ⎡ ⎤ 1+en Pn ISCC = ⎢ SCC ⎥ x⎢ SCC −1⎥ (2) ⎣1 +en ⎦ ⎣ Pn ⎦

scc where en and en are strains at the maximum load for load-elongation plot in air and

scc environment€ respectively. Pn and Pn are the maximum loads for load-elongation plot in air and environment, respectively. Substituting load with stress, we calculated ISCC for each case and the

results are as shown in Figure 3.12. The SCC susceptibility of LSP treated sensitized Alloy 600

is lower than that of counterpart untreated sensitized Alloy 600. It is interesting to note that the

ISCC is significantly lower even in a severe condition (low extension rate). With higher strain rate,

SCC susceptibility drops significantly from (2.52 to 0.09) for untreated sensitized Alloy 600.

Figure 3.12. Comparison of SCC susceptibility of untreated (Baseline) and LSP treated sensitized Alloy 600 in 0.001 M tetrathionate solution (high strain rate = 10-5 s-1, medium strain rate = 5*10-6 s- 1, low strain rate = 10-6 s-1).

Constant Load Tests were used to evaluate the effects of LSP on Alloy 600 under static loading conditions in 0.001 M tetrathionate and Figure 3.13 summarizes the effects of LSP in terms of time to failure. Sensitized but untreated coupons failed via SCC within 30 to 40 hours.

No cracks were observed on the LSP treated coupons after 120 hours in the same solution at the same stress following which the tests were terminated. The increased crack initiation time after

LSP may be attributed to the beneficial compressive residual stresses (Figure 3.8) induced by

LSP which reduced the effective stresses under constant loading conditions. The time to failure for sensitized Alloy 600 corresponds well with that reported by Bandy et al. [8]. Tsai et al. [12] had observed increased resistance to crack initiation in Alloy 600 in 0.01M thiosulfate solution at

95 °C after shot peening. Crack initiation times were higher for shot peened samples indicating increased resistance to SCC. They attributed the improved SCC resistance of sensitized Alloy

600 to changes in the near surface microstructure and shot peening induced compressive residual

stress. Sano et al. [13] also observed increased resistance to crack initiation after LSP treatment

on SS304 in high temperature pure water. They observed that LSP (without coating) prevented

SCC even if the residual stresses introduced after LSP were as low as -100 MPa. In addition,

LSP had prevented crack initiation and mitigated crack propagation. Peyre et al. [17] also

reported that SCC was inhibited in boiling MgCl2 solution under constant loading conditions after LSP treatment in SS316L. Results from our study and others indicate a beneficial effect of

LSP induced compressive residual stresses in mitigating SCC in Alloy 600.

Figure 3.13. Comparison of time to failure for sensitized (SA + SEN621 °C, 18 h) Alloy 600 in 0.001 M tetrathionate solution (room temperature) at constant load.

CONCLUSIONS

1. DLEPR tests were used to quantify the DOS for sensitized Alloy 600. Sensitized Alloy

600 shows precipitation of intergranular carbides along grain boundaries.

2. Sensitized Alloy 600 was susceptible to IGSCC in sodium tetrathionate solution at room

temperature. A higher DOS corresponded to greater susceptibility of sensitized Alloy 600

to SCC in tetrathionate solution.

3. Deep compressive residual stresses were introduced and a substantial increase in strength

was observed after LSP in Alloy 600. Near surface microstructure of LSP treated Alloy

600 shows deformation bands and dislocation cell networks.

4. Slow strain rate tests were used to evaluate IGSCC susceptibility of sensitized Alloy 600

in tetrathionate solution. LSP treatment led to a significant improvement in resistance to

SCC of sensitized alloy 600, which was manifested in the form of a lower SCC

susceptibility index and vastly increased crack initiation time in constant load tests.

ACKNOWLEDGEMENTS

The authors are grateful for financial support of this research by the Nuclear Energy

University Program (NEUP) of the US Department of Energy contract #102835 issued under prime contract DE-AC07-05ID14517 to Battelle Energy Alliance, LLC. We would like to acknowledge Dr. Bogdan Alexandreanu (Argonne National Laboratory) for providing the Alloy

600 material. We also gratefully acknowledge the contribution of the State of Ohio, Department of Development and Third Frontier Commission, which provided funding in support of “Ohio

Center for Laser Shock Processing for Advanced Materials and Devices” and the experimental and computational equipment in the Center that was used in this work. Any opinions, findings, conclusions, or recommendations expressed in these documents are those of the author(s) and do not necessarily reflect the views of the DOE, State of Ohio, Department of Development.

REFERENCES 1. R. Newman, K. Sieradzki, H. Isaacs, Stress-corrosion cracking of sensitized type 304 stainless steel in thiosulfate solutions, Metallurgical Transactions A. 13 (1982) 2015–2026.

2. G.S. Was, V.B. Rajan, The Mechanism of Intergranular Cracking of Ni-Cr-Fe Alloys in Sodium Tetrathionate, Metallurgical Transactions A. 18 (1987) 1313–1323. 3. G.S. Was, H.H. Tischner, R.M. Latanision, The Influence of Thermal Treatment on the Chemistry and Structure of Grain Boundaries in, Metallurgical Transactions A. 12 (1981) 1397–1408. 4. G.S. Was, V.B. Rajan, Technical Note: On the Relationship between the EPR Test, Sensitization, and IGSCC Susceptibility, Corrosion. 43 (1987) 576–579. 5. D. Van Rooyen, Review of the stress corrosion cracking of Inconel 600, Corrosion. 31 (1975) 327–337. 6. S.M. Bruemmer, G.S. Was, Microstructural and microchemical mechanisms controlling intergranular stress corrosion cracking in light-water-reactor systems, Journal of Nuclear Materials. 216 (1994) 348–363. 7. V.S. Raja, T. Shoji, eds., Stress Corrosion Cracking: Theory and Practice, Woodhead Publishing Limited, 2011. 8. R. Bandy, R. Roberge, R. Newman, Low temperature stress corrosion cracking of Inconel 600 under two different conditions of sensitization, Corrosion Science. 23 (1983) 995–1006. 9. J.J. Kai, C.H. Tsai, G.P. Yu, The IGSCC, sensitization, and microstructure study of Alloys 600 and 690, Nuclear Engineering and Design. 144 (1993) 449–457. 10. J. Kai, C. Tsai, T. Huang, M. Liu, The effects of heat treatment on the sensitization and SCC behavior of INCONEL 600 alloy, Metallurgical Transactions A. 20 (1989) 1077–1088. 11. R. Bandy, D. Van Rooyen, Effect of Thermal Stabilization on the Low Temperature Stress Corrosion Cracking of Inconel 600, Corrosion. 40 (1984) 281–289. 12. W. Tsai, C. Chang, J. Lee, Effects of shot peening on corrosion and stress corrosion cracking behaviors of sensitized alloy 600 in thiosulfate solution, Corrosion. 50 (1994). 13. Y. Sano, M. Obata, T. Kubo, N. Mukai, M. Yoda, K. Masaki, et al., Retardation of crack initiation and growth in austenitic stainless steels by laser peening without protective coating, Materials Science and Engineering: A. 417 (2006) 334–340. 14. P. Hurrell, D. Everett, A. Gregg, S. Bate, Review of residual stress mitigation methods for application in nuclear power plant, Proceedings of the ASME Pressure Vessels and Piping Conference. 6 (2006). 15. P. Peyre, R. Fabbro, Laser shock processing : a review of the physics and applications, Optical and Quantum Electronics. 27 (1995) 1213–1229. 16. Y. Sano, Y. Sakino, N. Mukai, M. Obata, I. Chida, T. Uehara, et al., Laser Peening without Coating to Mitigate Stress Corrosion Cracking and Fatigue Failure of Welded Components, Materials Science Forum. 580-582 (2008) 519–522. 17. P. Peyre, L. Berthe, R. Fabbro, Corrosion reactivity of laser-peened steel surfaces, Journal of Materials Engineering and Performance. 9 (2000) 656–662. 18. Y. Lim, H. Kim, J. Han, J. Kim, H. Kwon, Influence of laser surface melting on the susceptibility to intergranular corrosion of sensitized Alloy 600, Corrosion Science. 43 (2001) 1321–1335.

19. V. Kain, Y. Watanabe, Development of a single loop EPR test method and its relation to grain boundary microchemistry for alloy 600, Journal of Nuclear Materials. 302 (2002) 49– 59. 20. R.N. Parkins, Slow Strain Rate Testing - 25 Years Experience, in: Slow Strain Rate Testing for the Evaluation of Environmentally Induced Cracking: Research and Engineering Applications, ASTM STP 1210, R.D. Kane,Ed.,American Soceity for Testing and Materials, Philadelphia, 1993: pp. 7–21. 21. S. Abe, M. Kojima, Y. Hosoi, Stress Corrosion Cracking Susceptibility Index, ISCC, of Austenitic Stainless Steels in Constant Strain-Rate Test (STP 665), STP 665 Stress Corrosion Cracking—The Slow Strain-Rate Technique (ASTM International), 1979: pp. 294–304. 22. B. Alexandreanu, O.K. Chopra, W.J. Shack, Crack Growth Rates of Nickel Alloy Welds in a PWR Environment, NUREG/CR–6907, ANL–04/3, 2006.

4.0 Effects of Laser Shock Peening on SCC Behavior of Alloy 600 in Tetrathionate Solution

Abhishek Telanga*, Amrinder S. Gillb, Sebastien Teysseyrec, Seetha R. Mannavaa, Dong Qiand and Vijay K. Vasudevana aDepartment of Mechanical and Materials Engineering, University of Cincinnati, 2901 Woodside Dr., Cincinnati, OH 45221-0072 bAK Steel, Research Center, 705 Curtis St, Middletown, OH 45044 cIdaho National Laboratory, Idaho Falls, ID 83415-2218 dDepartment of Mechanical Engineering, University of Texas at Dallas, 800 West Campbell Rd, EC-38, Richardson, TX 75080-3021

(Published in Corrosion Science, Volume 90, Pages 434–444 (2015); doi:10.1016/j.corsci.2014.10.045)

Abstract

In this study, the effects of laser shock peening (LSP) on stress corrosion cracking (SCC) behavior of nickel based Alloy 600 in tetrathionate solution were investigated. The LSP induced compressive residual stresses and changes in the near surface microstructure, hardness were characterized. The effects of LSP on SCC susceptibility of Alloy 600 in tetrathionate solution were evaluated by slow strain rate tests and constant load tests. The results indicate a significantly longer time to failure and decreased susceptibility to SCC. These improvements were attributed to LSP induced compressive residual stresses, increased yield strength and hardening caused by near-surface microstructural changes.

1. Introduction

Alloy 600 has been known to be susceptible to stress corrosion cracking (SCC) at room temperature in polythionic acid environments [1,2] which can be attributed to a combination of susceptible microstructure (sensitization), aggressive species (thiosulfate and tetrathionate) and

34 tensile stresses (applied or residual stresses). Detailed studies on the effect of sensitization treatment, pH, solution concentration, temperature have been used to elucidate the mechanism of intergranular stress corrosion cracking (IGSCC) in polythionic acid environments[3–6]. In general, IGSCC can be mitigated by stabilization heat treatment to modify the microstructure, by changing the environment (pH, reducing the concentration of aggressive species) and by modifying the nature of stresses.

Laser Shock Peening (LSP) is an advanced mechanical surface treatment technique that has been widely used in the aerospace industry to improve component fatigue performance. LSP typically uses a Q-switched Nd:Glass or Nd:YAG (λ= 1064nm) laser of high energy (1-8 J) and short pulse duration (6-20 ns) that passes through a transparent confining medium (water or glass) to ablate a sacrificial thin coating (tape or opaque medium) on the material surface (Figure

4.1). As the laser beam passes through the water or glass, it is absorbed by the opaque overlay and only a thin layer (few µm) of this sacrificial layer is ablated creating plasma. Thermal effects on the material surface are suppressed [7]. As the generated plasma continues to absorb the rest of the laser energy, it is readily heated and expands between the material and the confining medium, thus generating a shock wave that propagates through the material. The volume affected by the shock wave is plastically deformed during its propagation to a depth beyond which the peak pressure does not exceed the Hugoniot Elastic limit of the material. The surrounding material in the sub-surface region is opposed to the lateral straining resulting in biaxial compressive stresses near and under the surface of the material. Physics of LSP technique have been reported elsewhere [7,8].

Recent interest in life extension of nuclear power plants has contributed to a drive towards mitigation strategies that counter materials degradation challenges in these harsh

35 environments. Fatigue damage from mechanical and/or environmental factors, environmentally assisted cracking (including primary water stress corrosion cracking, intergranular stress corrosion cracking and low temperature crack propagation), and weld residual stresses leading to

SCC have been cited as causes of failure in metallic components in service [9]. Surface treatments including laser shock peening, shot peening, water jet peening, low plasticity burnishing have proven to significantly increasing the fatigue resistance of aero alloys, SCC resistance [10] introducing deep compressive residual stresses in the material [11–13]. In theory, the presence of compressive residual stresses should mitigate or perhaps prevent stress corrosion cracking. Shot peening has been generally used to introduce compressive residual stresses to improve fatigue performance and also mitigate stress corrosion cracking [14,15].

Figure 4.1. Schematic of the Laser Shock Peening process In this paper, we present the effects of LSP on the SCC behavior of Alloy 600 in polythionic acids. We first quantified the degree of sensitization using double loop EPR tests and

36 performed detailed characterization of residual stresses, hardness and microstructure after LSP on Alloy 600. Furthermore, slow strain rate tests and constant load tests were used to evaluate the effects of LSP on the IGSCC resistance. The results obtained are described and discussed in detail.

2. Experimental

2.1 Material and Laser Shock Peening

Alloy 600 plate with 2 mm thickness (Goodfellow Corporation, USA) with chemical composition as listed in Table 4.1 was used in this study. Heat treatments, LSP parameters and their respective designations are as listed in Table 4.2.

Table 4.1. Chemical composition of the Inconel Alloy 600 used in this study.

C Mn Si S Cr Fe Co Cd Ti Cu P Al Ni 0.001 0.001 0.08 0.16 0.18 14.99 8.05 0.18 0.01 0.18 0.1 0.08 Bal. max. max.

Table 4.2. Designation for sensitization and LSP treatments considered in this study.

Designation Sensitization and LSP treatments AR As Received (annealed) S1 AR + 650 °C, 5h, air cooled S2 AR + 650 °C, 2h, air cooled LSP1 S1 + LSP 5.6 GW/cm2 LSP2 S2 + LSP 5.6 GW/cm2

To characterize the cross section microstructure, mounted samples were ground progressively to 1200 grit, followed by fine polishing with 0.05 µm colloidal silica suspension to achieve a mirror finish and etched with 10% oxalic acid solution (3V, 15s). Microstructure after sensitization and LSP treatments were characterized using a Keyence VH 600 digital optical microscope.

A schematic of the LSP process is shown in Figure 4.1. LSP was performed using a Q- switched Nd:YAG laser (Continuum Powerlite Plus) with infrared wavelength (1064 nm) and a frequency of 1 Hz. The target (Alloy 600 sample) was moved using a XY table to create a pattern with 50% overlay. Flowing water was used as confining medium and vinyl tape was used as the sacrificial ablative coating. Based on a series of experiments involving variations in the

LSP parameters and counterpart measurements of the surface residual stress, the conditions of pulse energy of 3J, spot size of 2 mm diameter and pulse width of 20 ns was used for treatment of all the coupons prior to SCC testing and other characterization. The laser power density for each pulse based on these parameters was ~ 5.6 GW/cm2.

2.2 Transmission Electron Microscopy

Characterization of the near-surface regions of the LSP-treated samples is important for understanding the fundamental microstructural changes introduced by these types of surface treatments. Enormous care was given to the preparation of thin foils for Transmission Electron

Microscopy (TEM) for observations of the near-surface microstructure of the LSP-treated samples, for which conventional twin-jet electro-polishing is not suitable. Consequently, thin foils from the specimens (sensitized, LSP-treated, etc.) for TEM observations of these microstructural changes were prepared by normal dimpling and ion milling procedures. To obtain foils from close to the LSP treated surface, a thin section was sliced off parallel to peened surface, polished to a thickness of ~100 µm, dimpled using a Fischione Instruments Model 150 dimpler from the side opposite to the peened surface to obtain a one-sided web of thickness of

~20 µm near the center, leaving the peened surface untouched and intact. The foil was then ion milled using a Fischione Instruments Model 1010 Ion Milling System only from one side using a shallow incidence angle (12°) and low energy (4 keV, 45˚ stage rotation) to ensure that there is

no damage induced by ion beam. This enables one to obtain thin areas which are very close to

surface. The near surface samples are important as they help in understanding the fundamental

microstructural changes introduced by these surface treatments. The same technique was

utilized to prepare thin foils of all samples, including those in the non-LSP treated sensitized

conditions.

TEM thin foils were observed with a Phillips/FEI CM20 TEM operated at 200 kV and

photographs from relevant regions were recorded under bright field (BF), dark field (DF), weak-

beam dark field (WBDF) and selected area diffraction (SAD) modes. Energy dispersive X-ray

(EDS) spectroscopy in TEM was conducted at several points at and across grain boundaries to

obtain chromium concentration. X ray intensities were collected at each position for 40 s by

stepping the electron beam (spot size of 15 nm diameter) in increments of 50-100 nm across

grain boundaries in the sensitized (S1 and S2) samples. Standardless analysis was used to

quantify the concentration of Cr and the other elements from the x-ray intensity data.

2.3 Double Loop Electrochemical Potentiokinetic Reactivation (DLEPR) tests

Electrochemical potentiokinetic reactivation (EPR) tests have been used previously to

quantify degree of susceptibility to IGA/IGSCC in Ni-based Alloy 600 [16,17]. The double loop electrochemical potentiokinetic (DLEPR) test is a fast, reproducible and quantitative method to determine degree of sensitization (DOS). The DOS after heat treatments was quantified as the ratio of maximum current in the reactivation loop to maximum current in the activation loop.

DLEPR tests were performed in accordance with ASTM G108-94 in a solution composed of

0.01 M H2SO4 + 20 ppm KSCN using a Gamry Potentiostat (Reference 600) [16]. The scan rate was set at 0.5 mV/s for activation and reactivation loop and the sample size was 1 cm2. Each heat

treated coupon was mechanically ground to 1200 grit, wet polished with 1 µm diamond

suspension and finished with 0.05 µm colloidal silica suspension. Freshly prepared solution was

de-aerated with high purity Ar gas before and during each test. All tests were performed at room

temperature. Samples were kept immersed in the test solution for 1 hour at open circuit potential

before the start of each test. The extent of intergranular attack after DLEPR tests was observed

using a FEI XL-30 Scanning Electron Microscope (SEM).

2.4 Residual Stress and Hardness Characterization

In the sin2ψ method, a number of XRD measurements are made at different psi (ψ) tilts where ψ is the angle between the normal of the sample and the normal of the diffracting plane.

Presence of strain causes peak shift (2-theta peak position) and inter-planar spacing (d) can be measured at each ψ tilt. The stress was then calculated by fitting the data in the d versus sin2ψ

plot and knowledge of the plane-specific x-ray diffraction elastic constants (S1 and S2) of the

material[18].

Residual stresses induced in the material after LSP treatment were measured using the

sin2ψ technique with a Proto LXRD instrument (single axis goniometer using Ω geometry) with parameters as listed in Table 4.3. Calibration of the system was checked by collecting a diffraction pattern from a strain free standard polycrystalline SS316 powder prior to conducting the experiment in accordance with ASTM E915-10 (“Verifying the Alignment of X-ray

Diffraction Instrumentation for Residual Stress Measurement”) [19].

Table 4.3. XRD parameters used for residual stress measurements.

Item Description

Detector Position sensitive scintillation detector(PSSD) 20º 2θ

Radiation MnKα1 (λ = 2.10314 Aº)

Tilt angles 0, ± 2.58, ± 9.07,± 12.45,± 18.8,± 23.0

Aperture size 1 mm diameter

Plane(Bragg’s angle) {311}, 156º

X-ray elastic constant S2/2: 5.66 x 10-6 MPa-1

To measure residual stress through depth, samples were electro-polished to remove 10 -

50µm layers using 87.5:12.5 vol.% CH3OH:H2SO4 solution. Thickness of layer removed was calculated by measuring the thickness of the sample with a precise micrometer before and after electro-polishing. Strain and layer removal corrections were applied to the residual stress values using established procedures included in the software.

The relaxation of LSP induced residual stresses due to applied stress was observed by measuring residual stresses on the gage section of the same sample before and after constant load

SCC tests. Before testing, residual stresses were measured at least 2 times along and perpendicular to the loading direction on each sample. After constant load tests, residual stresses were measured twice on the gage section of samples that did not fail in 500 hours. The average values of surface residual stresses on the gage section along the loading direction have been reported in this study.

Hardness measurements were performed using a CSM Instruments Nano-Micro

Indentation system with a Berkovich indenter, a load of 100 mN with loading/unloading rate of

200 mN/min and a pause of 5 seconds at the peak load. At least 3 hardness values were recorded at each depth on polished cross-sectioned LSP treated samples.

2.5 SCC tests

a. Slow Strain Rate Test (SSRT)

Flat samples with the dimensions as shown in Figure 4.2 were machined using wire EDM

from sensitized sheets of Alloy 600, polished to 1200 grit, degreased with acetone, dried and

immersed in the test solution for 1 hour prior to straining. Samples were strained at rates of 2 x

10-6 /s and 10-6 /s for S1 and S2 sensitization conditions respectively. LSP was performed using

vinyl tape as the ablative layer and water as the confining medium with the parameters described

earlier. Samples were LSP treated on 2 opposite sides of the gage section as shown in Figure 4.2.

Slow strain rate tests (SSRT) were performed at room temperature with a mechanical testing

system (Cortest Inc. Willoughby, Ohio) driven by a servo motor and fitted with a custom built

environmental chamber. Load and displacement values were recorded periodically using

calibrated load cells and 2 LVDT’s respectively and samples were tested to failure. Strain was

calculated from the average displacement measured by the 2 LVDT’s. Solutions with

concentration 0.001 M or 0.005 M were prepared using reagent grade Na2S4O6 sodium tetrathionate (Sigma-Aldrich) and distilled water. For certain SSRT tests, the solution pH was adjusted to 3 with the appropriate amount of dilute H2SO4.

Figure 4.2. Schematic of the sample used for slow strain rate tests and constant load tests. All dimensions in mm.

42 b. Constant Load Tests

Constant load tests were performed by loading the samples (dimensions as shown in

Figure 4.2) using calibrated Proof rings (Cortest Inc. Willoughby, Ohio) to the predetermined loads/stresses and the time to failure was recorded using a timer switch. The test solutions were prepared according to the same procedure described earlier. Fracture surfaces of failed samples from SSRT and constant load SCC tests were observed in an FEI XL-30 SEM to ascertain the mode of failure.

3. Results In this section, we first present the effects of LSP on the microstructure, residual stresses and hardness. Next, SCC test results including SSRT and constant load tests in tetrathionate solution are described.

3.1 Microstructure The cross-section optical micrographs recorded in Figure 4.3(a,b) from the untreated (S1) and LSP-treated (LSP1) samples respectively show no changes in the grain size following the

LSP treatment. Figure 4.3(c) shows the microstructure of the sensitized sample in which a small area was LSP treated. No significant difference in grain size can be observed between the untreated and LSP-treated regions in the micrograph. Other studies on effects of LSP on microstructure have reported no significant changes in grain size due to the relatively lower plastic strain induced during the treatment [20].

a b K K

S S t tLSP treated surface e e e e l l C C o o r r

p p o o untreate LSP treated cr r Ka d a t t Si i to o en n e lR R Ce e op p ro o

pr r Figureot 4.3. Optical micrographs from cross t sections of sensitized a) S1, b) LSP1 (LSP treated)rD condition samples and c) untreated andD LSP treated (right) regions from surface of S1.aa a tt t ie Carbides (M7C3, M23C6) along the graine boundaries usually results along the grain o: : boundariesnA usually results in depletion of Cr fromA the surrounding regions, the extent of which u u dependsRg on the sensitization treatment. Chromiumg concentration profiles across grain boundaries eu u in sensitizedps samples (S1 and S2) are shown in Figures 4.4(c). Cr concentration was as low as 9.4 ot t r wt.%1 and 13.4 wt.% at the grain boundary in1 S1 and S2 respectively. This suggests that t, , D precipitation2 of carbides during both sensitization 2treatments led to Cr depletion. a0 0 t 1 Heat treatment conditions were chosen such1 that Alloy 600 was sensitized to induce e3 3 : IGSCCA7 susceptibility in tetrathionate solutions at 7room temperature. TEM micrographs from the u0 0 sensitizedg5 samples (S1 and S2) showing in carbide5 precipitation along grain boundaries can be u sC C tu 44u 1r r ,t t 2i i 0s s 1 observed in Figure 4.4(a, b) respectively. LSP being a mechanical surface treatment technique

influences the microstructure and TEM micrographs from the near surface region after LSP

treatment have been presented in Figure 4.5(a-f). The pressure pulse/shock wave generated

during LSP causes plastic deformation, which results in the formation of a high dislocation

density and misoriented sub-grains/crystallites that have sizes in the range of 30-200 nm in the

near-surface regions, which can be clearly seen in the BF micrographs in Figure 4.5(a-c).

a b

Figure 4.4. Bright field TEM micrographs showing the presence of carbides along grain boundaries from (a) S1 and (b) S2 condition samples. Chromium concentration profiles at and with distance into the matrix from grain boundaries in S1 and S2 is shown in (c).

a b

c d

e f

Figure 4.5. TEM Micrographs from LSP1 sample (a, b and c) showing high dislocation density and 30-200 nm size sub-grains/crystallites. (d) Bright field micrograph revealing twins, which are confirmed by the extra reflections at the expected positions for twins in FCC crystals in the [011] SAD pattern in (e) from the encircled region in (d). (f) Dark field micrograph recorded from the twin reflection encircled in red in (e) revealing narrow twins. In addition, narrow twins were observed occasionally as seen in the BF micrograph in Figure

4.5(d) and confirmed by the extra twin reflections at the expected positions for FCC crystals in

46 the [011] SAD pattern in Figure 4.5(e) from the encircled area surrounding the twins and revealed more clearly in the DF micrograph in Figure 4.5(f) recorded from a twin reflection. This is in accordance with other studies that have reported a high dislocation density and twinning after shot peening and LSP treatments [7,11,21].

3.2 Residual stresses, FWHM and hardness

Residual stresses versus distance from the peened surface, measured using the sin2ψ method in two orthogonal directions (X and Y), are shown in Figure 4.6. The residual stresses were of the order of about -500 MPa at the surface and the depth of the compressive residual stress layer was about 0.4 mm (in a 2 mm thick sample). The surface residual stresses have been significantly modified from tensile (100 MPa) before LSP to compressive (-500 MPa) after LSP.

X-ray diffraction peak broadening characterized by full widths at half maximum

(FWHM) values, which are a qualitative indication of the level of plastic strain introduced in a material were also recorded during residual stress measurements and have been plotted in Figure

4.7. In LSP treated Alloy 600, the level of plastic strain is higher at the surface (FWHM of ~3.5) but drops rapidly in the first 100µm and decreases gradually through depth to a more or less constant value after ~400 µm.

Indentation hardness at the surface after LSP was about 4.99 GPa and drops gradually to

3.75 GPa at 600 µm below the surface. The indentation hardness before LSP was about 3.28 GPa and the increase in hardness can be attributed to the repeated impacts during the LSP process.

Pressure pulse generated during LSP process have short pulse duration and attenuates as it travels through the material. This explains the gradient nature of residual stresses, plastic deformation and hardness.

Figure 4.6. Residual stresses through depth in two orthogonal directions in LSP treated Alloy 600.

Figure 4.7. Nano-indentation hardness and FWHM of diffraction peak through depth in LSP treated Alloy 600

3.3 Double Loop Electrochemical Potentiokinetic Reactivation Tests

The DOS values obtained from DLEPR for solution annealed and sensitized samples are shown in Table 4.4. The sensitization treatment at 650 ˚C resulted in increased DOS from 0.56 for AR to 0.78 and 2.04 for S2 and S1 conditions respectively.

A higher degree of sensitization indicates a deeper Cr depletion zone and greater the susceptibility to intergranular stress corrosion cracking. SEM micrographs of the surface after

DLEPR tests for AR, S1 and S2 samples are shown in Figure 4.8(a-c) respectively. The increased attack on the grain boundaries can be clearly seen in microstructure of the sensitized samples and is indicative of the amount of Cr depletion due to the sensitization treatment, which is also consistent with the Cr concentration profiles in Figure 4.4 determined using EDS.

Table 4.4. Degree of sensitization for the test conditions considered in this study.

Designation DOS,% Sensitization time at 650˚C,h AR 0.56 ± 0.06 0 S1 2.04 ± 0.09 5 S2 0.78 ± 0.08 2

Figure 4.8. SEM micrographs showing surface microstructure after DLEPR test for a) As received (AR), b) S1 and c) S2 conditions.

3.4 Slow Strain Rate Tests (SSRT)

SSRTs were used to quantify the susceptibility of sensitized Alloy 600 to IGSCC in tetrathionate solution at room temperature and particularly to understand the effects of LSP on the SCC susceptibility. Figure 4.9 shows the effect of sensitization treatments (different DOS), solution concentration and LSP on the SCC susceptibility in tetrathionate solution. Strain to failure, maximum stress and mode of failure from SSR tests conducted in this study are tabulated in Table 4.5. For comparison, sets of untreated (S1) and LSP treated (LSP1) Alloy 600 (same sensitization condition) were strained at the same rate in the same environment. Both S1 and S2 sensitization conditions were observed to be susceptible to IGSCC in the 0.001 M tetrathionate solution. Maximum elongations were 31% and ~23% for S1 and S2 respectively in the tetrathionate solution (0.001 M) as compared to ~60% in air (neutral environment). DLEPR test results (Table 4.4) and Cr depletion profiles (Figure 4.4) indicated that S2 was less sensitized as compared to S1 and hence a lower strain rate was used for SSR tests. This ensured that the mode of failure was intergranular SCC. Once the baseline conditions (DOS, strain rate and environment) were established, LSP treated samples were tested under the same conditions.

Since the LSP treatment was done after sensitization treatment, any changes can be attributed to the effects of LSP. The stress-strain curves shown in Figure 4.9 indicate higher yield stress (YS) and ultimate tensile strength (UTS) for LSP treated Alloy 600 compared with the untreated condition. This effect was seen in all LSP treated samples irrespective of the sensitization condition or environment. For example, YS was ~40% higher after LSP for the same sensitization condition. The strain to failure was also higher for LSP treated samples compared with untreated samples.

While SSRTs can be used to provide information about the susceptibility of an alloy in the given environment, constant/cyclic load tests are more representative of the stress state. Also,

the residual stresses induced by LSP are likely to relax on straining beyond the yield stress of the

material. The increase in yield stress and ultimate tensile strength after LSP would possibly

increase the threshold applied stress for IGSCC in a given environment. A single crack that

propagated through the cross-section of the sample was observed in all samples that were tested

in 0.001 M Na2S4O6 solution. This may be attributed to a combination of less sensitization, less aggressive environment and strain rate. Secondary cracks were observed in samples that were tested in the acidified (pH 3) 0.001 M Na2S4O6 solution thereby indicating that the acidified

solution was more aggressive towards sensitized Alloy 600. Low magnification SEM

micrographs of fracture surfaces of failed samples after SSRTs have been shown in Figure 4.10

(a, c, e and g) whereas the higher magnification images in Figures 4.10 (b, d, f and h) show inter

granular cracks in more detail. The mode of failure was (partial) IGSCC for sensitized and LSP

treated samples.

Figure 4.9. Stress-strain curves of sensitized and LSP treated Alloy 600 tested in 0.001 M tetrathionate solution at room temperature. Strain rate for S1, LSP1 was 2 x 10-6 /s while strain rate for S2, LSP2 was 10-6 /s. For tests designated with pH3, appropriate amount of H2SO4 was added to reduce the pH to 3.

Table 4.5. Summary of all results from SSR tests on sensitized Alloy 600

Specimen/Environment YS, MPa UTS, MPa Elongation,% Test variable Mode of Failure

S1_0.001M 280 514 31.27 %DOS IGSCC

LSP1_0.001M 528 731 31.14 LSP IGSCC

S1_0.005M 310 514 33.82 Solution concentration IGSCC

LSP, Solution LSP1_0.005M 554 725 41.83 concentration IGSCC

S2_0.001M 296 441 23.10 %DOS, strain rate IGSCC

%DOS, strain rate, LSP2_0.001M 520 787 42.50 LSP IGSCC

S1_0.001M_pH3 312 496 18.89 Solution pH IGSCC

LSP1_0.001M_pH3 563 808 29.08 Solution pH, LSP IGSCC

S1_Air 330 864 66.78 Neutral environment Ductile

S2_Air 326 753 61.47 Neutral environment Ductile

a b

c d

e f

g h

Figure 4.10. SEM micrographs showing fracture surface from a), b) S1; c), d) LSP1; e), f) S1_pH3 and g), h) LSP1_pH3. Environment: 0.001 M Na2S4O6, pH3 denotes acidified 0.001 -6 -1 M Na2S4O6. Strain rate: 2 x 10 s . Black boxes in the low magnification SEM micrographs in (a), (c), (e) and (g) highlight the locations for corresponding high magnification micrographs in (b), (d), (f) and (h) and white arrows point to intergranular cracks. 3.5 Constant Load Tests

Constant load tests were used to evaluate the effects of LSP on Alloy 600 under static

loading conditions in different solution concentrations. Figure 4.11 summarizes the LSP effects

on SCC resistance in terms of applied stress and time to failure. For the untreated samples, the

failure time decreased with increase in applied stress in 0.001M tetrathionate solution. The

applied loads were chosen to be between 80-140% of the YS load for sensitized Alloy 600. In

general, LSP treated samples tested at same loads had much longer time to failure or did not fail

in 500 hours after which the tests were terminated. Sensitized samples failed in shorter time

compared with the LSP treated samples for all loads and solution concentrations/pH tested.

The increase in time to failure after LSP can be attributed to the beneficial compressive

residual stresses (Figure 4.6) induced by LSP which were as high as -550 MPa at the surface at

the start of test, as well as near-surface hardening (Figure 4.7) caused by the increased

dislocation density and formation of misoriented, 30-100 nm-size sub-grains/crystallites, whose

boundaries act as obstacles to dislocation glide (Figure 4.5). Cracks were not observed in

sensitized Alloy 600 (S1) after 500 hours in 0.001M tetrathionate solution at a load of 300 MPa

which suggests a stress threshold for IGSCC. SEM micrographs of the fracture surfaces from

constant load tests for sensitized (S1) and LSP treated (LSP1) are shown in Figure 4.12(a, b)

respectively, and intergranular cracking was observed in both cases.

The effects of LSP on resistance to IGSCC in acidified tetrathionate solution (0.001 M

Na2S4O6 pH 3) were also investigated. The time to failures dramatically decreased for sensitized samples (S1) from 176 hours in 0.001M tetrathionate to 30 hours when the pH was reduced to 3.

In this test, the LSP treated sample showed multiple cracks in the gage section after 423 hours

(applied stress = 350MPa). At a lower applied stress (300 MPa), the LSP treated sample did not

show cracking even after 500 hours, whereas the untreated sample failed in about 45 hours in the

acidified tetrathionate solution.

a b

Figure 4.11. SEM micrographs of fracture surface showing intergranular cracking after constant load test (applied stress = 400MPa) in 0.001M tetrathionate solution of a) sensitized (S1) and b) LSP treated (LSP1) Alloy 600 samples.

Figure 4.12. Summary of results of constant load SCC tests on sensitized and LSP treated Alloy 600 samples in 0.001M Na2S4O6 (tetrathionate) solution. Open and closed symbols indicate samples that did not fail in 500 hours and those that failed, respectively. pH3 indicates acidified 0.001 M tetrathionate solution and the red line denotes the yield stress for sensitized Alloy 600.

4. Discussion The SSRT and constant load test results suggest that LSP has a significant effect on

improving the IGSCC resistance of Alloy 600. In this section, SSRT data were used to calculate

the SCC susceptibility index and quantify the effect of LSP induced residual stresses. In addition,

we discuss the effects of LSP under different applied load, solution concentration and DOS

conditions, with attention to mechanistic aspects.

4.1 SCC Susceptibility Index To quantify the effect of LSP on the susceptibility to IGSCC in tetrathionate solutions,

the following equation (Equation 1) proposed by Abe et al. [21] was used to calculate the SCC

susceptibility index (ISCC),

⎡ ⎤ ⎡ ⎤ 1+ en Pn ISCC = ⎢ SCC ⎥ x⎢ SCC −1⎥ (1) ⎣1 +en ⎦ ⎣ Pn ⎦

where and �� are strains at the maximum load in the load-elongation plot in air and ��

€ environment respectively. and �� are the maximum loads in the load-elongation plot in air ��

and environment respectively. ISCC is used to quantify the degree of susceptibility to SCC in a

particular solution with respect to a neutral environment at the same strain rate. Table 4.6

summarizes the ISCC for the tests as per SCC susceptibility proposed by Abe et al. [22].

SCC susceptibility of LSP treated sensitized Alloy 600 was lower than that of counterpart

untreated sensitized Alloy 600. Increasing the solution concentration from 0.001M to 0.005M

Na2S4O6 did not increase the ISCC appreciably but decreasing the pH to 3 had a more significant

effect. S2 was strained at a slower rate than S1, which explains the relatively higher

susceptibility to SCC in the same solution despite having lower DOS. The higher yield stress and

UTS contributed to the reduced SCC susceptibility for LSP samples. ISCC was lower even in

more aggressive environment where the pH was 3. SSRT tests involve straining the samples

continuously and residual stresses relax as the applied stress exceeds the yield stress [18]. Hence,

residual stresses are likely to be ineffective in mitigating SCC if the samples are plastically

deformed to large strains.

Table 4.6. Comparison of SCC susceptibility index using strain and maximum load relationship for sensitized and LSP treated Alloy 600.

Specimen/Environment Max load, N Strain at max load, % ISCC S1_0.001M 359 16.58 0.88 LSP1_0.001M 510 15.02 0.24 S1_0.005M 359 17.82 0.87 LSP1_0.005M 506 12.82 0.26 S2_0.001M 308 10.58 1.31 LSP2_0.001M 549 29.93 0.11 S1_0.001M_pH3 346 12.33 1.00 LSP1_0.001M_pH3 564 22.61 0.09 S1-Air 603 50.90

As noted earlier, the mode of failure was IGSCC even for the LSP treated samples which

have lower ISCC. It is important to note that the LSP treated samples show a much higher yield stress as compared with the untreated samples. This increase in yield stress is a result of LSP induced deep compressive residual stresses and near surface hardening from the increased dislocation density and refined sub-grains/crystallites, whose boundaries act as obstacles to dislocation glide. ISCC proposed by Abe at al. takes into account the strain, load at the maximum

load and is better suited to quantify the effects of residual stresses on SCC susceptibility. The

higher yield stress for LSP1 vs S1 and LSP2 vs S2 contribute to reduced susceptibility.

4.2 Effect of Applied Load and Solution Concentration

Constant load test results presented in Figure 4.11 suggest that lower degree of sensitization increased the time to failure in the same solution under similar applied loads.

Residual stresses of ~ -150 MPa were measured after the constant load test (applied stress =

400MPa) in cases where samples did not fail in 500 hours. This is lower than the initial surface residual stresses of ~ -550 MPa before testing. This indicates that a part of the residual stresses relaxed during loading. It is known that residual stresses will redistribute and relax as cyclic/static load is applied especially if it exceeds the yield stress of the material [23]. This was confirmed by measuring residual stresses on the gage section constant load SCC tests and the results have been summarized in Table 4.7. Also, residual stresses were close ~ 10MPa on LSP treated samples that had failed before 500 hours. This indicates that compressive residual stresses induced by LSP had a significant effect in the mitigating SCC. Moreover, this effect was seen in cases where the solution is more acidic (pH 3). The presence of compressive residual stresses

(albeit lower than those in the as LSP treated condition) after constant load tests (500 hours) suggests that threshold stress was higher.

Table 4.7. Residual stresses on the gage section along the loading direction after constant load tests in tetrathionate solution.

Sample Designation, Applied stress for Residual stress Residual stress Environment 500 hours, MPa before test, MPa after test, MPa LSP1, 0.001 M 350 -550 -156 LSP2, 0.001 M 400 -550 -172 LSP1, 0.001 M 500 -550 -14 LSP1, 0.001M pH 3 300 -550 -165

Constant stress tests on the sensitized, but non-LSP treated samples were conducted at stress levels both below and above the yield stress. At applied stresses below the yield stress, for example 300 MPa, failure due to SCC was not observed in the normal pH solution but occurred

at short times in the more aggressive environment (pH 3). The applied stress was close to or

above the yield stress of the sensitized samples but below the yield stress of LSP treated samples.

It must be noted that this increase in threshold stress may be lower or higher in more or less

aggressive solution/sensitization conditions respectively.

Tests at stresses above the yield stress of the LSP-treated samples were not conducted

because plastic yielding would cause relaxation of the compressive residual stresses and little or

no improvement in SCC life would be expected. Nevertheless, the important conclusion to be

drawn is that LSP considerably increases the threshold stress and life for failure from SCC. The

better resistance of LSP treated samples to SCC under constant load in tetrathionate solutions

may be attributed to the higher yield stress (from the hardening caused by increased dislocation

density and refined sub-grains/crystallites whose boundaries serve as obstacles to slip).

Tsai et al. [14] had observed increased resistance of shot peened sensitized Alloy 600 to

SCC in thiosulfate solutions and the beneficial effect was attributed to the presence of

compressive residual stresses though the residual stresses were not quantified. Sano et al. [8] also

observed increased resistance to crack initiation after LSP treatment on SS304 in high

temperature pure water. They observed that LSP (without coating) prevented SCC even if the

residual stresses introduced after LSP were as low as -100 MPa. In addition, LSP had prevented

crack initiation and mitigated crack propagation. Peyre et al. [24] reported that LSP acted like a

tensile stress inhibitor which contributed to the improved SCC resistance in boiling MgCl2

solution in SS316L. Recently, Lu et al. [25] attributed the increase in SCC resistance of ANSI

304 austenitic stainless steel in boiling MgCl2 solution to modified grain refinement and residual stresses due to LSP treatment. Lu and colleagues also observed relaxation of residual stresses after LSP treated AISI 304 SS samples were bent to form U-bend specimen [26]. Additionally,

59 these samples failed by SCC in boiling MgCl2 solution as compared with samples that were LSP treated after being bent into U- bend specimens that did not fail. This indicates that LSP induced compressive residual stresses significantly reduce the susceptibility of austenitic stainless steels to SCC. Results from this study suggest that compressive residual stresses and near surface hardening induced by LSP treatment substantially increased the resistance of alloy 600 to SCC in tetrathionate solution.

5. Conclusions

The findings of this study have led to the following conclusions.

1. DLEPR tests were used to quantify the DOS for sensitized Alloy 600. Sensitized Alloy 600

was susceptible to IGSCC in sodium tetrathionate solution at room temperature. A higher

DOS corresponded to greater susceptibility of sensitized Alloy 600 to SCC in tetrathionate

solution.

2. Deep compressive residual stresses were introduced and a substantial increase in hardness

and strength was observed after LSP in Alloy 600. Near-surface microstructure of LSP

treated Alloy 600 shows deformation bands, high dislocation densities, misoriented sub-

grains of 30-200 nm size and occasionally narrows twins.

3. Slow strain rate tests used to evaluate IGSCC susceptibility of sensitized Alloy 600 in

tetrathionate solution revealed that LSP substantially reduced the susceptibility of sensitized

Alloy 600 to IGSCC for conditions with different DOS, solution concentration and pH.

4. LSP treatment led to a significant improvement in resistance to SCC indicated by longer time

to failure or no failure at same applied stress in sensitized alloy 600 as shown by constant

load tests for different DOS and pH in tetrathionate solutions.

5. The improvements in SCC resistance by LSP treatment are attributed to a combination of

deep compressive residual stresses, near-surface microstructural changes and the associated

hardening and increased yield strength of the alloy.

6.0 Acknowledgments

The authors are grateful for financial support of this research from the Nuclear Energy

University Program (NEUP) of the US Department of Energy contract #102835 issued under prime contract DE-AC07-05ID14517 to Battelle Energy Alliance, LLC. We also gratefully acknowledge the contribution of the State of Ohio, Department of Development and Third

Frontier Commission, which provided funding in support of “Ohio Center for Laser Shock

Processing for Advanced Materials and Devices” equipment in the Center that was used in this work. The authors are also grateful for the support of David Simmermon (Cincinnati State

Technical and Community College) for help with the LSP setup. Any opinions, findings, conclusions, or recommendations expressed in these documents are those of the author(s) and do not necessarily reflect the views of the DOE, State of Ohio, Department of Development.

7.0 References

[1] D. Van Rooyen, Corrosion 31 (1975) 327–337.

[2] R. Bandy, D. Van Rooyen, Corrosion 40 (1984) 281–289.

[3] J. Kai, C. Tsai, T. Huang, M. Liu, Metall. Trans. A 20 (1989) 1077–1088.

[4] S. Hsu, S. Tsai, J. Kai, C. Tsai, J. Nucl. Mater. 184 (1991) 97–106.

[5] R. Bandy, R. Roberge, R. Newman, Corros. Sci. 23 (1983) 995–1006.

[6] R.C. Newman, R. Roberge, R. Bandy, Corrosion 39 (1983) 386–390.

[7] P. Peyre, R. Fabbro, Opt. Quantum Electron. 27 (1995) 1213–1229.

[8] Y. Sano, Y. Sakino, N. Mukai, M. Obata, I. Chida, T. Uehara, M. Yoda, Y.C. Kim, Mater. Sci. Forum 580-582 (2008) 519–522.

[9] S.J. Green, Int. J. Press. Vessel. Pip. 25 (1986) 359–391.

[10] P. Peyre, X. Scherpereel, L. Berthe, C. Carboni, R. Fabbro, G. Be, Mater. Sci. Eng. A 280 (2000) 294–302.

[11] A. Gill, A. Telang, S.R. Mannava, D. Qian, Y.-S. Pyoun, H. Soyama, V.K. Vasudevan, Mater. Sci. Eng. A (2013).

[12] C.S. Montross, T. Wei, L. Ye, G. Clark, Y. Mai, Int. J. Fatigue 24 (2006) 1021–1036.

[13] C. Ye, A. Telang, A.S. Gill, S. Suslov, Y. Idell, K. Zweiacker, J.M.K. Wiezorek, Z. Zhou, D. Qian, S. Ramaiah Mannava, V.K. Vasudevan, Mater. Sci. Eng. A 613 (2014) 274–288.

[14] W. Tsai, C. Chang, J. Lee, Corrosion 50 (1994) 98–105.

[15] P. Hurrell, D. Everett, A. Gregg, S. Bate, Proc. ASME Press. Vessel. Pip. Conf. 6 (2006) 801–812.

[16] Y. Lim, H. Kim, J. Han, J. Kim, H. Kwon, Corros. Sci. 43 (2001) 1321–1335.

[17] M. Ahn, H. Kwon, J. Lee, Corros. Sci. 51 (1995) 441–449.

[18] P.J. Withers, H.K.D.H. Bhadeshia, Mater. Sci. Technol. 17 (2001) 355–365.

[19] Measurement, Standard Test Method for Verifying the Alignment of X-Ray Diffraction Instrumentation for Residual Stress, ASTM International, West Conshohocken, No Title, 2010.

[20] Y. Sano, M. Obata, T. Kubo, N. Mukai, M. Yoda, K. Masaki, Y. Ochi, Mater. Sci. Eng. A 417 (2006) 334–340.

[21] I. Nikitin, I. Altenberger, Mater. Sci. Eng. A 465 (2007) 176–182.

[22] S. Abe, M. Kojima, Y. Hosoi, in:, 1979, pp. 294–304.

[23] B.L. Boyce, X. Chen, J.O. Peters, J.W. Hutchinson, R.O. Ritchie, Mater. Sci. Eng. A 349 (2003) 48–58.

[24] P. Peyre, L. Berthe, R. Fabbro, J. Mater. Eng. Perform. 9 (2000) 656–662.

[25] J.Z. Lu, H. Qi, K.Y. Luo, M. Luo, X.N. Cheng, Corros. Sci. 80 (2014) 53–59.

[26] J.Z. Lu, K.Y. Luo, D.K. Yang, X.N. Cheng, J.L. Hu, F.Z. Dai, H. Qi, L. Zhang, J.S. Zhong, Q.W. Wang, Y.K. Zhang, Corros. Sci. 60 (2012) 145–152.

5.0 Surface Grain Boundary Engineering of Alloy 600 for Improved Resistance to Stress Corrosion Cracking

Abhishek Telanga*, Amrinder S. Gillb, Deepthi Tammanaa, Xingshuo Wena, Mukul Kumarc, S. Teysseyred, Seetha R. Mannavaa, Dong Qiane and Vijay K. Vasudevana aDepartment of Mechanical and Materials Engineering, University of Cincinnati, Cincinnati, OH bAK Steel, Research Center, 705 Curtis Street, Middletown, OH cLawrence Livermore National Laboratory, Livermore, CA dIdaho National Laboratory, Idaho Falls, ID eDepartment of Mechanical Engineering, University of Texas at Dallas, Richardson, TX

Published in Materials Science and Engineering A, Volume 648, pp. 280-288 (2015); doi:10.1016/j.msea.2015.09.074

Abstract

In this paper, we demonstrate a novel method for grain boundary engineering in Alloy 600 using iterative cycles of ultrasonic nanocrystal surface modification (UNSM) and strain annealing to modify the near surface microstructure (~ 250 µm) for improved stress corrosion cracking (SCC) resistance. These iterative cycles resulted in increased fraction of special grain boundaries whilst decreasing the connectivity of random grain boundaries in the altered near surface region. A disrupted random grain boundary network and a large fraction of low CSL boundaries (Σ3-Σ27) reduced the propensity to sensitization. Slow strain rate tests in tetrathionate solutions at room temperature show that surface GBE lowered susceptibility to intergranular SCC. Detailed analysis of cracks using Electron Back-scattered Diffraction showed cracks arrested at J1 (1-CSL) and J2 (2-CSL) type of triple junctions. The probability for crack arrest, calculated using percolative models, was increased after surface GBE and explains the increase in resistance to SCC.

1. Introduction

Grain boundary engineering (GBE) has been demonstrated as a viable method for improving the resistance to creep [1,2], hydrogen embrittlement [3], fatigue [4,5], corrosion [6–

8] and stress corrosion cracking [2,9–14] (SCC) in austenitic stainless steels (SS), Ni based alloys and superalloys. GBE involves increasing the frequency of coincident site lattice (CSL) grain boundaries whilst disrupting the random grain boundary network through thermomechanical processing routes. Low grain boundary energy, resistance to grain boundary sliding and intergranular degradation, less susceptibility to impurity or solute segregation are some reasons that contribute to the “special” nature of CSL boundaries.

Thermo-mechanical processing routes involving cold rolling or uniaxial tension/compression and subsequent annealing have been used to increase the frequency of CSL boundaries[6,15]. One approach involves a single cycle of pre-straining the material followed by annealing at comparatively lower temperature for a long time. A multi-cycle approach including steps of moderate strains (6-30%) followed by relatively high temperature annealing for short times has also shown to increase the special grain boundary fraction[16,17]. In addition, the multi-cycle approach results in a disrupted random grain boundary network that correlates to improvements in fatigue, creep and corrosion resistance.

Detailed studies carried out by Bi et al. [18] have established that twin boundaries(especially coherent Σ3) are more resistant to carbide precipitation and corrosion because the atomic structure is highly coherent as compared to high angle grain boundaries. In particular, Σ3 and Σ9 boundaries in grain boundary engineered SS304 have been observed to more resistant to sensitization while Σ27 and other CSL boundaries were not really “special” in terms of their resistance to sensitization and thus intergranular stress corrosion cracking

(IGSCC)[19]. Thus, it has been suggested that increased fraction of Σ3 and Σ9 boundaries would likely improve the corrosion and stress corrosion resistance.

Alloy 600 and austenitic stainless steels have been known to be susceptible to stress corrosion cracking (SCC) in polythionic acid environments[20–25]. Susceptibility to SCC at low temperature in tetrathionate and thiosulfate environments has been attributed to Cr depletion in the area surrounding the grain boundary. A reduction in Cr depletion by disrupting the random grain boundary network or increasing the fraction of special boundaries should decrease the susceptibility to sensitization and SCC [6,18,19]. While GBE has been studied extensively to improve resistance to intergranular cracking, surface GBE has not been explored to the same extent.

In this paper, we propose a novel approach to engineer the near surface region by using ultrasonic nanocrystalline surface modification (UNSM)/ultrasonic peening followed by annealing to increase the fraction of special boundaries. Further, we present and discuss the effect of this surface grain boundary engineered material on the SCC behaviour in tetrathionate solution. To the best of our knowledge, there are no studies investigating the effects of surface

GBE on SCC behaviour of Alloy 600.

2. Materials and Methods

2.1 Materials

Alloy 600 plate (2 mm thickness) with chemical composition as shown in Table 1 was sectioned into 15 mm x 15mm coupons using a wire EDM. The as received material was in annealed condition with a grain size of ~ 10 µm. UNSM is an advanced surface treatment that uses ultrasonic energy to strike a target (material surface) with a WC tip at a frequency of 20 kHz to induce strain in the near surface region of the material.

Table 1. Chemical composition of the Inconel Alloy 600 used in this study.

C Mn Si S Cr Fe Co Cd Ti Cu P Al Ni 0.001 0.001 0.08 0.16 0.18 15.05 8.05 0.16 0.01 0.18 0.1 0.08 Bal. max. max.

The amount of strain can be controlled by modifying the static and dynamic loads. A

schematic of the UNSM process is shown in Figure 1. Static load (Pst), amplitude of ultrasonic vibration, scan speed and overlap ratio can be controlled during processing. Details of UNSM have been reported elsewhere in literature [26,27]. For grain boundary engineering, coupons were peened using a LM20 UNSM system (DesignMecha) and subsequently annealed in a lab furnace for 10 minutes at 950 ˚C or 1000 ˚C, then water quenched (WQ). Processing details for surface GBE are listed in Table 2. AR and ARGBE conditions have been grouped together as Set

1 while SA and SAGBE are categorized as Set 2. The static load was 20 N and the amplitude of ultrasonic vibration was 8 µm. A scan speed of 3000 mm/minute and overlap interval of 30 µm was used for UNSM processing in this study.

Figure 1. Schematic of the ultrasonic nanocrystal surface modification (UNSM) technique.

After GBE treatments, samples were sectioned and cross sections were mounted in a

conducting epoxy. For EBSD, each sample was ground to 1200 grit, electropolished in 87.5:12.5

vol.% CH3OH:H2SO4 solution at 24 V, 15 s and finally polished with 0.05 µm colloidal silica

suspension. EBSD orientation mapping was performed in a FEI XL-30 SEM with step size of 2

µm at 30 kV. OIM scans were analysed with the TSL OIM Analysis (version 7.1) package to

calculate grain boundary character distribution (GBCD), grain size, boundary fractions and triple

junction fractions. CSL grain boundaries were categorized according to Brandon criterion of Δθ

≤ 15˚Σ-1/2 [28]. Boundaries with 3 < Σ < 29 were considered to be CSL boundaries whereas

boundaries with Σ > 29 were considered random high angle boundaries (HABs) and Σ =1 as low angle boundaries (LABs). For triple junction analysis, only Σ3, Σ9 and Σ27 were considered as

CSL boundaries.

Table 2. Designation and corresponding details of processing used in this study.

Designation Detail

AR As received

ARGBE AR + 3 cycles of (UNSM + annealing at 950˚C, 10 min, WQ)

SA AR + Solution annealing at 1050˚C,10 min, WQ

SAGBE SA + 3 cycles of (UNSM + annealing at 1000˚C, 10 min, WQ)

2.2 Residual stress and FWHM

Residual stresses were measured using sin2ψ technique with a Proto LXRD system,

MnKα radiation and (311) peak of the austenite phase. To measure residual stress through depth,

coupons were electropolished using 87.5:12.5 vol.% CH3OH:H2SO4 solution to remove 10-50

µm layers. Full width at half maximum (FWHM) data was also recorded for each depth.

2.3 Double loop electrochemical potentiokinetic reactivation (DLEPR) tests

Baseline and grain boundary engineered samples were given a sensitization treatment at

650˚C, 2h (water quenched) to induce precipitation of carbides. These samples were

mechanically ground to 1200 grit, wet polished with 1 µm diamond suspension and finished with

0.05 µm colloidal silica suspension DLEPR tests were performed in accordance with ASTM

G108-94 in a solution composed of 0.01 M H2SO4 + 20 ppm KSCN using a Gamry Potentiostat

(Reference 600). Samples were kept immersed in the test solution for 1 hour at open circuit

potential before the start of each test. The scan rate was set at 0.5 mV/s for activation and

reactivation loop and the sample size was 1 cm2. Freshly prepared solution was de-aerated with high purity Ar gas before and during each test. All tests were performed at room temperature.

The following procedure was used to quantify sensitization in the annealed and GBE material after sensitization[29]. The degree of sensitization is reported as DL-EPR value

(designated as R in %) which is the ratio of the current density in reactivation loop to that in the activation loop times 100.

�= ��/��×100 (1)

The DL-EPR value obtained is normalized with various parameters like grain boundary area

(GBA), grain size, mean lineal intercept length (MIL). It should be noted that twins have been excluded from grain size analysis. The DL-EPR value of a given alloy condition (with ASTM grain size number of G’) is normalized with the grain size (with ASTM grain size number of G) of the as-received material (SA) and is given by:

Rʹ = R x√2G ʹ -G (2)

The DL-EPR values were also normalized with grain boundary area (Sv, expressed in

mm2/mm€ 3):

RGBA = R /Sv (3)

68 € The DL-EPR values were also normalized with mean intercept length (designated by l):

RMIL = R /Sv (4)

Here l is expressed in µm. The grain boundary area and mean lineal intercept length can be € obtained from the number of intercepts per unit length (NL) of the test line [30] . The NL can be calculated from the following equation [30]:

G = − 3.2877+6.6439log10 NL (5) where NL is a number of intercepts per unit length and G is the ASTM grain size number. The € mean lineal intercept (l) and the GBA, Sv, are given by [30]:

l =1/NL Sv = 2NL (6)

2.4 Slow strain rate tests (SSRT)

€ GBE samples were prepared by UNSM treating 2 sides a 2 mm thick Alloy 600 (in AR and SA conditions), then strain annealing. This was repeated 3 times with the same process parameters as listed in Table 2. All samples were then given a sensitization treatment at 650 ˚C,

2h. Flat samples with gage length of 6 mm were fabricated using wire EDM, polished to 1200 grit, degreased with acetone, dried and immersed in the test solution for 1 hour prior to straining.

Samples were strained at rate of 2 x 10-6 s-1 for all tests. Slow strain rate tests (SSRT) were

performed with a SSRT system (Cortest Inc. Willoughby, Ohio) driven by a servo motor and

fitted with a custom built environmental chamber. Load and displacement values were recorded

periodically and samples were strained to failure. Some SSRTs were interrupted after 10%

nominal strain and held at constant load for 24 hours or failure. Test solutions (0.001- 0.01 M)

were prepared using reagent grade Na2S4O6 sodium tetrathionate (Sigma-Aldrich) and distilled water. For certain tests, solution pH was adjusted with the appropriate amount of dilute H2SO4.

1. Results

3.1 Residual stresses and FWHM

Residual stresses after ultrasonic peening were measured at different depths and plotted in Figure 2(a). The in-plane residual stresses are of the order of -1200 MPa and -700 MPa along

(phi 90) and perpendicular (phi 0) to the scanning direction respectively. Differences in residual stresses in the 2 directions are attributed to the processing parameters and have been reported previously [26,31]. In addition, FWHM data plotted in Figure 2(b) shows high level of plastic strain at the surface which decreases gradually through depth. Significant peak broadening is observed at the surface and FWHM decreases to ~50% as represented by the normalized FWHM scale (secondary axis). The magnitude of residual stresses and FWHM indicate a high level of strain and plastic deformation that varies through depth. Significant strain and plastic deformation is observed within the first 250-300 microns after ultrasonic peening and hence the near surface region is likely to show strain induced microstructural changes after annealing.

b a

Figure 4. (a) Residual stresses and (b) FWHM after UNSM in Alloy 600. 3.2 Microstructure

In this study, GBE treatment was applied to 2 sets of Alloy 600 with different initial starting conditions (AR and SA). An experimental matrix with different UNSM parameters and annealing temperatures was used to observe changes in grain boundary character distribution

(GBCD) in the near surface region (~250µm). GBCD statistics obtained were compared with

those from thermo-mechanical processing routes in the FCC materials including Alloy 600 that

have been reported previously [6,7,16,19].

Grain boundary maps from the AR and SA conditions are shown in Figure 3 (a) and (c),

respectively. The microstructure obtained as a result of repeating the (peening + annealing) cycle

for 3 times, designated as ARGBE and SAGBE are shown in Figure 3 (b) and (d), respectively.

All orientation maps were obtained from 200-250µm layer perpendicular to the peened side. For

analysis of triple junctions, only Σ3, Σ9 and Σ27 have been considered. Triple junctions with no

CSL boundaries, 1 CSL boundary, 2 CSL and 3 CSL boundaries have been classified as J0, J1,

J2 and J3, respectively.

Grain boundary fractions (length), grain size and triple junction fractions for the 2 sets

have been summarized in Figures 4 (a-c). In set 1(AR), the HAB fraction (length) decreased

from 58% to 28%in ARGBE while the CSL fraction increased from 36% to 66%. The fraction of

Σ3 boundaries increased from 27% in the AR condition to 53% after GBE. More importantly,

fractions of Σ9 + Σ27 increased from 3.4% (AR) to 10.2% (ARGBE). An increase in the Σ9, Σ27 fractions has been attributed to the Σ3 regeneration model proposed by Randle et al. [32,33]. In addition, triple junction fractions extracted from the orientation maps show a sharp drop in fraction of J0 type from 48% to 10%. J2 and J3 fractions increase from 3%(AR) to

12%(ARGBE) and 4%(AR) to 29%(ARGBE) respectively. These changes in triple junction fractions are consistent with the GBE model proposed by Kumar et al. [16,34].

In set 2, the HAB fraction (length) decreased from 49% to 27%in SAGBE while the CSL fraction increased from 37% to 63%. The fraction of Σ3 boundaries increased from 31% in the

SA condition to 54% after GBE. Σ9 + Σ27 fractions increased from 2.2% (SA) to 7% (SAGBE).

Similarly, J0 fraction dropped from 43% to 12% while J2 and J3 fractions increased from 3%

(SA) to 11% (SAGBE) and 3% (SA) to 21% (SAGBE) respectively.

a Surface

b Surface GBE layer

c Surface

d Surface

Figure 5. EBSD orientation maps showing the microstructure in the cross-section for (a) AR (b) ARGBE, (c) SA and (d) SAGBE conditions. Black lines denote random high angle grain boundaries and grey lines denote CSL boundaries (Σ ≤ 27). Grain size increased modestly from 9 µm to ~12 µm in set 1 and from 11 µm to 15 µm in set 2 after grain boundary engineering.

Figure 6. (a) Grain boundary character distribution and grain size , (b) CSL boundary fraction and (c) triple junction fraction for each processing condition. Note that boundary and triple junction fractions are from ~ 200 µm from the top surface. 3.3 Sensitization

After a sensitization treatment of 650˚C, 2h (water quenched) and DLEPR tests were used to quantify the degree of sensitization for each condition. The as-received material was in annealed condition with a relatively finer grain size (~9 µm). In case of the SA condition, the larger grain size and more carbon in the solid solution contributed to the increase in susceptibility to sensitization. DOS values increased from 0.56 to 2.58 after the annealing treatment prior to sensitization as summarized in Table 3.

Table 3. CSL fraction and corresponding degree of sensitization (R) values for each processing condition.

Sample CSL fraction,% R R' RGBA RMIL AR 36.2 0.56 0.5593 0.0053 0.0299 ARGBE 65.9 0.06 0.0217 0.0015 0.0012 SA 37.7 2.58 2.5766 0.0323 0.1032 SAGBE 62.5 0.04 0.0285 0.0006 0.0011

a b

c d

Figure 7. SEM images showing microstructure after DLEPR test for (a) AR, (b) ARGBE, (c) SA and (d) SAGBE conditions. Samples were given a sensitization treatment for 2 hours at 650˚C to induce sensitization prior to DLEPR tests.

More importantly, DOS values (R) for ARGBE was 0.06% as compared with 0.56% for the

AR condition (Set 1). Similarly, DOS value was much lower (0.04%) for SAGBE condition as

compared with (2.58%) for the SA condition. Micrographs obtained after DLEPR tests for each

condition are shown in Figure 5. Note that DLEPR tests were performed on the surface (within

50 µm from peened surface) and after sensitization. DOS values normalized by grain size, (R’),

grain boundary area (RGBA) and mean intercept length (RMIL) also show similar trends. The dramatic reduction in DOS has been correlated with increased CSL fractions after thermomechanical processing [6,19,29]. The decrease in the DOS indicates higher SCC resistance and results from SSRTs in tetrathionate solution are presented in the next section.

3.4 Slow strain rate tests

The increase in the CSL fractions and J2/(1-J3) fractions are indicative of the disruption in the grain boundary network after grain boundary engineering. A decrease in DOS (calculated from DLEPR tests) implies less Cr depletion after grain boundary engineering. The lower DOS can improve the SCC resistance of Alloy 600 in thiosulfate and tetrathionate solutions. Hence,

SSR tests were performed to observe and quantify the decreased susceptibility to IGSCC after

GBE. Stress-strain curves obtained from SSR tests in tetrathionate solution are as shown in

Figure 6. Tests were performed at the same strain rate and in the same environment show a difference in SCC behavior. For Set 1, the elongation to failure was 85% for ARGBE as compared to 68% for AR condition in 0.01 M Na2S4O6 solution. However, both AR and ARGBE

showed ductile mode of a failure with no intergranular cracking. DOS values obtained from

DLEPR tests had indicated that both the AR and ARGBE conditions were not severely sensitized

so a more aggressive environment was used during SSRT. In acidified 0.001 M Na2S4O6 solution

with pH 3 (pH reduced by addition of dilute H2SO4), the strain to failure for AR was 26% as

compared to ~50% for ARGBE condition. Addition of sulfuric acid to reduce the pH of the

solution has been shown to be more aggressive towards austenitic stainless steels and Ni alloys

[22,35].

In case of Set 2, the elongation to failure was only 15% for SA condition while SAGBE

condition was about 26% in 0.01 M Na2S4O6 solution. Analysis of the fracture surface after tests

showed mixed mode of failure for SAGBE condition while SA condition showed 100%

intergranular cracking. DLEPR tests conducted previously had indicated higher sensitization as

compared to the as received material. The low elongation to failure in the SA condition and

intergranular mode of failure indicates that the material was more severely sensitized as

compared to the AR condition. Secondary cracks from the gage sections of samples were

analyzed in greater detail and discussed in the next section.

To quantify the effect of LSP on the susceptibility to IGSCC in tetrathionate solutions,

the following equation proposed by Abe et al. [36] was used to calculate the SCC index (ISCC),

⎡ ⎤ ⎡ ⎤ 1+ en Pn ISCC = ⎢ SCC ⎥ x⎢ SCC −1⎥ (7) ⎣1 +en ⎦ ⎣ Pn ⎦

where and �� are strains at the maximum load for load-elongation plot in air and ��

€ environment respectively. and �� are the maximum loads for load-elongation plot in air and ��

environment respectively. Table 4 summarizes the results from SSR tests in 0.01 M Na2S4O6

solution. For Set 1, the ISCC drops from 0.6 to 0.13 after GBE in acidified 0.001 M Na2S4O6

solution(pH = 3). In case of Set 2, GBE reduces the ISCC to 1.21 as compared to 2.17 for the SA

condition.

Kikuchi patterns degrade due to strain and SAGBE samples had undergone close to 30%

strain before failure during the SSR test in 0.01 M Na2S4O6 solution. Further, SA and SAGBE

samples were strained to 10% (nominal strain) and held for 24 hours (or failure) at the same

stress level (± 5N). The SA sample failed in 8 hours after 10% strain while the SAGBE sample

did not fail in 24 hours (after 10% strain) and the test was terminated. Cracks were observed on

the gage section of both samples and were analyzed in greater detail. EBSD orientation maps

from regions including crack tips and their analysis are presented in the next section.

Table 4. Summary of Stress-strain data from SSRT for each processing condition. All tests were performed at room temperature Strain rate was 2 x 10-6 /s.

Figure 8. Stress-strain curves for baseline and grain boundary engineered samples (after -6 sensitization at 650˚C, 2h, water quenched) in 0.01 M Na2S4O6 solution. Strain rate: 2 x10 /s. For tests designated PH3, dilute H2SO4 was added to 0.001 M Na2S4O6 solution to reduce pH to 3.

3.5 SCC behavior

Detailed analysis of the fracture surface and secondary cracks from the gage section after

SSR tests shows intergranular nature of cracking. EBSD orientation maps from areas including cracked boundaries were used to assess the cracking behavior. In case of SA condition, elongation to failure was 16% and the mode of failure was intergranular in nature as shown in

Figure 7 (b). A number of secondary cracks were observed in the gage section as shown in

Figure 7 (a, c). Image quality (IQ) map overlaid with grain boundaries in Figure 7(d) shows the cracks following an intergranular path.

Figure 9. (a) SEM micrograph showing fracture surface and secondary cracking for SA condition after SSRT. (b) Micrograph showing secondary cracks in the gage section that were analyzed using EBSD. (c) High magnification micrograph of the fracture surface showing intergranular cracking. (d) Image quality EBSD map showing intergranular cracking overlaid with high angle boundaries in black and CSL boundaries in red. White arrow indicates the loading direction.

In case of SAGBE condition, a mixed mode of failure was observed as shown in Figure 8 (a, b). Very few cracks were observed on the gage section and even these did not propagate the sample width or depth. A secondary crack, shown in Figure 8(c), that was analyzed in greater detail showed crack deflection at multiple locations. For instance, the crack was observed to have been arrested at a Σ3 boundary. Note that the sample was strained to almost 30% and the kikuchi patterns degrade in locations of large strain especially near crack tips.

Figure 10. (a) SEM micrograph showing fracture surface and secondary cracking for SAGBE condition after SSRT. (b) High magnification micrographs of the fracture surface showing mixed mode of cracking. (c) Micrograph showing the location of a secondary crack that was analyzed in detail. (d) Image quality (IQ) map showing intergranular cracking overlaid with high angle boundaries in black and CSL boundaries in red. White arrow indicates the loading direction.

To analyze the interaction of SCC cracks with CSL boundaries in greater detail, we interrupted the SSRT test after nominal strain was 9% and the sample was held at this load/stress

(± 5N) for 24 hours or till failure. The SAGBE sample did not fail (test stopped after 24 hours) while the SA sample failed in less than 6 hours.

c d

Figure 11. (a, c) IQ maps showing crack arrest locations (black arrows) in SAGBE sample. (b, d) Corresponding skeletonized orientation maps showing the character of grain boundaries. Random high angle grain boundaries are colored black while CSL boundaries are colored red. The direction of loading is horizontal. In case of SAGBE sample, multiple instances of crack arrest were observed on the gage section after the test. Orientation maps were recorded from multiple locations around the cracks.

IQ and corresponding grain boundary character maps presented in Figure 9 (a, c) and (b, d)

81 respectively show the character of grain boundaries ahead of the crack front. Cracks were observed to have been “arrested” on encountering a J2 type of triple junction. Several other instances where cracks were arrested or deflected from their path were observed in the SAGBE sample. In sharp contrast, the SA sample showed no instances of crack arrest or deflection as seen in Figure 10. More importantly, the cracks propagated the entire width of the sample and failed in 6 hours.

Figure 12. IQ map overlaid with grain boundaries from SA condition after the interrupted SSR test. Black lines denote random high angle boundaries while CSL boundaries are colored red. 4. Discussion

4.1 Microstructure evolution

Strain induced by UNSM was qualitatively measured by FWHM (shown in Figure 2) after the surface treatment. After 3 cycles of UNSM and subsequent annealing, the microstructure in the near surface region was significantly different as compared with the original microstructure.

The network of random grain boundaries was observed to have broken down to smaller clusters after GBE. This disruption in random grain boundary networks has been reported after

82 conventional thermo-mechanical processing routes in the literature [6,34]. The increase in CSL boundary fractions (by length) observed after SGBE in the near surface regions is similar to those reported after conventional single and multi-step thermo-mechanical processing. Also, the fracture characteristics are dependent on the distribution and interconnectivity of boundaries prone to crack propagation [16]. Therefore, changes to the triple junction distributions provide a direct co-relation to failure characteristics. In this study, we observed J2 and J3 fractions were significantly higher while the J0 fraction decreased after GBE in the near surface region. These characteristics are consistent with other observations of triple junction distribution after GBE

[16,34,37].

The mechanism is likely to be similar to conventional GBE with the UNSM induced strain that decreases gradually from the near surface region to the interior. The multi-cycle treatment allows more possibilities for grain boundary migration and this further disrupts the random grain boundary network. Multiple interactions of twin boundaries (and its variants) with other random boundaries may introduce disruptions in the boundary network[34,38,39]. Though the strain induced by UNSM has not been quantified, residual stress and FWHM (Figure 2) provide a qualitative description of the strain in the near surface region. UNSM or other mechanical surface treatments can induce strain in regions to ~ 250-500 µm from the surface and thus could be used for near net shaped components or areas particularly prone to cracking. Alyousif et al.

[40] reported a single step shot peening and annealing treatment to modify the microstructure in

304 stainless steel. They observed a decreasing gradient in twin fractions from the surface up to

100 µm. In this study, we used a multi-cycle approach to increase the CSL fractions and disrupt the random high angle grain boundary network to a depth of 250 µm.

4.2 Effect of increased CSL fraction on corrosion behavior

The decrease in DOS after GBE in austenitic stainless steels and Alloy 600 has been reported previously in a number of studies [6,7,18,19,29]. More significantly, this decrease in DOS has been attributed to increased fraction of twin boundaries and its variants. Additionally, CSL boundaries have been shown to be more resistant to precipitation of carbides and Cr depletion

[18,41]. Jones and Randle [19] reported that Σ3 (~ 97%) and Σ9 (~ 80%) were resistant to sensitization while Σ27 and random high angle boundaries were attacked and not particularly resistant. A decrease in DOS by almost an order of magnitude suggests that the high CSL fraction as well as numerous disruptions in HAB network reduced Cr depletion indicated by the lower DOS.

We observed that that cracks were deflected or mitigated on encountering J2 type of triple junctions after surface GBE as shown in Figure 9. Similarly, improvements in high temperature fatigue [4] and hydrogen embrittlement [3] have been attributed to an increase in fractions of special boundaries particularly when the mode of failure is intergranular. Intergranular attack and stress corrosion cracks usually initiate at the surface and propagate to the interior along random grain boundaries. Though modification in grain boundary network is confined to ~200-

300 µm in the near surface region, cracks would encounter more resistant CSL boundaries. Other studies show that Σ3 boundaries are significantly more resistant to cracking than HAB

[10,38,42]. While a few Σ3 boundaries have been observed to crack and the “special” nature of

Σ3n variants (n = 2, 3) with respect to their cracking resistance is under debate. Additionally, even some random HABs have been observed to “resistant” if they are unfavorably oriented to the applied stress direction [38,43].

A few percolation models have been proposed to determine the probability for crack deflection or arrest when the crack encounters a resistant grain boundary [16,44–46]. Marrow et

84 al.[45] observed cracks that were arrested by J1 and J2 type of triple junctions in sensitized 304 stainless steels after SCC tests in tetrathionate solutions at room temperature. Gertsman and

Bruemmer[38] also observed cracks that were arrested at J2 type junctions and coherent Σ3 boundaries were resistant to SCC. They also reported several Σ9 and Σ27 boundaries and Σ3 boundaries with large deviations from ideal misorientation had cracked. The probability for crack arrest at a triple junction was calculated according to the 3 models proposed by Kumar et al. [39], Palumbo et al. [46] and Marrow et al. [45] and have been plotted in Figure 11.

Palumbo et al. also proposed the following relation (Equation 8) to calculate the probability

(P) of crack arrest,

P = f 2 +2 f f 1− f (8) sp [ 0 sp ()sp ]

Where fsp is the fraction of special boundaries and f0 is the fraction of interfaces in the distribution€ that are unfavorably oriented to the stress axis. This model relates the probability for crack arrest to the fsp but does not consider the spatial arrangement of these special boundaries in the microstructure.

It has been observed that fracture characteristics are primarily dependent on spatial distribution and interconnectivity of boundaries prone to crack propagation. To quantify the effect of improvement in the GBCD on the spatial connectivity of grain boundaries in two dimensions, triple junction distribution was evaluated [16]. Assuming J3 type of junctions to be non-entities in crack propagation, Kumar et al.[16] proposed (Equation 9) that the percolative paths in the microstructure would be broken if the following inequality holds,

J2 P = ≥ 0.35 (9) R (1− J3)

€ 85

Marrow et al.[45] observed cracks arrested at both J2 and J1 type of triple junctions and

proposed the following (Equation 10),

( f J2+ f J1) P = a b (10) (1− J3)

where fa and fb (considered 1 and 0.5 respectively) are geometrical factors to account for unfavorably€ oriented sensitized boundaries at the triple junction. These three probabilistic models provide a basis to evaluate microstructures that are indicative of the resistance to SCC.

The probability for crack arrest was observed to increase from 0.3 to about 0.5 after SGBE according to the model proposed by Marrow et al. and from 0.3 to 0.6 according to the model proposed by Palumbo et al. Similarly, PR was observed to increase from 0.03 to 0.18 as per the model proposed by Kumar et al. The decrease in SCC susceptibility can therefore be attributed to increase in the probability of crack arrest and percolation ratio calculated using the three models.

It has been suggested that as the ratio of J2/(1-J3) is equal to or exceeds 0.35, percolative paths in the microstructure will be broken [16]. Tsurekawa et al. also reported higher resistance to percolative intergranular corrosion in SS304L for microstructures with higher PR [47]. They also

found that triple junction distribution fraction was a more important parameter to rank than

maximum random boundary cluster length or GBCD. In this study, surface GBE increased the

ratio (PR) by five times that in the baseline microstructure and this reduced the SCC susceptibility significantly. This further demonstrates that surface GBE increased the resistance to SCC in Alloy 600 in tetrathionate solutions. Since only Σ3, Σ9 and Σ27 CSL boundaries have

been considered for triple junction analysis, this type of analysis is a more conservative approach

to evaluate the relative resistance of microstructure to SCC.

Figure 13. Probability for crack arrest as proposed by Palumbo et al. [46] and Marrow et al. [45] and percolation ratio by Kumar et al. [16], for all conditions. Only Σ3, 9 and 27 CSL boundaries have been considered in the triple junction analysis. 5. Conclusions

The following conclusions can be drawn from this study,

1. The fraction of special boundaries was increased in the 200 – 300 µm layer underneath

the surface after iterative steps of ultrasonic peening (inducing high strain in the near

surface region) and high temperature annealing.

2. The increase in fraction of low CSL boundaries, J1, J2 and J3 type of triple junctions and

corresponding decrease in J0 type was similar to that observed in single or multi-step

conventional GBE processing schemes.

3. Surface GBE decreased the degree of sensitization significantly (an order of magnitude)

and a SSRT results show an increase in elongation to failure and decrease in SCC

susceptibility.

4. Detailed analysis of cracks on the gage section after SSR tests show cracks being

deflected or mitigated at J2 type to triple junctions. Thus increase in fraction of special

grain boundaries may have a dual effect of decreasing Cr depletion and mitigating

intergranular cracks.

5. Probability for crack arrest and ratio (PR) calculated using percolative models was shown

to have increased the probability for crack arrest after SGBE.

Acknowledgements

The authors are grateful for financial support of this research by the Nuclear Energy

Frontier Commission, which provided funding in support of ‘‘Ohio Center for Laser Shock

Processing for Advanced Materials and Devices’’ and the equipment in the Center that was used in this work. MK was supported by the U.S. Department of Energy (DOE), Office of Basic

Energy Sciences, Division of Materials Science and Engineering under FWP# SCW0939. This work was partly performed under the auspices of the U.S. Department of Energy by Lawrence

Livermore National Laboratory under Contract DE-AC52-07NA27344. Any opinions, findings, conclusions, or recommendations expressed in these documents are those of the author(s) and do not necessarily reflect the views of the DOE and the State of Ohio, Department of Development.

References

[1] B. Alexandreanu, B.H. Sencer, V. Thaveeprungsriporn, G.S. Was, Acta Mater. 51 (2003) 3831–3848.

[2] E.M. Lehockey, G. Palumbo, P. Lin, A.M. Brennenstuhl, 36 (1997) 1211–1218.

[3] S. Bechtle, M. Kumar, B.P. Somerday, M.E. Launey, R.O. Ritchie, Acta Mater. 57 (2009) 4148–4157.

[4] Y. Gao, M. Kumar, R.K. Nalla, R.O. Ritchie, Metall. Mater. Trans. A 36 (2005) 3325– 3333.

[5] S. Kobayashi, M. Nakamura, S. Tsurekawa, T. Watanabe, J. Mater. Sci. 46 (2011) 4254– 4260.

[6] M. Shimada, H. Kokawa, Z.. Wang, Y.. Sato, I. Karibe, Acta Mater. 50 (2002) 2331– 2341.

[7] S.K. B., B.S. Prasad, V. Kain, J. Reddy, Corros. Sci. 70 (2013) 55–61.

[8] L. Tan, X. Ren, K. Sridharan, T.R. Allen, Corros. Sci. 50 (2008) 3056–3062.

[9] C. Cheung, U. Erb, G. Palumbo, Mater. Sci. Eng. A 185 (1994) 39–43.

[10] B. Alexandreanu, B. Capell, G.S. Was, Mater. Sci. Eng. A 300 (2001) 94–104.

[11] D.C. Crawford, G.S. Was, 23 (1992).

[12] G. Palumbo, K.T. Aust, E.M. Lehockey, U. Erb, P. Lin, Scr. Mater. 38 (1998) 1685–1690.

[13] Y. Pan, B.L. Adams, T. Olson, N. Panayotou, Acta Mater. 44 (1996) 4685–4695.

[14] A. King, G. Johnson, D. Engelberg, W. Ludwig, J. Marrow, Science (80-. ). 321 (2008) 382–386.

[15] V. Randle, M. Coleman, Acta Mater. 57 (2009) 3410–3421.

[16] M. Kumar, W.E. King, A.J. Schwartz, Acta Mater. 48 (2000) 2081–2091.

[17] Q. Li, B.M. Guyot, N.L. Richards, Mater. Sci. Eng. A 458 (2007) 58–66.

[18] H. Yun Bi, H. Kokawa, Z. Jie Wang, M. Shimada, Y.S. Sato, Scr. Mater. 49 (2003) 219– 223.

[19] R. Jones, V. Randle, Mater. Sci. Eng. A 527 (2010) 4275–4280.

[20] G.S. Was, V.B. Rajan, Corrosion 43 (1987) 576–579.

[21] R. Bandy, R. Roberge, R. Newman, Corros. Sci. 23 (1983) 995–1006.

[22] R. Newman, K. Sieradzki, H. Isaacs, Metall. Trans. A 13 (1982) 2015–2026.

[23] S.M. Bruemmer, G.S. Was, J. Nucl. Mater. 216 (1994) 348–363.

[24] M. Tsai, W. Tsai, J. Lee, Corros. Sci. 34 (1993) 741–757.

[25] A. Telang, A.S. Gill, S. Teysseyre, S.R. Mannava, D. Qian, V.K. Vasudevan, Corros. Sci. 90 (2015) 434–444.

[26] C. Ye, A. Telang, A.S. Gill, S. Suslov, Y. Idell, K. Zweiacker, J.M.K. Wiezorek, Z. Zhou, D. Qian, S. Ramaiah Mannava, V.K. Vasudevan, Mater. Sci. Eng. A 613 (2014) 274–288.

[27] C.-M. Suh, G.-H. Song, M.-S. Suh, Y.-S. Pyoun, Mater. Sci. Eng. A 443 (2007) 101–106.

[28] D.G. Brandon, Acta Metall. 14 (1966) 1479–1484.

[29] P.M. Ahmedabadi, V. Kain, B.K. Dangi, I. Samajdar, Corros. Sci. 66 (2013) 242–255.

[30] ASTM E112-13, Standard Test Methods for Determining Average Grain Size, ASTM International, West Conshohocken, PA, 2013.

[31] A. Gill, A. Telang, S.R. Mannava, D. Qian, Y.-S. Pyoun, H. Soyama, V.K. Vasudevan, Mater. Sci. Eng. A (2013).

[32] V. Randle, M. Coleman, M. Waterton, Metall. Mater. Trans. A 42 (2010) 582–586.

[33] V. Randle, G. Owen, Acta Mater. 54 (2006) 1777–1783.

[34] C.A. Schuh, M. Kumar, W.E. King, Acta Mater. 51 (2003) 687–700.

[35] J. Kai, C. Tsai, T. Huang, M. Liu, Metall. Trans. A 20 (1989) 1077–1088.

[36] S. Abe, M. Kojima, Y. Hosoi, in Stress Corrosion Cracking Susceptibility Index, ISCC, of Austenitic Stainless Steels in Constant Strain-Rate Test (STP 665), STP 665 Stress Corrosion Cracking—The Slow Strain-Rate Technique (ASTM International):, 1979, pp. 294–304.

[37] P. Fortier, W. a. Miller, K.T. Aust, Acta Mater. 45 (1997) 3459–3467.

[38] V.Y. Gertsman, S.M. Bruemmer, Acta Mater. 49 (2001) 1589–1598.

[39] M. Kumar, A.J. Schwartz, W.E. King, Acta Mater. 50 (2002) 2599–2612.

[40] O.M. Alyousif, D.L. Engelberg, T.J. Marrow, J. Mater. Sci. 43 (2007) 1270–1277.

[41] E.. Trillo, L.E. Murr, J. Mater. Sci. 33 (1998) 1263–1271.

[42] S. Rahimi, D.L. Engelberg, J. a Duff, T.J. Marrow, J. Microsc. 233 (2009) 423–31.

[43] T.J. Marrow, L. Babout, A.P. Jivkov, P. Wood, D. Engelberg, N. Stevens, P.J. Withers, R.C. Newman, J. Nucl. Mater. 352 (2006) 62–74.

[44] C. A. Schuh, M. Kumar, W.E. King, Acta Mater. 51 (2003) 687–700.

[45] T.J. Marrow, L. Babout, A. P. Jivkov, P. Wood, D. Engelberg, N. Stevens, P.J. Withers, R.C. Newman, J. Nucl. Mater. 352 (2006) 62–74.

[46] G. Palumbo, P.J. King, K.T. Aust, U. Erb, P.C. Lichtenberger, Scr. Metall. Mater. 25 (1991) 1775–1780.

[47] S. Tsurekawa, S. Nakamichi, T. Watanabe, Acta Mater. 54 (2006) 3617–3626.

6.0 Iterative Thermomechanical Processing of Alloy 600 for Improved Resistance to Corrosion and Stress Corrosion Cracking

Abhishek Telang1, Amrinder S. Gill2, Mukul Kumar3, Sebastien Teysseyre4, Dong Qian5, S.R. Mannava1, Vijay K. Vasudevan1 1Department of Mechanical and Materials Engineering, University of Cincinnati, Cincinnati, OH, USA 45221-0072 2AK Steel, Research Center, 705 Curtis Street, Middletown, OH, USA 3Lawrence Livermore National Laboratory, Livermore, CA, USA 4Idaho National Laboratory, Idaho Falls, ID, USA 5The University of Texas at Dallas, Richardson, TX, USA

Published in Acta Materialia, Volume 113, pp. 180-193 (2016); http://dx.doi.org/10.1016/j.actamat.2016.05.009

Abstract The effects of thermomechanical processing (TMP) with iterative cycles of 10% cold work and strain annealing, on corrosion and stress corrosion cracking (SCC) behavior of alloy 600 was studied. The associated microstructural and cracking mechanisms were elucidated using transmission (TEM) and scanning electron microscopy (SEM), coupled with precession electron diffraction (PED) and electron back scatter diffraction (EBSD) mapping. TMP resulted in increased fraction of low coincident site lattice (CSL) grain boundaries whilst decreasing the connectivity of random high angle grain boundaries. This disrupted random grain boundary network and large fraction of low CSL boundaries reduced the propensity to sensitization, i.e. carbide precipitation and Cr depletion. After TMP, alloy 600 (GBE) also showed higher intergranular corrosion resistance. Slow strain rate tests in 0.01 M Na2S4O6 solution at room temperature show TMP lowered susceptibility to intergranular SCC. To better understand the improvements in corrosion and SCC resistance, orientation maps of regions around cracks were used to analyze the interactions between cracks and various types of grain boundaries and triple junctions (TJs). Detailed analysis showed that cracks were arrested at J1 (1-CSL) and J2 (2-CSL) type of TJs. The probability for crack arrest at special boundaries and TJs, calculated using percolative models, was found to have increased after TMP, which also explains the increase in resistance to corrosion and SCC in GBE alloy 600. A clear correlation and mechanistic understanding relating grain boundary character, sensitization, carbide precipitation and susceptibility to corrosion and stress corrosion cracking was established.

1. Introduction Grain boundary engineering (GBE) has been demonstrated as a viable method for improving the resistance to corrosion [1–3] and stress corrosion cracking (SCC) [4–8] in austenitic stainless steels (SS) and Ni based alloys (alloy 600, alloy 690). GBE involves increasing the frequency of coincident site lattice (CSL) grain boundaries whilst disrupting the random grain boundary network through thermo-mechanical processing (TMP) routes. These routes involve cold rolling or uniaxial tension/compression followed by annealing and have been used to increase the frequency of CSL boundaries [1,3,9–11]. One approach involves pre- straining the material followed by annealing at comparatively lower temperature for a long time

(24-72 hours). Another multi-cycle approach uses two or more iterations of cold work (6-30%) followed by relatively high temperature annealing for short times (10-30 minutes) and has also been shown to increase the low CSL boundary fraction [11,12]. In addition, the multi-cycle approach results in numerous disruptions in the random high angle grain boundary network.

Further, a significant improvement in the corrosion and stress corrosion cracking has been attributed to these changes in the microstructure [13,14].

Alloy 600 and austenitic stainless steels have been known to be susceptible to stress corrosion cracking (SCC) in polythionic acid environments [15–18]. Susceptibility to SCC at low temperature in tetrathionate and thiosulfate environments has been attributed to Cr depletion in the area surrounding the grain boundary. A reduction in Cr depletion by disrupting the random grain boundary network or increasing the fraction of special boundaries should decrease the susceptibility to SCC [19].

Twin boundaries are more resistant to carbide precipitation and corrosion because the atomic structure is highly coherent as compared with high angle grain boundaries. In particular,

Σ3 and Σ9 boundaries in GBE SS304 have been observed to be more resistant to sensitization,

93 whereas Σ27 and other CSL boundaries were not really “special” in terms of their resistance to sensitization and thus intergranular stress corrosion cracking (IGSCC). Thus, it has been suggested that increased fraction of Σ3 and Σ9 boundaries would likely improve the corrosion and SCC resistance [20–23].

The purpose of this study was to evaluate the susceptibility of GB of known character to corrosion and their resistance to SCC. In this study, we use multi-step TMP to increase the fraction of special boundaries. Electron back-scattered diffraction (EBSD) is used to characterize changes in the microstructure after TMP. Transmission Electron microscopy (TEM) and

Precession Electron Diffraction (PED) were also used to characterize the microstructure and precipitation of carbides. Double loop Electrochemical Potentiokinetic Reactivation (DLEPR) and intergranular corrosion tests were used to evaluate the corrosion resistance. We also performed slow strain rate tests to evaluate the susceptibility to intergranular SCC after TMP.

EBSD mapping of regions around crack tips after DLEPR tests, corrosion (ASTM G28) tests and interrupted SSRT were used to study the interaction of cracks with various types of boundaries and triple junctions to explain the improvements in grain boundary engineered Alloy 600.

2. Materials and Methods

2.1 Thermo-mechanical processing and characterization

Alloy 600 plate (2 mm thickness) with chemical composition as shown in Table 1 was given an annealing treatment at 1050˚C for 10 minutes and then water quenched (WQ). For

TMP, these plates were cold rolled 10% and subsequently annealed in a furnace for 10 minutes at 900 ˚C or 1000 ˚C, then water quenched. This cycle was repeated 3 times to obtain 2 heats of grain boundary engineered Alloy 600, i.e GBE1 and GBE2. Processing details for TMP are listed in Table 2.

Table 5. Chemical composition of the Inconel Alloy 600 used in this study.

C Mn Si S Cr Fe Co Cd Ti Cu P Al Ni 0.001 0.001 0.08 0.16 0.18 14.99 8.05 0.18 0.01 0.18 0.1 0.08 Bal. max. max.

Table 6. Designation and corresponding details of processing used in this study.

Designation Detail

SA As received + Annealing at 1050˚C, 10min, WQ

GBE1 SA + 3 cycles of (10% cold work + 1000˚C,10min,WQ)

GBE2 SA + 3 cycles of (10% cold work + 900˚C,10min,WQ)

For EBSD, each coupon was electropolished in 12.5:87.5 sulfuric acid to methanol solution at 24V, 15s and finally polished with 0.05µm colloidal silica suspension. EBSD orientation mapping was performed in a FEI XL-30 scanning electron microscope (SEM) with step size of 0.5-1.5 µm at 30kV. OIM scans were analysed with the TSL OIM Analysis 7 package to calculate grain boundary character distribution, grain size, types of triple junctions.

CSL grain boundaries were categorized according to Brandon criterion of Δθ ≤ 15˚Σ-1/2 [24].

Boundaries with 3 < Σ < 29 were considered to be CSL boundaries whereas boundaries with Σ >

29 were considered random high angle boundaries (HABs) and Σ =1 as low angle boundaries

(LABs) [25]. Triple junctions (TJs) with no CSL boundaries, 1 CSL boundary, 2 CSL and 3 CSL boundaries have been classified as J0, J1, J2 and J3 respectively. For TJ analysis, only Σ3, Σ9 and Σ27 were considered as CSL boundaries.

TEM was used to characterize the precipitation behavior of carbides on grain boundaries of known character. Thin foils were prepared by conventional method including grinding, dimpling and finally ion-milling. Foils were obtained from SA and GBE1 samples after sensitization (650˚C, 2h) and analyzed with a Phillips CM-20 TEM operating at 200 kV.

Nanoscale lateral spatial resolution OIM analysis was performed by automated acquisition and indexing of precession electron diffraction (PED) patterns with a JEM 2100F TEM equipped with the DIGISTAR/ASTAR system from NanoMEGAS at the University of Pittsburgh.

Precessed illumination, 0.6° precession angle, and electron beam focused to ~3 nm in diameter at the TEM specimen section surface was scanned across a pre-selected area of interest with step- size of 20 nm to obtain maps of PED patterns, which were indexed automatically by optimized matching to computer generated reciprocal lattice based templates of the Nickel and Cr7C3

(carbide) phases of interest here. The PED based TEM OIM data sets provide information akin to that available via EBSD based OIM in the SEM but with lateral resolution in areal maps being on the order of 2 nm, limited essentially by the electron beam diameter used in the TEM instrument. The raw data sets of the PED based orientation indexed areal maps were processed and analyzed further with the TSL OIM Data Analysis software.

2.3 Double loop electrochemical potentiokinetic reactivation (DLEPR) and corrosion tests

All samples (SA, GBE1 and GBE2) were given a sensitization treatment in a laboratory furnace at 650˚C, 2h (water quenched) to induce precipitation of carbides. These samples were mechanically ground to 1200 grit, wet polished with 1 µm diamond suspension and finished with

0.05 µm colloidal silica suspension. DLEPR tests were performed in a solution composed of 0.01

M H2SO4 + 20 ppm KSCN using a Gamry Potentiostat (Reference 600) [26,27]. The scan rate was set at 0.5 mV/s for activation and reactivation loop and the sample size was 1 cm2. Freshly prepared solution was de-aerated with high purity Argon gas before and during each test and all tests were performed at room temperature. Samples were kept immersed in the test solution for 1 hour at open circuit potential before the start of each test. The following procedure was used to quantify sensitization in the annealed and GBE material after sensitization [28].

The degree of sensitization is reported as DL-EPR value (designated as R in %) which is

the ratio of the current density in reactivation loop to that in the activation loop times 100.

�= ��/��×100 (1)

The DL-EPR value obtained is normalized with various parameters like grain boundary area

(GBA), grain size, mean lineal intercept length (MIL). The DL-EPR value of a given alloy

condition (with ASTM grain size number of G’) is normalized with the grain size (with ASTM

grain size number of G) of the as-received material (SA) and is given by:

Rʹ = R x√2G ʹ -G (2)

The DL-EPR values were also normalized with mean intercept length (designated by l):

€ RMIL = R /Sv (4)

G = − 3.2877+6.6439log10 NL (5) where NL is a number of intercepts per unit length and G is the ASTM grain size number. The € mean lineal intercept (l) and the GBA, Sv, are given by [30]:

l =1/NL Sv = 2NL (6)

To evaluate susceptibility to intergranular corrosion after GBE, all samples were given a sensitization€ treatment at 650 ºC, 2h and then tested as per ASTM G28 standard in boiling ferric sulfate (25%)- sulfuric acid (50%) solution for 24 hours [30]. All samples were given a sensitization treatment at 650˚C, 2h, WQ before testing. Samples were mechanically ground to

1200 grit finish and weighed before and after testing. On completion of tests, samples were

97 sectioned and polished for SEM and EBSD analysis. For a few tests, samples were polished (For

EBSD analysis) before testing. This allowed for orientation mapping of the sample after the test.

2.4 Slow strain rate tests (SSRT)

Flat tensile samples of alloy 600 (all three conditions), with dimensions as shown in

Figure 1, were fabricated using a wire EDM. Subsequently, these samples were polished to 1200 grit, degreased with acetone and ethanol, dried and immersed in the test solution for 1 hour prior to straining. A strain rate of 2 x 10-6 /s was chosen for all tests. Slow strain rate tests (SSRTs) were performed with a system driven by a servo motor and fitted with a custom built environmental chamber. Load and displacement values were recorded periodically and samples were tested to failure. For interrupted SSRTs, samples were prepared using the same procedure that was used for EBSD sample preparation mentioned previously. SSRTs were interrupted at

12% and 26% nominal strain and the samples were examined using SEM/EBSD and orientation maps were recorded from areas on the gage section of the sample. Test solutions were prepared using reagent grade sodium tetrathionate (Na2S4O6) and distilled water.

Figure 1. Schematic of the sample used for slow strain rate tests and constant load tests. All dimensions in mm. 3. Results

3.6 Microstructure and sensitization

EBSD generated grain boundary maps from the SA condition (initial microstructure) are shown in Figure 2 (a, b). The microstructure obtained as a result of repetitive strain annealing cycles, designated GBE1 and GBE2 are shown in Figures 2 (c, d) and 2 (e, f) respectively. All orientation maps were obtained from the surface and are representative of the microstructure after GBE. Grain boundary fractions (length), grain size and TJ fractions for all conditions has been summarized in Figures 3 (a-c). The HAB fraction (length) decreased from 50% to ~ 20% after TMP while the CSL fraction increased from 37% to ~ 70%. The fraction of Σ3 boundaries increased from 30% in the SA condition to 55-60% for the GBE1, GBE2 conditions. More importantly, fractions of Σ9 + Σ27 increased from 2% (SA) to 10.2% (GBE1 and GBE2). In addition, TJ fractions extracted from the orientation maps show a sharp drop in fraction of J0 type from 43% to 6-10%. J3 fractions increased from 3% (SA) to 38% for GBE1 and 35% for

GBE2. These changes in TJ fractions are consistent with the GBE model proposed by Kumar et al. [11,31]. Grain size after GBE increased modestly from ~11 µm to ~17 µm for GBE1 and 14

µm for GBE2, respectively, after grain boundary engineering.

a b

c d

e f

Figure 2. EBSD orientation maps showing the microstructure for (a, b) SA (c, d) GBE1, (e, f) GBE2 conditions. Black lines denote random high angle grain boundaries and grey lines denote CSL boundaries (Σ ≤ 27).

100

Figure 3. (a) Grain boundary character distribution and grain size, (b) CSL boundary fraction and (c) triple junction fraction for each processing condition.

TEM micrographs from SA and GBE1 samples (after sensitization) are shown in Figures

4 (a, b) and (c, d) respectively. In the SA condition, carbides were observed to have precipitated

along grain boundaries. In addition, PED was used to resolve the nature of grain boundaries and

the type of carbides. In case of SA sample, the carbides of type Cr7C3 were present along grain boundaries and these were random HABs as shown in Figure 5(c). These were confirmed by electron diffraction to be Cr7C3 type of carbides (Figure 5d) that precipitate in alloy 600 when

101

sensitized in the 600-800 ˚C temperature range. In case of GBE1 condition, no carbides were

observed along grain boundaries but intra-granular carbides as shown in Figure 6.

Trillo and Murr observed that carbides (M23C6) precipitate first on HABs then at non-

coherent twins in stainless steels, whereas no carbides were observed on coherent twin

boundaries [32]. Zhou et al. also reported a higher resistance to carbide precipitation in special

GB (Σ≤29) as compared with random GB [33]. In the SA condition, carbides GBE1 condition,

shown in Figure 5, intra-granular carbides were observed. No carbides were observed along

other random high angle GBs. The spot pattern #2 in Figure 6 confirms that the carbide is of

Cr7C3 type. A number of Σ3 twins show no carbide precipitation.

Figure 4. TEM micrographs from SA (a, b) and GBE1 (c, d) conditions after sensitization.

102

Figure 5. (a) PED results from SA condition after sensitization including (a) IPF; (b) IQ overlaid with IPF; (c) IQ map showing carbides; (d) index map and corresponding spot patterns.

Figure 6. TEM micrograph and corresponding OIM map showing grain boundaries and carbides from (a) SA and (b) GBE1 condition respectively. GBs colored black are random high angle boundaries while those colored red are CSL (Σ3, 9, 27) boundaries. Spot patterns recorded from scans and simulated patterns used to index the grains, carbides are shown.

103

The area observed in the TEM was limited due to grain size as well as lower random

HAB fraction, which made the quantitative, statistical analysis of the precipitation behavior on grain boundaries of known character difficult. Hence EBSD-based orientation maps were recorded after DL-EPR tests to provide both a larger area for analysis and better statistics of the precipitation behavior at various types of boundaries in the SA and GBE conditions.

3.2 Double loop electrochemical potentiokinetic reactivation (DLEPR) and corrosion weight loss tests After a sensitization treatment of 650˚C, 2h (water quenched) and DLEPR tests were used to quantify the degree of sensitization (DOS) for each condition. The DOS value (R) for SA was 0.56%, but much lower for GBE1 and GBE2 conditions, being 0.06 and 0.12%, respectively. As the CSL fraction in the microstructure increased from 36% to ~ 70%, the susceptibility to sensitization was lowered as seen by the DOS or R value in Table 3. SEM micrographs recorded after DLEPR test from the samples are shown in Figure 7 (a-c). The corresponding normalized DOS values (Table 3) i.e. R’, RGBA, and RMIL all follow similar trends.

It should be noted that twin boundaries were excluded from grain size analysis used for normalizing these values.

Table 7. CSL fraction and corresponding degree of sensitization values for each processing condition.

Sample CSL fraction,% DOS (R), % R’ RGBA RMIL SA 37.7 0.56 0.4195 0.0070 0.0224 GBE1 72.9 0.06 0.0157 0.0021 0.0008 GBE2 65.3 0.12 0.0404 0.0033 0.0022

The SEM micrograph and its corresponding skeletonized grain boundary map in Figure

7(d,e) recorded after DLEPR test from the SA sample shows the presence of localized attack on random HABs (colored black). No attack was observed on low CSL boundaries (Σ3 colored red and Σ9, Σ27 colored blue). The precipitation of carbides during sensitization results in Cr

104

depletion in Alloy 600 and the DLEPR test gives good correlation with SCC susceptibility in

tetrathionate and thiosulfate solutions [27,34]. The DLEPR test parameters used in this study

have been optimized to attack the deeper Cr depletion zones as compared with shallower broad

zones, which are more immune [27]. Cr7C3 type of carbides precipitated along HABs during the sensitization treatment (shown in Figure 4 and 5) resulted in localized Cr depletion zones that were preferentially attacked during the DLEPR test. In both the GBE1 and GBE2 conditions, no localized attack was observed on HABs (colored black) or CSL boundaries (Σ3 colored red and

Σ9, Σ27 colored blue). A representative set of SEM and orientation maps from the GBE2 condition sample are shown in Figure 6 (f, g). These results indicate that the discontinuous random HAB network in the GBE conditions reduced the propensity for Cr carbide precipitation and Cr depletion as compared with the SA condition. The decrease in sensitization has been well correlated with CSL fractions in other instances in Alloy 600 as well as austenitic stainless steels in literature[1,2,20,33,35,36].

105

e) d)

f) g)

Figure 7. SEM images showing microstructure after DLEPR test for (a) SA, (b) GBE1, (c) GBE2 conditions. SEM micrographs and corresponding GB map showing locations of localized attack (circles) for SA (d, e) and GBE2 (f, g) conditions. Random HABs are in black while Σ3 and Σ9, Σ27 boundaries are colored blue respectively.

Corrosion weight loss tests were also performed to evaluate the effect of increased CSL fractions and disrupted grain boundary network on the corrosion behavior. Baseline and GBE samples (after sensitization) were immersed in boiling Fe2(SO4)3 -50% H2SO4 solution for 24 hours as per ASTM G28 standard [30]. SEM micrographs from the surface of samples after the test are shown in Figure 8 (a-f). Weight loss after tests and corrosion rates calculated after testing are summarized in Table 4. Weight loss after 24 hours decreased from 5% (SA) to 0.67% while corrosion rates decreased from 16 mpy (SA) to ~ 2 mpy after GBE. The attack was noticeably

106

lower on the GBE1 and GBE2 samples as compared with the SA condition. Grain dropping was

clearly observed in the SA condition while GBE1 and GBE2 showed discontinuous attack along

grain boundaries.

Table 4. Results of ASTM G28 weight loss tests for all conditions.

Sample Weight Loss, % Corrosion rate, mpy SA 4.989 0.411 GBE1 0.656 0.035 GBE2 0.664 0.041

Like DLEPR tests, ASTM G28 corrosion weight loss tests are commonly used as quick tests

to evaluate Cr depletion susceptibility to intergranular cracking. Figure 9 shows the SEM

micrograph and corresponding grain boundary map for GBE1 condition after corrosion test

(G28) for 24 hours. J0 type junctions consisting of all random HABs were attacked in all

instances. J3 types of junctions were the most resistant with minimal or no intergranular attack.

In J1 and J2 types, the CSL segments showed no attack in a majority of cases. This is consistent

with other observations of termination of intergranular attack at low CSL boundaries during G28

tests [37].

3.3 Slow strain rate tests

The increase in the CSL fractions and lower DOS are indicative of less Cr depletion after

TMP and may reduce the SCC susceptbility of alloy 600 in thiosulfate and tetrathionate

solutions. Hence, SSR tests were performed to observe and quantify the susceptibility to IGSCC

after GBE. Stress-strain curves obtained from SSR tests in tetrathionate (0.01 M Na2S4O6) solution are as shown in Figure 10. Tests were performed at the same strain rate and in the same environment for diect comparison of SCC behavior. In case of SA, the elongation to failure in

107

0.01 M Na2S4O6 solution was only 15% for SA condition while those for GBE1 and GBE2 condition were about 66% and 35% respectively.

Analysis of the fracture surface after tests showed ductile and mixed mode of failure for

GBE1 and GBE2 conditions, respectively while SA condition showed 100% intergranular cracking. The significant difference in elongation to failure as well as failure analysis indicates

that TMP improved resistance to SCC in alloy 600. Secondary cracks from the gage sections of

samples were analyzed in greater detail and the results are discussed in the next section.

Figure 8. Micrographs after G28 test for (a, b) SA, (c, d) GBE1 and (e, f) GBE2 conditions.

108

Figure 9. (a) SEM micrograph and (b) corresponding EBSD map recorded from GBE1 sample after ASTM G28 test for 24 hours. Circles show locations of J0, J1, J2 and J3 type of triple junctions.

Figure 10. Stress-strain curves for (SA) and grain boundary engineered (GBE1, GBE2) -6 samples (after sensitization at 650˚C, 2h) in 0.01 M Na2S4O6 solution. Strain rate: 2 x10 /s.

3.1 Crack Analysis Some SSRT tests were interrupted to analyze intergranular cracks along grain boundaries

of known character. Cracks were observed on the gage section of the GBE2 sample and

109 orientation, grain boundary maps from the gage section after 11% strain are shown in Figure

11(a-e). Conventional inverse pole figure (IPF) and grain boundary map with low CSL and

HABs show locations of crack arrest in Figure 11(a) and 11(b) respectively. Cracks were observed to have been mitigated or arrested at a J1 or J2 type of TJs and a few locations have been circled (black). In addition, orientation and grain boundary maps, in Figure 11(c) and 11(d) respectively, have been colored according based on the technique proposed by Patala and Schuh

[38]. This method uses quaternions and provides information about both the misorientation angle and the misorientation axis. For example, Σ3 boundaries (60º (111), colored blue) and Σ9 (39º

(110), colored green) are part of J1, J2 type of junctions that were crack arrest locations. This analysis indicates that disruptions in the cracks along along HABs, encounter greater resistance to propagation at J1 and J2 type of junctions. More statistical analysis that examines the susceptiblity of certain axis/angle orientations to intergranular corrosion would elucidate if certain other boundaries exhibit relatively higher resistance.

Figure 12 (a, b) show the kernal average misorientation (KAM) map and grain orientation spread map from the same area area of a GBE2 sample interrupted after 11% strain during

SSRT. KAM values were relativley higher around the crack tips (Figure 12(a)) as compared with the grain interior indicating higher local strain as the crack encounters a resistant TJ. The crack arrest locations in Figure 12b suggest that as the crack front encounters a resistant TJ, grains adjoining the crack paths are strained resulting in higher misorientations (lattice rotation). In addition, the low Taylor factor grains (Figure 12(c)) correspond well with high lattice rotation in the strained grains adjoining the crack arrest locations. Furthermore, the Schmid factor (Figure

12(d)) for grains adjoining the crack front are very high, which indicates high critically resolved shear stress on the slip planes. Therefore, we can infer that the grains adjoining and ahead of the

110 crack front undergo severe plastic deformation. Similarly, Figure 13 (a, b) shows intergranular cracks on the gage section after 26% strain. As in the case of the 11% strained sample, numerous instances of the crack arrest at TJs can be seen.

Figure 11. (a) Inverse pole figure (IPF) image and (b) corresponding grain boundary map showing intergranular cracks on the gage section of GBE2 sample interrupted after 11% strain during SSRT. (c) Misorientation map, (d) grain boundary map in quarternion color

111 from the same area and (e) Fore scattered detector (FSD) image from the same area. (f) color legend for homophase misorientations, built using stereographic projections of surfaces of constant misorientation angle used to interpret the (c) and (d) plots. Locations showing crack deflection or mitigation are circled in black and the loading direction is horizontal.

Figure 12. (a) Kernal average misorientation map, (b) grain orientation spread map, (c) Taylor factor map and (d) Schmid factor map from the same region of GBE2 sample interrupted after 11% strain during SSRT. Locations showing crack deflection or mitigation are circled in black. Loading direction is horizontal.

112

Figure 13. (a) FSD image and corresponding GB map (b) showing intergranular cracks on the gage section of GBE2 sample interrupted after 26% strain during SSRT.

4. Discussion

The results of this comprehensive study have led to a number of new findings on the behavior of grain boundaries and on their sensitization, carbide precipitation, and hence corrosion and SCC behavior of alloy 600 and related mechanistic aspects, which are discussed below.

The present results confirm a decrease in DOS after GBE has been attributed to increased fraction of twin boundaries and its variants [1,2,20,28]. Additionally, CSL boundaries have been shown to be more resistant to precipitation of carbides and the resulting Cr depletion. TEM micrographs in Figures 4 and 5 show Cr carbides precipitated along random high angle boundaries. These Cr7C3 type carbides precipitated during the sensitization treatment. EBSD on

SA sample surface after DLEPR tests showed local attack on random HABs. In contrast, CSL

boundaries particularly Σ3, 9 and 27 did not show any attack. For the GBE2 sample, no attack

was observed along either HABs or CSL boundaries. This is consistent with the absence of Cr

carbides along GBs in the GBE1 sample during TEM analysis. Bi et al. [19] reported

precipitation of M23C6 type of carbides in austenitic SS304 along random HABs while CSL

113 segments were precipitation free [19]. In addition to reduced Cr carbide precipitation, Cr depletion was suppressed. Zhou et al. [33] reported that more than 90% of special GBs (Σ≤ 29) were resistant to carbide precipitation. Kurban et al. [36] also observed higher resistance to carbide precipitation in grain boundary engineered SS304. The decrease in DOS by almost an order of magnitude suggests that the higher CSL fraction as well as numerous disruptions in

HAB network reduced Cr depletion. In the GBE1 condition, no instances of localized attack were observed along random HAB or CSL boundaries, indicating that the Cr depletion was suppressed as seen from micrographs recorded after DLEPR tests (Figure 6).

Figure 14. Probability for crack arrest as proposed by Kumar et al. [11], Palumbo et al. [39] and Marrow et al. [40] for all conditions.

The probability for crack arrest at a TJ was calculated according to the three models proposed by Kumar et al. [41], Palumbo et al. [39] and Marrow et al. [40] and have been plotted in Figure 14.

Palumbo et al. [39] also proposed the following relation (Equation 7) to calculate the probability (P) of crack arrest,

114

P = f 2 +2 f f 1− f (7) sp [ 0 sp ()sp ]

where fsp is the fraction of special boundaries and f0 is the fraction of interfaces in the

distribution€ that are unfavorably oriented to the stress axis.

Assuming J3 type of junctions to be non-entities in crack propagation, Kumar et al. [11]

proposed (Equation 8) that the percolative paths in the microstructure would be broken if,

J2 P = ≥ 0.35 (8) R (1− J3)

Marrow et al. [40] observed cracks arrested at both J2 and J1 type of TJs and proposed the following€ (Equation 9),

( f J2+ f J1) P = a b (9) (1− J3) where fa and fb (considered 1 and 0.5 respectively) are geometrical factors to account for

unfavorably€ oriented sensitized boundaries at the triple junction. These probabilistic models

provide a basis to evaluate microstructures that are indicative of the resistance to SCC.

The probability for crack arrest was observed to increase from 0.3 to about 0.5 after GBE

according to the model proposed by Marrow et al. and from 0.3 to 0.6 according to the model

proposed by Palumbo et al. Similarly, PR was observed to increase from 0.03 to 0.18 as per the model proposed by Kumar et al. Only Σ3, 9 and 27 CSL boundaries have been considered for TJ analysis. This allows for a more conservative approach to evaluate the relative SCC resistance.

These models provide a basis for comparing relative resistance of microstructures to percolation of intergranular cracks. Tsurekawa et al. also reported higher resistance to percolative intergranular corrosion in SS304L for microstructures with higher PR [14]. They also found that

triple junction distribution fraction was a more important parameter to rank than maximum

random boundary cluster length or GBCD. In this study, GBE increased the ratio (PR) by five

115 times that in the baseline microstructure and this reduced the SCC susceptibility significantly.

The increase in probability of crack arrest (P) after GBE explains the increase in resistance to

SCC observed experimentally. Cracks were more likely to encounter a resistant junction, and the reduced sensitization of random grain boundaries due to altered Cr diffusion behavior contributed to reduced SCC susceptibility.

A number of crack tips were used for statistical analysis of crack arrest at TJs of specific kind. Figure 15 shows observed fractions of crack arrest at a TJ, probability of finding a crack tip/TJ of a specific type, and fraction of types of TJs in the microstructure. About 52 crack arrest locations were identified in the GBE2 sample and considered for this analysis. The probability of finding a TJ of a specific type was calculated as per the procedure used by Seita et al. [42]. The assumption for this calculation is that for a given TJ, each GB (including low CSL such as Σ3,

Σ9, and Σ27) is likely to be a cracked boundary. Subsequently, each of the three possible permutations of cracked vs. non-cracked GBs for each TJ type are considered to determine the fraction of permutations that result in each type of crack tip. These fractions are then summed as shown in Equation 10(a-c) below to determine the total probability of finding each type of crack tip/TJ.

1 P0Σ = P(J1) + P(J0) 10(a) 3

2 2 P1Σ = P(J2) + P(J1) 10(b) 3 3 € 1 P2Σ = P(J3) + P(J2) 10(b) 3 € Here, P(Ji) is the fraction of TJs with i low-Σ GBs in the microstructure [42]. €

116

Figure 15. (a) Probability of crack arrest at a triple junction of specific type, fraction of TJ junction type that were observed to be crack arrest locations and occurrence of TJ types in the bulk microstructure is summarized. (b) Inclination angle with respect to tensile axis for cracked boundaries.

Although no CSL (Σ3, Σ9, Σ27) boundaries were observed to have cracked in this study, Σ9 and 27 are likely to crack at higher strain and in a more aggressive environment [22]. It is clear that disruption of HAB connectivity deflects and/or arrests cracks at some J0, but almost all J1

117 and J2 types. About 51% of crack arrest locations were J1 type of TJs while 32% were J2 type and the rest (15%) were J0 type (Figure 15). The probability for crack arrest at J0, J1 and J2 type of TJ is 0.22, 0.37 and 0.42, respectively, and shows good correlation with the values of crack arrest locations determined from the EBSD observations and analysis of interactions of cracks with the different types of TJs. the observed values. In addition, a majority of cracked boundaries

(all of them were random HABs) were observed to have inclination angle close to 70-90˚ with respect to the tensile axis. This is consistent with observations of intergranular cracks along boundaries inclined at angles close to 90˚ along the tensile axis [42–44]. It is evident that random HABs may be most susceptible to localized intergranular attack, crack initiation and cracks tend to propagate along the random HAB network till they encounter a J1 or J2 type of junction.

Serial sectioning combined with 3D EBSD is currently being applied to further investigate the interaction between cracks and the 3D grain boundary network and further develop and extend the percolative models for quantification and prediction of crack arrest in 3D.

5. Conclusions

Iterative cold rolling and annealing was used to modify the GBCD in Alloy 600 and its effect on carbide precipitation, Cr depletion behavior, intergranular corrosion and stress corrosion cracking was investigated in this study. The following conclusions can be drawn from this study,

6. The fraction of low CSL boundaries was increased after iterative steps of cold rolling and

high temperature annealing and the random HAB network was disrupted.

7. TMP of alloy 600 decreased the degree of sensitization significantly (by an order of

magnitude), carbide precipitation and chromium depletion at grain boundaries and, in

118

turn, the corrosion rates were also significantly lower as compared with those in the SA

condition.

8. SSRT results show an increase in elongation to failure and significant decrease in SCC

susceptibility following GBE. Detailed analysis of cracks with the microstructural

elements (different types of grain boundaries and TJs) on the gage section of the samples

after interrupted SSRT show cracks being deflected or arrested at J1 and J2 type of TJs.

9. The probability for crack arrest at special GBs as well as at a particular type of TJ (J1,

J2), calculated using percolative models, was found to have increased after GBE, which

also explains the increase in resistance to corrosion and SCC in GBE alloy 600.

Acknowledgements

The authors are grateful for financial support of this research by the Nuclear Energy University

Program (NEUP) of the US Department of Energy contract #102835 issued under prime contract DE-

AC07-05ID14517 to Battelle Energy Alliance, LLC. This work was partly performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-

AC52-07NA27344. The efforts of MK were supported by the U.S. Department of Energy (DOE), Office of Basic Energy Sciences, Division of Materials Science and Engineering under FWP# SCW0939. We also gratefully acknowledge the contribution of the State of Ohio, Department of Development and Third

Frontier Commission, which provided funding in support of ‘‘Ohio Center for Laser Shock Processing for

Advanced Materials and Devices’’ equipment in the Center that was used in this work. The authors would also like to thank Kai Zweiacker and Dr. Jorg M. K. Wiezorek at the University of Pittsburgh for help with the TEM/PED characterization. Any opinions, findings, conclusions, or recommendations expressed in these documents are those of the author(s) and do not necessarily reflect the views of the DOE or the

State of Ohio, Department of Development.

119

References

[1] M. Shimada, H. Kokawa, Z.. Wang, Y.. Sato, I. Karibe, Optimization of grain boundary character distribution for intergranular corrosion resistant 304 stainless steel by twin- induced grain boundary engineering, Acta Mater. 50 (2002) 2331–2341. doi:10.1016/S1359-6454(02)00064-2. [2] B. Sunil Kumar, B.S. Prasad, V. Kain, J. Reddy, Methods for making alloy 600 resistant to sensitization and intergranular corrosion, Corros. Sci. 70 (2013) 55–61. doi:10.1016/j.corsci.2012.12.021. [3] T. Watanabe, Grain boundary engineering: Historical perspective and future prospects, J. Mater. Sci. 46 (2011) 4095–4115. doi:10.1007/s10853-011-5393-z. [4] C. Cheung, U. Erb, G. Palumbo, Application of grain boundary engineering concepts to alleviate intergranular cracking in Alloys 600 and 690, Mater. Sci. Eng. A. 185 (1994) 39– 43. doi:10.1016/0921-5093(94)90925-3. [5] B. Alexandreanu, B. Capell, G.S. Was, Combined effect of special grain boundaries and grain boundary carbides on IGSCC of Ni–16Cr–9Fe–xC alloys, Mater. Sci. Eng. A. 300 (2001) 94–104. doi:10.1016/S0921-5093(00)01705-6. [6] D.C. Crawford, G.S. Was, The Role of Grain Boundary Misorientation in Intergranular Cracking of Ni-16Cr-9Fe in 360 C Argon and High-Purity Water, Scr. Mater. 23 (1992). [7] E.M. Lehockey, G. Palumbo, P. Lin, A.M. Brennenstuhl, On the Relationship between Grain Boundary Character Distribution and Intergranular Corrosion, Scr. Mater. 36 (1997) 1211–1218. [8] G. Palumbo, K.T. Aust, E.M. Lehockey, U. Erb, P. Lin, On a More Restrictive Geometric Criterion for “Special” CSL Grain Boundaries, Scr. Mater. 38 (1998) 1685–1690. doi:10.1016/S1359-6462(98)00077-3. [9] V. Randle, M. Coleman, A study of low-strain and medium-strain grain boundary engineering, Acta Mater. 57 (2009) 3410–3421. doi:10.1016/j.actamat.2009.04.002. [10] V. Randle, Grain boundary engineering: an overview after 25 years, Mater. Sci. Technol. 26 (2010) 253–261. doi:10.1179/026708309X12601952777747. [11] M. Kumar, W.E. King, A.J. Schwartz, Modifications to the microstructural topology in f.c.c. materials through thermomechanical processing, Acta Mater. 48 (2000) 2081–2091. doi:10.1016/S1359-6454(00)00045-8. [12] Q. Li, B.M. Guyot, N.L. Richards, Effect of processing parameters on grain boundary modifications to alloy Inconel 718, Mater. Sci. Eng. A. 458 (2007) 58–66. doi:10.1016/j.msea.2006.12.129. [13] V. Randle, R. Jones, Grain boundary plane distributions and single-step versus multiple- step grain boundary engineering, Mater. Sci. Eng. A. 524 (2009) 134–142. doi:10.1016/j.msea.2009.06.018. [14] S. Tsurekawa, S. Nakamichi, T. Watanabe, Correlation of grain boundary connectivity with grain boundary character distribution in austenitic stainless steel, Acta Mater. 54 (2006) 3617–3626. doi:10.1016/j.actamat.2006.03.048.

120

[15] G.S. Was, V.B. Rajan, Technical Note: On the Relationship between the EPR Test, Sensitization, and IGSCC Susceptibility, Corrosion. 43 (1987) 576–579. doi:10.5006/1.3583904. [16] R. Bandy, R. Roberge, R. Newman, Low temperature stress corrosion cracking of Inconel 600 under two different conditions of sensitization, Corros. Sci. 23 (1983) 995–1006. http://www.sciencedirect.com/science/article/pii/0010938X83900264 (accessed May 27, 2013). [17] R. Newman, K. Sieradzki, H. Isaacs, Stress-corrosion cracking of sensitized type 304 stainless steel in thiosulfate solutions, Metall. Trans. A. 13 (1982) 2015–2026. http://link.springer.com/article/10.1007/BF02645947 (accessed April 24, 2013). [18] S.M. Bruemmer, G.S. Was, Microstructural and microchemical mechanisms controlling intergranular stress corrosion cracking in light-water-reactor systems, J. Nucl. Mater. 216 (1994) 348–363. doi:10.1016/0022-3115(94)90020-5. [19] H. Yun Bi, H. Kokawa, Z. Jie Wang, M. Shimada, Y.S. Sato, Suppression of chromium depletion by grain boundary structural change during twin-induced grain boundary engineering of 304 stainless steel, Scr. Mater. 49 (2003) 219–223. doi:10.1016/S1359- 6462(03)00261-6. [20] R. Jones, V. Randle, Sensitisation behaviour of grain boundary engineered austenitic stainless steel, Mater. Sci. Eng. A. 527 (2010) 4275–4280. doi:10.1016/j.msea.2010.03.058. [21] E.A. West, G.S. Was, IGSCC of grain boundary engineered 316L and 690 in supercritical water, J. Nucl. Mater. 392 (2009) 264–271. doi:10.1016/j.jnucmat.2009.03.008. [22] V.Y. Gertsman, S.M. Bruemmer, Study of grain boundary character along intergranular stress corrosion crack paths in austenitic alloys, Acta Mater. 49 (2001) 1589–1598. doi:10.1016/S1359-6454(01)00064-7. [23] A. King, G. Johnson, D. Engelberg, W. Ludwig, J. Marrow, Observations of Intergranular Stress Corrosion Cracking in a Grain-Mapped Polycrystal, Science (80-. ). 321 (2008) 382–386. [24] D.G. Brandon, The strucutre of high angle grain boundaries, Acta Metall. 14 (1966) 1479– 1484. [25] B. Alexandreanu, B.H. Sencer, V. Thaveeprungsriporn, G.S. Was, The effect of grain boundary character distribution on the high temperature deformation behavior of Ni– 16Cr–9Fe alloys, Acta Mater. 51 (2003) 3831–3848. doi:10.1016/S1359-6454(03)00207- 6. [26] Y. Lim, H. Kim, J. Han, J. Kim, H. Kwon, Influence of laser surface melting on the susceptibility to intergranular corrosion of sensitized Alloy 600, Corros. Sci. 43 (2001) 1321–1335. http://www.sciencedirect.com/science/article/pii/S0010938X00001499 (accessed June 5, 2013). [27] M. Ahn, H. Kwon, J. Lee, Predicting Susceptibility of Alloy 600 to Intergranular Stress Corrosion Cracking Using a Modified Electrochemical Potentiokinetic Reactivation Test, Corros. Sci. 51 (1995) 441–449. [28] P.M. Ahmedabadi, V. Kain, B.K. Dangi, I. Samajdar, Role of grain boundary nature and

121

residual strain in controlling sensitisation of type 304 stainless steel, Corros. Sci. 66 (2013) 242–255. doi:10.1016/j.corsci.2012.09.026. [29] ASTM E112-13, Standard Test Methods for Determining Average Grain Size, ASTM International, West Conshohocken, PA, 2013. [30] ASTM G28-02, Standard Test Methods for Detecting Susceptibility to Intergranular Corrosion in Wrought, Nickel-Rich, Chromium-Bearing Alloys, ASTM International, West Conshohocken, PA, 2008. [31] C.A. Schuh, M. Kumar, W.E. King, Analysis of grain boundary networks and their evolution during grain boundary engineering, Acta Mater. 51 (2003) 687–700. doi:10.1016/S1359-6454(02)00447-0. [32] E.. Trillo, L.E. Murr, A TEM investigation of M23C6 carbide precipitation behaviour on varying grain boundary misorientations in 304 stainless steels, J. Mater. Sci. 33 (1998) 1263–1271. [33] Y. Zhou, K.. Aust, U. Erb, G. Palumbo, Effects of grain boundary structure on carbide precipitation in 304L stainless steel, Scr. Mater. 45 (2001) 49–54. doi:10.1016/S1359- 6462(01)00990-3. [34] A. Telang, A.S. Gill, S. Teysseyre, S.R. Mannava, D. Qian, V.K. Vasudevan, Effects of Laser Shock Peening on SCC Behavior of Alloy 600 in Tetrathionate Solution, Corros. Sci. 90 (2015) 434–444. doi:10.1016/j.corsci.2014.10.045. [35] H. Kokawa, M. Shimada, Y.S. Sato, Grain-boundary structure and precipitation in sensitized austenitic stainless steel, Jom. 52 (2000) 34–37. doi:10.1007/s11837-000-0159- 0. [36] M. Kurban, U. Erb, K.T. Aust, A grain boundary characterization study of boron segregation and carbide precipitation in alloy 304 austenitic stainless steel, Scr. Mater. 54 (2006) 1053–1058. doi:10.1016/j.scriptamat.2005.11.055. [37] P. Lin, G. Palumbo, U. Erb, K.T. Aust, Influence of grain boundary character distribution on sensitization and intergranular corrosion of alloy 600, Scr. Metall. Mater. 33 (1995) 1387–1392. doi:10.1016/0956-716X(95)00420-Z. [38] S. Patala, C.A. Schuh, A continuous and one-to-one coloring scheme for misorientations, Acta Mater. 59 (2011) 554–562. doi:10.1016/j.actamat.2010.09.058. [39] G. Palumbo, P.J. King, K.T. Aust, U. Erb, P.C. Lichtenberger, Grain boundary design and control for intergranular stress-corrosion resistance, Scr. Metall. Mater. 25 (1991) 1775– 1780. doi:10.1016/0956-716X(91)90303-I. [40] T.J. Marrow, L. Babout, a. P. Jivkov, P. Wood, D. Engelberg, N. Stevens, et al., Three dimensional observations and modelling of intergranular stress corrosion cracking in austenitic stainless steel, J. Nucl. Mater. 352 (2006) 62–74. doi:10.1016/j.jnucmat.2006.02.042. [41] M. Kumar, A.J. Schwartz, W.E. King, Microstructural evolution during grain boundary engineering of low to medium stacking fault energy fcc materials, Acta Mater. 50 (2002) 2599–2612. doi:10.1016/S1359-6454(02)00090-3. [42] M. Seita, J.P. Hanson, S. Gradečak, M.J. Demkowicz, The dual role of coherent twin

122

boundaries in hydrogen embrittlement, Nat. Commun. 6 (2015) 6164. doi:10.1038/ncomms7164. [43] E.A. West, G.S. Was, Strain incompatibilities and their role in intergranular cracking of irradiated 316L stainless steel, J. Nucl. Mater. 441 (2013) 623–632. doi:10.1016/j.jnucmat.2012.10.021. [44] M.D. McMurtrey, G.S. Was, L. Patrick, D. Farkas, Relationship between localized strain and irradiation assisted stress corrosion cracking in an austenitic alloy, Mater. Sci. Eng. A. 528 (2011) 3730–3740. doi:10.1016/j.msea.2011.01.073.

123

7.0 Effect of Thermo-Mechanical Processing on Sensitization and Corrosion in Alloy 600 Studied by SEM- and TEM-Based Diffraction and Orientation Imaging Techniques

Abhishek Telang1, Amrinder S. Gill2, Kai Zweiacker3, Can Liu3, Jorg M. K. Wiezorek3, S. R.Mannava,1 D. Qian,4 Vijay K. Vasudevan1* 1Department of Mechanical and Materials Engineering, University of Cincinnati, Cincinnati, OH, USA 45221-0072 2AK Steel, Research Center, 705 Curtis Street, Middletown, OH, USA 45044 3Department of Mechanical and Materials Engineering, University of Pittsburgh, Pittsburgh, PA, USA 15261 4Department of Mechanical Engineering, University of Texas at Dallas, Richardson, TX (Submitted to Journal of Nuclear Materials in December 2016)

Abstract

In this study, we investigate the effect of thermo-mechanical processing (TMP) on the evolution of low coincident site lattice (CSL) grain boundaries (3 ≤ Σ ≤ 27) and thence on the precipitation behavior of carbides at various types of grain boundaries in Alloy 600. Detailed analysis of the microstructure using Electron Back-scattered Diffraction (EBSD) in the scanning electron microscope (SEM) and Precession Electron Diffraction (PED) and Energy Dispersive

Spectroscopy (EDS) in the transmission electron microscope (TEM) has been used to study the effects of TMP on the precipitation of Cr carbides and Cr depletion. After TMP, the fraction of low-CSL grain boundaries is increased appreciably and the precipitation behavior of carbides is modified resulting in lower sensitization. The results showed that Σ3-type coincident site lattice

(CSL) boundaries were more resistant to carbide precipitation as compared with random high angle boundaries. Furthermore, the increased fraction of low-sigma CSL boundaries and a disrupted random high angle grain boundary network resulted in lower Cr depletion and lower sensitization along grain boundaries. A clear correlation and mechanistic understanding relating grain boundary character distribution, carbide precipitation and susceptibility to sensitization was established.

124

1. Introduction

Thermo-mechanical processing (TMP) has been shown to increase the fraction of low-

coincident site lattice (CSL) boundaries (3 ≤ Σ ≤ 27) in Alloy 600 and austenitic stainless steels

[1–3]. TMP usually involves single or multi-step cycles of strain followed by annealing to

“engineer” the microstructure. The main objective of the TMP is to decrease the connectivity of

random high angle boundaries (HAB) whilst increasing the fraction of the CSL boundaries.

Various methods of deformation including tension [4,5], compression [6,7], cold rolling [2,6,8,9]

and even surface treatments[10,11] followed by annealing have been shown to increase the

fraction of low-CSL boundaries particularly Σ3n (n ≤ 3) in materials with low stacking fault

energy. Furthermore, a discontinuous HAB network and higher fraction of low-CSL boundaries

has been shown to improve resistance to percolation of intergranular attack in austenitic stainless

steels and Ni alloys [2,7–9,12–14].

The precipitation of M7C3 or M23C6 type of carbides and the resulting Cr-depletion in the area adjoining the grain boundary results in localized attack in an aggressive environment [15–

17]. Cr-rich carbide precipitation and the resulting sensitization has been shown to be related to the grain boundary character and energy [18–22]. Trillo and Murr reported preferential precipitation of M23C6 type of carbides on the HAB, which exhibit high grain boundary energy, while coherent twin boundaries with comparatively much lower grain boundary energy were immune to precipitation [18,19]. Bi et al. have also reported lower Cr depletion across low CSL boundaries as compared with HAB [23]. The introduction of new low energy segments including incoherent and coherent Σ3 boundaries and disruption in random HAB by iterative TMP has been shown to improve the overall resistance to sensitization in austenitic stainless steels

[12,24,25]. The objective of this study is to develop a comprehensive understanding of carbide

125 precipitation and Cr depletion across grain boundaries of known character in Alloy 600. Further, we also investigate the effect of iterative TMP using sequential cold rolling and annealing cycles on carbide precipitation after sensitization. Electron back-scattered diffraction (EBSD) was used to characterize the microstructure before and after TMP. Further, Precession Electron Diffraction

(PED) and Energy Dispersive Spectroscopy (EDS) in transmission electron microscope (TEM) was used to characterize the grain boundary character, carbide precipitation and Cr depletion.

Electrochemical potentiokinetic reactivation (EPR) tests were used to investigate the effect of

TMP on the Cr depletion and localized attack along grain boundaries of known character was investigated in detail. The results are discussed with a focus on establishing mechanistic understanding of correlation relating grain boundary character, sensitization, carbide precipitation and susceptibility to corrosion.

2. Material and Methods

2.1 Thermo-mechanical processing

Alloy 600 plate (2 mm thickness) with chemical composition as shown in Table 1 was annealed at 1050˚C for 10 minutes and then water quenched (WQ). For grain boundary engineering, these plates were cold rolled 10% and subsequently annealed in a lab furnace for 10 minutes at 1000 ˚C, then water quenched. This cycle was repeated 3 times to obtain grain boundary engineered Alloy 600, designated as TMP, and has been shown to increase low-CSL boundary fraction in our previous study [26]. Processing details for solution annealed (SA) and

TMP conditions are listed in Table 2.

Table 8. Chemical composition in weight percent (wt.%) of the Inconel Alloy 600 used in this study.

C Mn Si S Cr Fe Co Cd Ti Cu P Al Ni 0.001 0.001 0.08 0.16 0.18 14.99 8.05 0.18 0.01 0.18 0.1 0.08 Bal. max. max.

126

Table 9. Designation and corresponding details of processing used in this study.

Designation Detail

SA As received + Annealing at 1050˚C, 10min, WQ

TMP SA + 3 cycles of (10% cold work + 1000˚C,10min,WQ)

2.2 Microstructural characterization

For EBSD, each coupon was electropolished in 12.5:87.5 sulfuric acid to methanol

solution at 24V, 15 s and finally polished with 0.05 micron colloidal silica suspension. EBSD

orientation mapping was performed in a FEI XL-30 scanning electron microscope (SEM) with

step size of 1-2 µm at 30kV. OIM scans were analysed with the TSL OIM Analysis 7 package to

calculate grain boundary character distribution (GBCD), grain size and triple junction analysis.

Boundaries with (3 ≤ Σ ≤ 27) were classified as CSL boundaries were categorized according to

Brandon criterion of Δθ ≤ 15˚Σ-1/2 [27]. All non CSL boundaries (3 ≤ Σ ≤ 27) with misorientation angle θ > 15º were classified as HABs. Triple junctions (TJs) with no CSL boundaries, 1 CSL boundary, 2 CSL and 3 CSL boundaries have been classified as J0, J1, J2 and J3 respectively.

For TJ analysis, only Σ3n, (n ≤ 3) CSL boundaries were considered.

Transmission electron microscopy was used to characterize the precipitation behavior on

grain boundaries of known characters. Thin foils were prepared by conventional method

including grinding, dimpling and finally ion-milling. Foils were obtained from SA and TMP

samples after sensitization (650˚C, 2h) and analysed with a Phillips/FEI CM-20 TEM operating

at 200 kV. Nanoscale lateral spatial resolution OIM has been performed by automated

acquisition and orientation indexing of precession electron diffraction (PED) patterns with field-

emission gun equipped (FEG) TEM instruments operated at 200kV, namely, a JEM 2100F TEM

and a FEI Tecnai F20 G2 UT, using the DIGISTAR/ASTAR and Topspin systems from

127

NanoMEGAS, respectively. Precessed illumination, 0.6° precession angle, with electron beam

focused to 2-3 nm in diameter at the TEM specimen section surface was scanned across a pre-

selected area of interest with step-size of 5-20 nm to obtain maps of PED patterns, which were

indexed automatically by optimized matching to computer generated reciprocal lattice based

templates of the austenite (Nickel) and Cr7C3 (carbide) phases of interest here. The orientation

data sets provide information akin to that available via EBSD in the SEM but with lateral

resolution in areal maps being on the order of the step-size of a few nanometers. Ultimately, the

spatial resolution of PED acquired orientation data sets is limited essentially by the electron

beam diameter used in the TEM instrument, i.e. ~2 nm for a precessed beam in a FEGTEM. The

raw data sets of the orientation data sets were processed and analyzed further with the TSL OIM

Data Analysis software.

2.3 Double loop electrochemical potentiokinetic reactivation (DLEPR) tests

SA and TMP samples were given a sensitization treatment in a laboratory furnace at

650˚C for 2h followed by water quenching to induce precipitation of carbides. These samples

were mechanically ground to 1200 grit, wet polished with 1 µm diamond suspension and

finished with 0.05 µm colloidal silica suspension. DL-EPR tests were performed in a solution

composed of 0.01 M H2SO4 + 20 ppm KSCN using a Gamry Potentiostat (Reference 600)

[17,28]. The scan rate was set at 0.5 mV/s for activation and reactivation loop and the sample

size was 1 cm2. Freshly prepared solution was de-aerated with high purity Ar gas before and

during each test and all tests were performed at room temperature. Samples were kept immersed

in the test solution for 1 hour at open circuit potential before the start of each test. The degree of

sensitization (R) was calculated as the ratio of the current density in the reactivation loop (Ir) to that in the activation loop (Ia) times 100 (Equation 1).

128

�= ��/��×100 (1)

After DL-EPR tests, EBSD data was collected from specific locations from the sample surface

exposed to the DLEPR test solution. Fiducial marks (indents) were used to identify the locations

for orientation mapping and corresponding optical/SEM micrographs.

3. Results and Discussion

In this section, we first present the effect of TMP on the microstructure of Alloy 600.

This is followed by characterization of the Cr carbides (after sensitization) using TEM. Further,

we present and discuss the effect of TMP on the corrosion behavior of Alloy 600 using DLEPR

tests. EBSD orientation maps and corresponding SEM/Optical micrographs are used to correlate

the sensitization behavior of grain boundaries of known characters.

3.1 Microstructure

In this study, we used a multi-cycle TMP route to modify the GBCD and the connectivity

of HABs (colored black) in Alloy 600. EBSD orientation maps obtained before and after TMP

are shown in Figure 1 (a, b) and (c, d) respectively. The TMP introduces low-sigma CSL

boundaries (colored grey), mainly Σ3, Σ9 and Σ27 in the microstructure. Moreover, the

connectivity of HABs has been broken up considerably after TMP (Figure 1d) as compared with

the SA microstructure (Figure 1b). A brief summary of GBCD, Σ3n (n = 1-3) fraction and TJ fraction for SA and TMP conditions are shown in Table 3. TJs have been classified into 4 types, i.e J0 (0-CSL, 3-HAB), J1 (1-CSL, 2-HAB), J2 (2-CSL, 1-HAB) and J3 (3-CSL). The significant decrease in the J0 TJ fraction and corresponding increase in J2 and J3 fractions indicates the breakdown in the HAB network.

129

Figure 1. Inverse Pole Figure (IPF) maps and corresponding skeletonized grain boundary maps for SA (a, b) and TMP (c, d) conditions.

Table 3. Summary of grain boundary character distribution, Σ length fractions and triple junction fractions for the SA and TMP conditions.

Sample CSL LAB HAB ∑3 ∑9 ∑27 J0 J1 J2 J3 SA 37.7 12.4 49.8 31.3 1.6 0.6 43.2 50.1 3.2 3.5 TMP 72.9 7.3 19.1 61.9 6.3 3.8 7.0 43.0 11.6 38.4

A new stereological technique for measuring the five-parameter grain boundary distribution (FPGBD), developed by Rohrer et al. [29,30], provides statistical data on the distribution of boundary planes in addition to misorientations. The 60º/[111] misorientation (Σ3)

130 features a very strong maximum at (111), associated with coherent annealing twins, as shown in

Figure 2. Also, the five parameter stereological analyses reveals that peak multiples of a random distribution (MRD) value at (111) plane of Σ3 orientation has increased to 1385 as a result of

TMP (Figure 2b) as compared to 885 for SA (Figure 2a). This indicates that Σ3 boundaries established during TMP predominantly terminated on (111) planes. In addition to coherent annealing twin boundaries, a large spread about the peak corresponds to incoherent segments of

Σ3 boundaries. These incoherent segments have higher mobility and lower energy than random

HABs and have replaced parts of the random boundary network.

Figure 2. Grain boundary plane distribution for Σ3, Σ9, Σ27a boundaries for (a, c, e) SA and (b, d, f) TMP conditions respectively. The (100), (110) and (111) poles are marked with (□), (O) and (∆) respectively.

131

The 39º/[110] and 32º/[110] misorientations are associated with Σ9 and Σ27a boundaries respectively. Another feature of this TMP method is that the number fraction of Σ9 and Σ27

boundaries is enhanced because interactions between two Σ3 (Shown in Table 3). This feature is

also seen in the stereograms for SA and TMP conditions in Figure 2 by comparing panels for Σ9

(Figure 2c, d) and Σ27a boundaries (Figure 2e-f). The three- to four-fold increase in these

sections after TMP is indicative of multiple twinning reactions due to the iterative strain and

annealing cycles. Multiple overlapping peaks are observed along the [110] zone as the

misorientations on [110] have a strong tendency to exist as asymmetrical tilt boundaries [30].

Similar to prior observations, the population of Σ9 grain boundaries peaks along the zone of pure

tilt grain boundaries. In the TMP condition, for 39º/[110] section (Σ9), diffuse maxima (MRD =

11) occur at the orientations of the several planes including (-117), (1-15), (1-12) and (1-11).

Further, the 32º/[110] section (Σ27a) shows discrete peaks (MRD =13) at several planes

including low energy symmetrical plane like (1-15) and other orientations like (-117), (2-23).

The MRD values (~3) for these planes are much lower in the SA condition. These observations

are similar to those reported in Ni by Randle et al. [18] and in alloy 617 by Katnagallu et al. [19]

showing no particular preference of Σ9 and Σ27 boundaries to terminate on low index planes.

3.2 Grain Boundary Character and Carbide Precipitation

To investigate the effect of TMP on the precipitation behavior of carbides, TEM

observations of samples from the TMP and SA conditions (after sensitization) were conducted.

Representative BF micrographs are shown in Figure 3 (a, b) and (c, d), respectively. In the TMP

condition carbides were observed at intra-granular locations and also inter-granularly along

HABs (e.g. as marked in Figure 3a and 3b), while many boundary segments appeared to be

132

without carbide precipitation. In contrast, in the SA condition a majority of the carbides where

observed along the majority of the grain boundaries (e.g. as marked in Figure 3c and 3d).

Figure 3. TEM micrographs from TMP (a, b) and SA (c, d) conditions.

Although the fields of view examined by TEM analysis are inherently quite limited, significant differences in the carbide precipitation behavior emerged and have been detected after

TMP of the SA samples. PED assisted automated crystal orientation mapping in TEM has been performed to determine whether carbides precipitated preferentially on specific types of boundaries. The virtual bright field image in Figure 4a has been constructed from the intensity of the transmitted beam in the map of PED patterns obtained from a 2µm by 2µm area scan using a

20nm step size and shows the carbides precipitated on grain boundaries in a SA sample. The

133

precipitates were determined to be of Cr7C3-type based on corresponding PED patterns obtained

from intergranular locations (e.g. location 3 marked in Figure 4a). Examples of PED spot

patterns obtained from locations marked as (1)-(5) in the virtual bright field image of Figure 4a

and the corresponding indexing by the software are also shown in Figure 4. Image quality (IQ)

map and a composite map of inverse pole figure (IPF) based orientations overlaid with an IQ

map are shown Figure 4(b,c). They reveal a few additional carbides decorating the grain

boundaries. The boundaries between the grains marked as (1) and (2), (1) and (4), (4) and (2),

and (4) and (5) were determined to be random HABs (black/dark lines shown in Figure 4d),

while a Σ3 boundary segment (red in Figure 4d) was also been identified. The relatively large

size of carbides and a fine step size for the PED scan (~10-20 nm) allows for reliable indexing of

carbides as well as the matrix as seen in Figure 4. Carbides can be seen along the other two

branches of the J0 type of triple junction where the HABs separating grains (1), (2) and (4) from

each other meet. The J0 type junctions represent potent nucleation sites from the standpoint of

classical heterogeneous precipitation mechanisms.

In case of the TMP condition, significantly fewer carbides were observed to have

precipitated along HABs. One such example of a carbide on a HAB and spot patterns from

locations 1-3 in the TMP sample is shown in bright field image in Figure 5a. A phase map of the

same area in Figure 5b shows the matrix (red) and the Cr7C3 type of carbide (yellow). Figure 5c shows that the carbide had precipitated on a HAB (colored black) and Figure 5d shows the IPF map of the same area. A few intra-granular Cr7C3 type of carbides were also observed in the area

observed in TEM, e.g. Figure 3 (a, b). The decrease in HAB fraction after TMP coupled with the

limited field of view accessible to TEM examination rendered statistically significant analyses of

134

carbide precipitation along grain boundaries of various types unfeasible. Nevertheless, the

limited results suggest that carbide precipitation on the SA samples occurs largely on the HABs.

Figure 4. PED results from SA condition after sensitization including (a) virtual bright field and (b) IQ map showing Cr7C3 type of carbides (c) IQ overlaid with IPF with color key for

Nickel and Cr7C3 (d) grain boundary map and corresponding spot patterns (1-5) for locations labeled in (a).

135

Figure 5. PED results from TMP condition after sensitization including (a) Index map and corresponding spot patterns (b) phase map with Nickel (red) and Cr7C3 (yellow); (c) IQ overlaid with IPF; (d) IPF map showing carbides.

STEM-EDS line profiles were recorded to evaluate the differences in Cr carbide

precipitation (and Cr depletion) across multiple grain boundaries. Prior to STEM-EDS,

orientation data from this area was analyzed to identify the grain boundary characters. In case of

the TMP condition, an area shown by BF image in Figure 6a was selected for analysis. IQ

overlaid with IPF maps from two locations (marked in white in Figure 6a) are presented in

Figure 6b, 6c and Figure 6d shows the character of grain boundaries from the location shown in

Figure 6c. Further, STEM-EDS profiles across grain boundaries labeled A-D in Figure 6e are

shown in Figure 6f. Cr concentration across HAB (labeled A) compared with Σ3 boundaries

(labeled B, C and D) shows no significant difference. The absence of Cr enrichment (due to

precipitation) and depletion (adjoining the precipitate) across HAB and Σ3 boundaries in the

TMP condition is different as compared to the SA condition as presented in Figure 7 and 8.

136

Figure 6. (a) Bright field image of an area in the TMP condition; (b) and (c) Image quality maps overlaid with Inverse Pole Figure map; (d) grain boundary map showing Σ3 and HABs from the location shown in (c). Location of STEM-EDS profiles (e) and (f) Cr concentration across grain boundaries.

137

Figure 7. (a) Bright field image of an area in the SA condition; (b) Image quality map from a location marked in red in (a) showing intergranular and intra-granular precipitates. (c) IPF map showing grain boundary character. (d, e) Location of STEM-EDS profiles (Line A and B) showing Cr, Ni and Fe concentration across grain boundaries.

138

Figure 8. (a) Bright field image of an area in the SA condition; (b) Location of STEM-EDS profile showing Cr concentration across a grain boundary with an intergranular precipitate. Note the Cr depletion in locations adjacent to the precipitate.

BF image from the SA condition shown in Figure 7a includes intergranular as well as

intra-granular precipitates and various grain boundaries. IQ and IPF maps for the location

(marked by a red box) are shown in Figures 7b and 7c respectively. STEM-EDS line profiles

across LAB (labeled A) and HAB (labelled B) are shown in Figures 7d and 7e respectively.

Additionally, Figure 8a shows a precipitate on a HAB and Cr concentration across the

boundary/precipitate is shown in Figure 8b. Cr depletion observed across the HABs (Figure 7d,

Figure 8b) as well as LAB (Figure 7e), while no Cr precipitation was observed along the Σ3

boundary.

From the SA sample, we can infer that Cr7C3 type of carbides precipitated along HABs but were not observed to precipitate on any low-sigma CSL boundary (including twin boundaries). Previously, Bi et al.[23] reported no carbide precipitation and suppression of Cr

139

depletion across Σ13b and coherent Σ3 boundaries, while random HABs showed significant Cr depletion. Trillo and Murr [18,19] have also observed that M23C6 type of carbides precipitated along HABs and even certain low-CSL boundaries, including incoherent twin boundaries, while coherent twin boundaries were resistant to carbide precipitation in SS304. They explained precipitation behavior of these Cr-rich carbides with the differences in grain boundary free energy. In this study, we use the same methodology as Trillo and Murr [19] to calculate the grain boundary free energies for Alloy 600 at 650 ˚C. The average grain boundary free energy (γgb)

2 and coherent twin boundary energy (γctb) for Alloy 600 was estimated to be 880 mJ/m and 19 mJ/m2, respectively, at 1060 ºC [32]. Assuming the temperature coefficients for SS304 and Alloy

600 are similar and that they remain constant to 650 ºC (sensitization temperature), we can then calculate the γgb, γctb coherent twin boundary energy at this temperature.

Table 4. Summary of calculations in Alloy 600 for average grain boundary free energy (γgb) and coherent (γctb) twin boundaries.

SS304 @ Alloy 600 @ Alloy 600 @ 1060 ºC 1060 ºC 650 ºC Average grain boundary free energy 835 [19] 880 [32] 1081 2 (γgb), mJ/m Average coherent twin boundary free 19 [19] 19 [32] 16 2 energy (γctb), mJ/m Temperature coefficient, -0.49 [19] -0.49* dγgb/dT Temperature coefficient, +0.007 [33] +0.007* dγctb/dT *Assumed to be same as that for SS304

A summary of these calculations is given in Table 4. The average grain boundary energy was 1081 mJ/m2 while the coherent twin boundaries have an interfacial free energy of 16 mJ/m2.

The Σ9 boundary and other low CSL boundaries have comparatively higher energies than the Σ3 boundaries in pure Nickel [34] and Alloy 600 [35]. It must be noted that C content, temperature and time strongly influence precipitation behavior [15,36,37]. Furthermore, the nucleation and

140 growth of Cr rich precipitates is dependent on the average grain boundary energy and grain size

[38]. While these grain boundary energy calculations can explain the differences in carbide precipitation behavior, a different method is needed to estimate the sensitization susceptibility of the overall microstructure. Following the approach of Wasnik et al.[38], the average or effective grain boundary energy (EGBE) was calculated as shown in Equation 2.

��=��(4�)�� (2)

In Equation 2, fi is the fraction of each class of grain boundary (Σ3-Σ27 and random), γi is the corresponding energy and d is the grain size. The γi for CSL boundaries (Σ3-Σ27) was calculated using Equation 3, where γmax is the energy of the random boundaries and γΣ is the energy of a

CSL boundary of the corresponding Σ notation.

��=(1− 1�)�� (3)

The EGBE for the SA and the TMP condition Alloy 600 have been determined as 0.246 and

0.126, respectively. The iterative TMP in Alloy 600 significantly reduced the EGBE of the engineered microstructure. To evaluate the effect of the TMP on the precipitation of Cr carbides and Cr depletion, orientation maps were recorded by EBSD in the SEM from samples after DL-

EPR tests. Precipitation of carbides along grain boundaries may lead to severe Cr depletion in the areas adjoining the grain boundary precipitates (e.g. Figures 7 and 8). With the DLEPR test, areas where Cr concentration is depleted below a certain threshold are preferentially attacked.

Hence, this method was used to observe many grain boundaries (of different character) for large fields of view providing statistically significant additional information that is complementary to the locally resolved TEM and STEM analyses regarding the susceptibility of microstructure of

Alloy 600 in the SA and TMP conditions to intergranular attack.

141

3.3 DL-EPR

After a 2h sensitization treatment at 650˚C followed by water quenching, DL-EPR tests were performed to quantify the degree of sensitization (DOS) of Alloy 600 for both SA and TMP conditions. The DOS value (R) for the SA condition was R=0.57%, which is an order of magnitude larger than for the TMP condition with R=0.06% and these values are similar to our previous work [26]. It is pertinent here to note that the J0 fraction decreased (from 43% to 7%), while the fraction of J2 (3% to 11%) and J3 (4% to 38%) increased significantly after TMP. As the CSL fraction of grain boundaries in the grain boundary networks of the Alloy 600 microstructure increased from 38% to ~ 73%, the random grain boundary network was disrupted and the susceptibility to sensitization was lowered as a result of TMP. The decrease in DOS for the TMP condition as compared with the SA condition is also consistent with the lower EGBE attained for the TMP condition (see section 3.2). The results obtained here for Alloy 600 are similar to observations of a decrease in DOS with increase in low-CSL boundary fraction in austenitic stainless steels [38,39]. In the SA condition, grain boundaries that show localized attack (circled in black) are shown in Figure 9a. Skeletonized grain boundary maps, shown in

Figure 9b, from the same area were used to identify the character of the grain boundaries that show localized attack. A number of grain boundaries that show localized attack were identified as random HAB (circled black). During the DL-EPR test, grain boundaries with deep but narrow

Cr depletion zones are attacked [17]. This results in preferential attack on grain boundaries with

Cr carbides (Cr7C3) that induced locally Cr depletion to sub-critical levels. Σ3-type CSL

boundaries, shown in red in Figure 9b and also color-coded in Figure 9c do not show any

evidence of localized attack. Further, one of the Σ9-type boundaries (circled blue), of which only

a few segments were recorded in the field of view, shows localized attack, while a large fraction

of the random HABs show localized attack (Figure 9a,b). The extent of localized attack on

142

random HABs (or other boundaries) will depend on the amount of carbon available in the solid

solution, sensitization temperature and time. Figure 9c and 9d (color key) is another

representation where the grain boundary segments are colored as per the method proposed by

Patala and Schuh [40]. This method uses quaternions and provides information about both the

misorientation angle and the misorientation axis. For example, Σ3 and Σ9 boundaries are colored blue and green respectively provide information about the grain boundary axis angle orientation with the contrast in colors representing misorientations distance. Black and blue arrow markers were used to indicate examples of HABs and Σ9 boundaries immune to intergranular attack. This variability in sensitization behavior within populations of Σ9 boundaries may be related to the deviation from the exact axis/angle misorientations from ideal condition indicated by coloring differences for each boundary in Figure 9c. Similarly, HABs with certain axis/angle misorientations may also show relatively higher resistance to sensitization.

Figure 9. (a) SEM micrograph from SA sample (sensitized) after DLEPR test. (b) Skeletonized grain boundary map; (c) grain boundary misorientations map and (d) color

143 scheme showing axis-angle misorientations. Locations of intergranular attack on HAB and Σ9 boundaries are circled black and blue respectively. Examples of HABs and Σ9 boundaries that were immune are indicated using black and blue arrows respectively.

Ortner and Randle reported partial sensitization of incoherent twin boundaries and Σ9 boundaries, while coherent twin boundaries remained completely un-sensitized [41].

Furthermore, variable sensitization resistance of HAB has been reported and attributed to the range of precipitation and diffusion rates along the different types of HABs [41]. Ebrahimi et al. also observed that Σ3 boundaries were more resistant to intergranular attack as compared with random HABs [42]. Yamada et al. reported that grain boundary engineered SS304 suppressed intergranular corrosion even after strain sensitization treatment due to the breakdown of the random grain boundary network [25]. Figures 9 shows that not all grain boundaries are sensitized uniformly to the same extent. This observation was also reported by Rahimi et al. [13] in sensitized thermo-mechanically processed SS304.

In case of the Alloy 600 in the TMP condition, localized attack was not observed on either the HABs or the low CSL boundaries as shown in Figure 10. Optical and high magnification

SEM micrographs from the same area analyzed after the DL-EPR test are shown in Figure 10a,b.

Skeletonized grain boundary maps from the area shown in Figure 10a are presented in Figure 10 c with Σ3n (n ≤ 3) and HAB highlighted for correlation. Figure 10d shows forward scatter detector image of the same area) overlaid with grain boundary misorientation map with the same color scheme as Figure 9d. The absence of localized intergranular attack on HABs and low CSL boundaries (Σ3n, n ≤ 3) indicates that Cr concentrations were consistently above the minimum threshold detected by the onset of localized attack in the DLEPR test method used in this study.

From the DLEPR tests and subsequent correlation with grain boundary maps, it was observed that Σ3 boundaries (coherent and incoherent) were resistant to intergranular attack in SA as well

144 as TMP conditions. However, other low-CSL boundaries, such as Σ9 and Σ27, were not completely resistant to intergranular attack in the SA condition but appear to be comparatively immune in the TMP condition. This reduction in susceptibility of sensitization of in the TMP condition can be related to the lower EGBE and the disruption of the random HAB network.

However, additional grain boundary plane information may be needed with higher statistical sampling frequency from different sensitization conditions to establish more detailed correlations between grain boundary character and their susceptibility to sensitization in Alloy 600.

Figure 10. (a) Optical and (b) SEM micrograph from TMP sample (sensitized) after DLEPR test. (c) Skeletonized grain boundary map and (d) grain boundary misorientation map overlaid with forward scatter detector map of the same area. Arrows indicate HABs immune to intergranular attack.

145

4. Conclusions

Iterative cold rolling and annealing was used to modify the GBCD in Alloy 600 and its effect

on carbide precipitation and intergranular corrosion was investigated in this study. A significant

increase (from 38% to 73%) in low-CSL boundary fraction was observed after TMP and TJ

characteristics indicated a breakdown/disruption of the random HAB network. FPGBD analysis

indicated that a majority of newly added Σ3 boundaries predominantly terminated on the (111), a

low energy configuration. TEM and PED were used to identify locally resolved the grain

boundary types and the intergranular and intra-granular Cr carbides that precipitated after sensitization. STEM-EDS plots across HAB and Σ3 boundaries in TMP condition (after

sensitization) show no Cr carbide precipitation/Cr depletion. In contrast, HABs in the SA

condition (after sensitization) show Cr carbide precipitation and Cr depletion in areas adjacent to

the precipitates. Orientation mapping and DLEPR tests were used in conjunction to observe the relative resistance of grain boundaries of different character to localized intergranular attack.

Additionally, the EGBE calculated using grain boundary characteristics explain the decrease in the susceptibility to sensitization after TMP. HABs and occasionally Σ9 boundaries in the SA condition show localized attack, while Σ3 boundaries were immune. However, intergranular attack was not observed on HABs or other low CSL boundaries in the TMP condition. Through experimental data and calculations of grain boundary free energy, low CSL boundaries

(coherent, incoherent twins) were found to be more resistant to sensitization and intergranular attack as compared with random HABs.

Acknowledgements

The authors are grateful for financial support of this research by the Nuclear Energy

University Program (NEUP) of the US Department of Energy contract #102835 issued under

146 prime contract DE-AC07-05ID14517 to Battelle Energy Alliance, LLC. We also gratefully acknowledge the contribution of the State of Ohio, Department of Development and Third

Frontier Commission, which provided funding in support of ‘‘Ohio Center for Laser Shock

Processing for Advanced Materials and Devices’’ equipment in the Center that was used in this work. Any opinions, findings, conclusions, or recommendations expressed in these documents are those of the author(s) and do not necessarily reflect the views of the DOE or the State of

Ohio, Department of Development. Use of the facilities of the Materials Micro-Characterization

Laboratory of the Department of Mechanical Engineering and Materials Science of the

University of Pittsburgh and assistance from Dr. S. Tan of the Department of Electrical and

Computer Engineering at the University of Pittsburgh with EDS mapping is gratefully acknowledged.

References

[1] M. Kumar, W.E. King, A.J. Schwartz, Modifications to the microstructural topology in f.c.c. materials through thermomechanical processing, Acta Mater. 48 (2000) 2081–2091. doi:10.1016/S1359-6454(00)00045-8. [2] M. Shimada, H. Kokawa, Z.. Wang, Y.. Sato, I. Karibe, Optimization of grain boundary character distribution for intergranular corrosion resistant 304 stainless steel by twin- induced grain boundary engineering, Acta Mater. 50 (2002) 2331–2341. doi:10.1016/S1359-6454(02)00064-2. [3] T. Watanabe, Grain boundary engineering: Historical perspective and future prospects, J. Mater. Sci. 46 (2011) 4095–4115. doi:10.1007/s10853-011-5393-z. [4] V. Randle, R. Jones, Grain boundary plane distributions and single-step versus multiple- step grain boundary engineering, Mater. Sci. Eng. A. 524 (2009) 134–142. doi:10.1016/j.msea.2009.06.018. [5] S. Rahimi, D.L. Engelberg, J.A. Duff, T.J. Marrow, In situ observation of intergranular crack nucleation in a grain boundary controlled austenitic stainless steel., J. Microsc. 233 (2009) 423–31. doi:10.1111/j.1365-2818.2009.03133.x. [6] A.J. Schwartz, W.E. King, M. Kumar, Influence of processing method on the network of grain boundaries, Scr. Mater. 54 (2006) 963–968. doi:10.1016/j.scriptamat.2005.11.052. [7] S. Tsurekawa, S. Nakamichi, T. Watanabe, Correlation of grain boundary connectivity

147

with grain boundary character distribution in austenitic stainless steel, Acta Mater. 54 (2006) 3617–3626. doi:10.1016/j.actamat.2006.03.048. [8] P.M. Ahmedabadi, V. Kain, B.K. Dangi, I. Samajdar, Role of grain boundary nature and residual strain in controlling sensitisation of type 304 stainless steel, Corros. Sci. 66 (2013) 242–255. doi:10.1016/j.corsci.2012.09.026. [9] B. Sunil Kumar, B.S. Prasad, V. Kain, J. Reddy, Methods for making alloy 600 resistant to sensitization and intergranular corrosion, Corros. Sci. 70 (2013) 55–61. doi:10.1016/j.corsci.2012.12.021. [10] O.M. Alyousif, D.L. Engelberg, T.J. Marrow, Surface grain boundary engineering of shot- peened type 304 stainless steel, J. Mater. Sci. 43 (2007) 1270–1277. doi:10.1007/s10853- 007-2252-z. [11] A. Telang, A.S. Gill, D. Tammana, X. Wen, M. Kumar, S. Teysseyre, et al., Surface grain boundary engineering of Alloy 600 for improved resistance to stress corrosion cracking, Mater. Sci. Eng. A. 648 (2015) 280–288. doi:10.1016/j.msea.2015.09.074. [12] R. Jones, V. Randle, Sensitisation behaviour of grain boundary engineered austenitic stainless steel, Mater. Sci. Eng. A. 527 (2010) 4275–4280. doi:10.1016/j.msea.2010.03.058. [13] S. Rahimi, D.L. Engelberg, T.J. Marrow, A new approach for DL-EPR testing of thermo- mechanically processed austenitic stainless steel, Corros. Sci. 53 (2011) 4213–4222. doi:10.1016/j.corsci.2011.08.033. [14] P. Lin, G. Palumbo, U. Erb, K.T. Aust, Influence of grain boundary character distribution on sensitization and intergranular corrosion of alloy 600, Scr. Metall. Mater. 33 (1995) 1387–1392. doi:10.1016/0956-716X(95)00420-Z. [15] J. Kai, C. Tsai, T. Huang, M. Liu, The effects of heat treatment on the sensitization and SCC behavior of INCONEL 600 alloy, Metall. Trans. A. 20 (1989) 1077–1088. http://link.springer.com/article/10.1007/BF02650143 (accessed May 27, 2013). [16] G.S. Was, V.B. Rajan, The Mechanism of Intergranular Cracking of Ni-Cr-Fe Alloys in Sodium Tetrathionate, Metall. Trans. A. 18 (1987) 1313–1323. [17] M. Ahn, H. Kwon, J. Lee, Predicting Susceptibility of Alloy 600 to Intergranular Stress Corrosion Cracking Using a Modified Electrochemical Potentiokinetic Reactivation Test, Corros. Sci. 51 (1995) 441–449. [18] E.A. Trillo, L.E. Murr, Effects of carbon content, deformation, and interfacial energetics on carbide precipitation and corrosion sensitization in 304 stainless steel, Acta. Mater. 47 (1998) 235–245. doi:10.1016/S1359-6454(98)00322-X. [19] E.. Trillo, L.E. Murr, A TEM investigation of M23C6 carbide precipitation behaviour on varying grain boundary misorientations in 304 stainless steels, J. Mater. Sci. 33 (1998) 1263–1271.

148

[20] H.U. Hong, B.S. Rho, S.W. Nam, Correlation of the M23C6 precipitation morphology with grain boundary characteristics in austenitic stainless steel, Mater. Sci. Eng. A. 318 (2001) 285–292. doi:10.1016/S0921-5093(01)01254-0. [21] H. Kokawa, M. Shimada, Y.S. Sato, Grain-boundary structure and precipitation in sensitized austenitic stainless steel, Jom. 52 (2000) 34–37. doi:10.1007/s11837-000-0159- 0. [22] Y.S. Lim, J.S. Kim, H.P. Kim, H.D. Cho, The effect of grain boundary misorientation on the intergranular M 23C6 carbide precipitation in thermally treated Alloy 690, J. Nucl. Mater. 335 (2004) 108–114. doi:10.1016/j.jnucmat.2004.07.038. [23] H. Yun Bi, H. Kokawa, Z. Jie Wang, M. Shimada, Y.S. Sato, Suppression of chromium depletion by grain boundary structural change during twin-induced grain boundary engineering of 304 stainless steel, Scr. Mater. 49 (2003) 219–223. doi:10.1016/S1359- 6462(03)00261-6. [24] K. Kurihara, H. Kokawa, S. Sato, Y.S. Sato, H.T. Fujii, M. Kawai, Grain boundary engineering of titanium-stabilized 321 austenitic stainless steel, J. Mater. Sci. 46 (2011) 4270–4275. doi:10.1007/s10853-010-5244-3. [25] G. Yamada, H. Kokawa, Y. Yasuda, Effect of post-GBE strain-sensitisation on corrosion resistance of grain boundary engineered 304 austenitic stainless steel, Philos. Mag. 93 (2013) 1443–1453. doi:10.1080/14786435.2012.762467. [26] A. Telang, A.S. Gill, M. Kumar, S. Teysseyre, D. Qian, S.R. Mannava, et al., Iterative thermomechanical processing of alloy 600 for improved resistance to corrosion and stress corrosion cracking, Acta Mater. 113 (2016) 180–193. doi:10.1016/j.actamat.2016.05.009. [27] D.G. Brandon, The strucutre of high angle grain boundaries, Acta Metall. 14 (1966) 1479– 1484. [28] Y. Lim, H. Kim, J. Han, J. Kim, H. Kwon, Influence of laser surface melting on the susceptibility to intergranular corrosion of sensitized Alloy 600, Corros. Sci. 43 (2001) 1321–1335. http://www.sciencedirect.com/science/article/pii/S0010938X00001499 (accessed June 5, 2013). [29] D.M. Saylor, A. Morawiec, G.S. Rohrer, The relative free energies of grain boundaries in magnesia as a function of five macroscopic parameters, Acta Mater. 51 (2003) 3675– 3686. doi:10.1016/S1359-6454(03)00182-4. [30] V. Randle, G.S. Rohrer, H.M. Miller, M. Coleman, G.T. Owen, Five-parameter grain boundary distribution of commercially grain boundary engineered nickel and copper, 56 (2008) 2363–2373. doi:10.1016/j.actamat.2008.01.039. [31] S.S. Katnagallu, S. Mandal, A. Cheekur Nagaraja, B. de Boer, S.S. Vadlamani, Role of Carbide Precipitates and Process Parameters on Achieving Grain Boundary Engineered Microstructure in a Ni-Based Superalloy, Metall. Mater. Trans. A. 46 (2015) 4740–4754. doi:10.1007/s11661-015-3064-4.

149

[32] L.E. Murr, P.J. Smith, C.M. Gilmore, Relative interfacial free energies in pure nickel, dispersion hardened nickel, and a precipitation hardened nickel-base alloy, Philos. Mag. 17 (1968) 89–106. doi:10.1080/14786436808218183. [33] L.E. Murr, G.I. Wong, R.J. Horylev, Measurement of Interfacial Free Energies and Associated Temperature Coefficients in 304 Stainless Steel, Acta Metall. 21 (1973) 595– 604. [34] J. Li, S.J. Dillon, G.S. Rohrer, Relative grain boundary area and energy distributions in nickel, Acta Mater. 57 (2009) 4304–4311. doi:10.1016/j.actamat.2009.06.004. [35] T. Skidmore, R.G. Buchheit, M.C. Juhas, Grain boundary energy vs. misorientation in Inconel® 600 alloy as measured by thermal groove and OIM analysis correlation, Scr. Mater. 50 (2004) 873–877. doi:10.1016/j.scriptamat.2003.12.004. [36] B. Alexandreanu, B. Capell, G.S. Was, Combined effect of special grain boundaries and grain boundary carbides on IGSCC of Ni–16Cr–9Fe–xC alloys, Mater. Sci. Eng. A. 300 (2001) 94–104. doi:10.1016/S0921-5093(00)01705-6. [37] S.C. Yoo, K.J. Choi, T. Kim, S.H. Kim, J.Y. Kim, J.H. Kim, Microstructural evolution and stress-corrosion-cracking behavior of thermally aged Ni-Cr-Fe alloy, Corros. Sci. (2016). doi:10.1016/j.corsci.2016.04.051. [38] D.. Wasnik, V. Kain, I. Samajdar, B. Verlinden, P.. De, Resistance to sensitization and intergranular corrosion through extreme randomization of grain boundaries, Acta Mater. 50 (2002) 4587–4601. doi:10.1016/S1359-6454(02)00306-3. [39] N. Parvathavarthini, S. Mulki, R.K. Dayal, I. Samajdar, K. V Mani, B. Raj, Sensitization control in AISI 316L ( N ) austenitic stainless steel : Defining the role of the nature of grain boundary, Corros. Sci. 51 (2009) 2144–2150. doi:10.1016/j.corsci.2009.05.045. [40] S. Patala, C.A. Schuh, A continuous and one-to-one coloring scheme for misorientations, Acta Mater. 59 (2011) 554–562. doi:10.1016/j.actamat.2010.09.058. [41] S.R. Ortner, V. Randle, A Study of the Relation Between Grain Boundary Type and Sensitization in a Partially Sensitized AISI 304 Stainless Steel Using Electron Back- Scattering Patterns, Scr. Metall. 23 (1989) 1903–1908. [42] N. Ebrahimi, P. Jakupi, A. Korinek, I. Barker, D.E. Moser, D.W. Shoesmith, Sigma and Random Grain Boundaries and their Effect on the Corrosion of Ni-Cr-Mo Alloys, J. Electrochem. Soc. 163 (2016) 232–239. doi:10.1149/2.1111605jes.

150

8.0 Effect of Laser Shock Peening on Stress Corrosion Cracking of Alloy 600 in Simulated Pressurized Water Reactor Environment

Abstract

Alloy 600 and its weldments are known to be susceptible to SCC in high temperature pure water. It is widely recognized that occurrence of SCC requires a combination of sufficiently high tensile stresses, susceptible material, and the high temperature corrosive environment.

Mitigation of SCC can be possible by addressing one or more of these factors. In this study, we investigate the effects of laser shock peening (LSP) on stress corrosion cracking (SCC) behavior of Alloy 600 in simulated pressurized water reactor (PWR) environment. Compressive residual stresses induced by LSP in the near surface region were characterized using X-Ray Diffraction

(XRD). Slow strain rate tests (SSRT) and U-Bend tests in simulated PWR environment were used to evaluate the effect of LSP on SCC behavior. Stress-strain data and crack length analysis indicates that LSP increased the resistance to SCC in PWR environment.

1. Introduction Stress Corrosion Cracking (SCC) is a major issue in the safe operation of boiling water reactors (BWR) and pressurized water reactors (PWR) [1,2]. Alloy 600 and its weldments

(82/182) that are widely used for different components in BWR/PWR are known to be particularly susceptible to SCC. In general, SCC is dependent on the presence of tensile stresses due to welding or operational stresses, a susceptible material and corrosive environment. Thus,

SCC mitigation techniques are used to address one or more these factors. Weld overlays, coolant chemistry control and surface stress modification have been used to mitigate SCC [3-5]. In recent years, surface stress improvement has been used to address SCC issues in commercial nuclear power plants. Water jet peening, shot peening, laser peening and others have been used to modify the nature of stresses locally in different components of Alloy 600 and its weld metals.

151

Laser Shock Peening (LSP) is an advanced mechanical surface treatment that has been widely used in the aerospace industry to improve component fatigue performance. LSP typically uses a Q-switched Nd:Glass or Nd:YAG (λ = 1064 nm) laser of high energy (1-8 J) and short pulse duration (6-20 ns) that passes through a transparent confining medium (water or glass) to ablate a sacrificial thin coating (tape or opaque medium) on the material surface (Figure 7.1).

After the laser beam passes through the water or glass, it is absorbed by the opaque overlay and only a thin layer (few microns) of this sacrificial layer is ablated. Thermal effects on the target material surface are suppressed [6]. As the generated plasma continues to absorb the rest of the laser energy, it is readily heated and expands between the material and the confining medium, thus generating a shock wave that propagates through the material. The volume affected by the shock wave is plastically deformed during its propagation to a depth beyond which the peak pressure does not exceed the Hugoniot Elastic limit of the material. The surrounding material in the sub-surface region is opposed to lateral straining resulting in biaxial compressive stresses near the surface.

A few studies have reported effects of residual stress mitigation techniques but the mechanism of improvement in SCC resistance is not well understood [7-9]. The present work was undertaken to systematically evaluate the effects of laser shock peening on SCC resistance of Alloy 600. Residual stresses through depth were measured after LSP treatment on Alloy 600.

Effect of LSP on SCC behavior was evaluated using slow strain rate tests (SSRT) and U-bend tests in simulated PWR environment. After SCC tests, detailed crack analysis was performed using SEM micrographs.

2. Experimental

2.1 Material and Laser Shock Peening

152

Alloy 600 (Heat NX1310) with chemical composition as listed in Table 1 was used in this

study. SSRT samples were machined using a wire –EDM and then polished to 1200 grit before

LSP treatment in the gage section. U bend samples as shown in Figure 7.2 were fabricated from

the same material.

Laser shock peening was performed using a Q-switched Nd:YAG laser (Continuum

Powerlite Plus) with infrared wavelength (1064 nm) and a frequency of 1 Hz. The target (Alloy

600 sample) was moved using a XY table to create a pattern with 50% overlay. Flowing water

was used as confining medium and vinyl tape was used as the sacrificial ablative coating. The

pulse energy was set at 3J with spot size of 2 mm diameter and pulse width of 22 ns. A

schematic of the laser shock peening process is shown in Figure 7.1.

Table 1. Chemical composition (wt.%) of Alloy 600 (Heat NX1310)

Figure 1. Schematic of the LSP process

2.2 Transmission Electron Microscopy

153

Thin foils from the SSRT samples after testing were prepared with an FEI Helios

NanoLab™ 600 DualBeam FIB/SEM. These thin foils from areas around cracks on the gage

sections were observed with a FEI Titan S/TEM operated at 300 kV and a Philips CM20 TEM operated at 200 kV. Micrographs from relevant regions modes were recorded under bright field

(BF), dark field (DF), and selected area diffraction (SAD) modes. STEM EDS analysis of local chemical composition around the cracks was also performed.

2.3 Residual stress

Residual stresses induced in the material after LSP treatment were measured using the sin2ψ technique with a Proto LXRD instrument with parameters as listed in Table 7.2.

Measurements were also done on the gage section of the LSP treated samples before and after

constant load tests. To measure residual stress through depth, samples were electro-polished

using 87.5:12.5 vol.% CH3OH:H2SO4 solution to remove layers of predetermined thickness.

Strain and layer removal corrections were applied to the residual stress values. Residual stresses

were also measured after some constant load tests using the techniques mentioned above.

Table 2. XRD parameters used for residual stress measurements

Item Description

Detector Position sensitive scintillation detector(PSSD) 20º 2θ

Radiation MnKα1 (λ = 2.10314 Aº)

Tilt angles 0, ± 2.58, ± 9.07,± 12.45,± 18.8,± 23.0

Aperture size 1 mm diameter

Plane(Bragg’s angle) {311}, 156º

X-ray elastic constant S2/2: 5.66 x 10-6 MPa-1

2.5 SCC tests

154

Samples for slow strain rate tests (SSRT) were fabricated from Alloy 600 in the as- received condition using wire EDM as per dimensions shown in Figure 7.2. Some of the samples were LSP treated in the gage section with the same parameters mentioned previously. For U

Bend tests, samples were fabricated as per ASTM G30-97 standard. U Bend samples were fabricated from the same heat of Alloy 600 in as-received and 20% cold worked conditions.

Alloy 600 nuts, bolts and spacers. LSP treatment was applied to U-bend samples along the outer radius as shown in Figure 7.2. After tests, samples were characterized using an FEI XL-30 SEM.

SSRT’s were conducted in an autoclave mounted in a constant extension rate tensile

(CERT) system at an initial strain rate of 7×10−7 s−1 at 340 °C, 2500 psi. The stress corrosion cracking tests were performed in refreshed testing loops in a simulated PWR environment. A schematic of the loop and test system is shown in Figure 7.3. The water was continuously refreshed with a flow rate of about 200 ml/min for a 4 liter autoclave and the water chemistry was continuously monitored and controlled. Dissolved gas concentration was controlled by applying an overpressure of mixed gas at room temperature before the water flows in the high pressure high temperature part of the loop. The ion content in the water was controlled by flowing water through an ion exchanger to remove corrosion products. The simulated PWR environment consisted of flowing deaerated, high purity (20.6 mS/cm) water with the addition of

1000 ppm B + 2 ppm Li and 2 ppm H2. U bend tests were also conducted in the same environment using custom made sample holder.

155

Figure 2. Schematic of the sample used for slow strain rate tests and U Bend tests. All dimensions in mm.

Figure 3. Schematic of the refreshed autoclave loop used for SSRT and U bend tests in simulated PWR environment. 3. Results and Discussion

In this section, we present and discuss the effects of LSP on the residual stresses followed by effect on stress corrosion cracking behavior in simulated PWR environment.

3.1 Residual stresses

Residual stresses versus distance from the peened surface, measured using the sin2ψ

method in two orthogonal directions (X and Y), are shown in Figure 7.4 [10]. The residual

156 stresses were of the order of about -500 MPa at the surface and the depth of the compressive residual stress layer was about 0.4 mm (in a 2 mm thick sample). The surface residual stresses have been significantly modified from tension (100 MPa) before LSP to compression (-500

MPa) after LSP. The magnitude and depth of compressive residual stresses were optimized by varying the LSP parameters like pulse energy, spot size and overlap ratio. Residual stresses were also measured on the surface of SSRT and U-bend samples before and after LSP. Residual stresses were of the order of ~300 MPa on the apex of the U bends in the as-fabricated condition.

After LSP, residual stresses were -400 MPa at the apex.

Figure 4. Residual stresses through depth in X and Y directions after LSP in Alloy 600.

3.2 Stress Corrosion Cracking tests

Slow strain rate tests conducted in simulated PWR environment were used to evaluate the effect of LSP on SCC behavior. Figure 7.5 shows stress strain curves from SSRTs for as- received (AR) and LSP treated (LSP) conditions. Analysis of stress strain curves shows an

157 increase in yield strength (YS), ultimate tensile strength (UTS) and elongation to failure and the results are summarized in Table in 7.3.

Table 3. Summary of results from all SSRTs in simulated PWR environment.

Yield Strength, Ultimate tensile strength, Maximum Sample MPa MPa elongation, % AR 344 678 59.97 LSP 547 805 70.32 LSP(repeat) 572 807 70.12

Figure 5. Stress-strain curves for AR and LSP samples tested in simulated PWR environment.

The increase in YS, UTS and elongation to failure can be attributed to residual stresses induced by LSP. These results are consistent with our work on effect of LSP on the SCC in tetrathionate solution at room temperature [10,11]. In case of the AR sample, SEM micrographs from the fracture surface shown in Figure 7.6 indicate regions of IGSCC. In addition, a number

158 of cracks were observed on the gage section of the sample and were intergranular in nature. In contrast, LSP treated samples show very few cracks as compared with the AR sample as shown in Figure 7.7. The fracture surface of the LSP treated sample showed 100% ductile cracking.

Results of analysis of the average crack lengths and number of cracks shows are shown in Table

7.4. LSP treated samples clearly show far less number of cracks and lower average crack length.

This indicates that LSP increased the resistance to SCC in simulated PWR environment.

Table 4. Summary of results from crack length analysis for AR and LSP treated samples.

Sample No. of Average crack length, Max. crack length, cracks/mm2 µm µm AR 80 78.77 212 LSP 15 33.81 47

Figure 6. (a, b) SEM micrographs showing cracks in the gage section and (c) IGSCC region on the fracture surface of Alloy 600 sample in AR condition after SSRT

159

Residual stresses induced by surface treatments including LSP relax under thermal and

mechanical loading [12,13]. As the LSP treated samples were loaded beyond the YS of the

sample, residual stresses relaxed. The effect of LSP on SCC can be evaluated by U bend tests. In

this study, U bend tests were used for this purpose. After LSP treatment on the outer edge of the

U bend samples, compressive residual stresses of about -450 MPa were induced on the surface at

the apex. Preliminary results after 500 hours of exposure to simulated PWR conditions show

very little cracking on the LSP treated samples. Long duration tests are being performed to

confirm these results.

Figure 7. (a, b) SEM micrographs from the gage section and (c) IGSCC region on the fracture surface of Alloy 600 sample in LSP condition after SSRT

To evaluate the effect of LSP on the oxidation behavior and microstructure, thin foils were obtained using FIB from cracks on the gage sections of the AR and LSP conditions. Figure

7.8 shows TEM micrographs from the AR and LSP conditions after SSRT in simulated PWR

160 environment. Figure 7.8 c shows the formation of voids ahead of the crack tip while narrow twins and dislocation cells can be observed in Figure 7.8 d.

Figure 8. TEM micrographs from AR (a, b) and LSP (c, d) samples after SSRT.

Figure 9. HAADF image along a crack and corresponding elemental EDS maps

161

High angle annular dark field and corresponding elemental EDS maps for O, Cr, Fe and

Ni from area around a crack from the AR sample is shown in Figure 7.9. Preferential enrichment of Cr and O is observed near the crack tip while Ni and Fe are depleted.

4. Conclusions

The following conclusions can be drawn from this study.

1) Deep compressive residual stresses were induced in Alloy 600 after LSP. Residual

stresses were of the order of -500 MPa on the surface and extended to 0.4 mm.

2) SSRTs in simulated PWR environment show increase in YS, UTS and elongation to

failure after LSP as compared with AR sample. Average crack length, number of cracks

and maximum crack length were significantly lower after LSP as compared with

untreated Alloy 600.

3) U-bend tests show increased resistance to SCC in simulated PWR environment after LSP

treatment.

5. References [1] S.J. Zinkle, G.S. Was, Acta Mater. 61 (2013) 735–758. [2] Mater. Reliab. Progr. Tech. Basis Prim. Water Stress Corros. Crack. Mitig. by Surf. Stress Improv. (MRP-267, Revis. 1), EPRI, Palo Alto, CA 1025839. (2012). [3] W. Sagawa, T. Aoki, T. Itou, K. Enomoto, E. Hayashi, T. Ishikawa, Nucl. Eng. Des. 239 (2009) 655–664. [4] S.S. Hwang, J. Nucl. Mater. 443 (2013) 321–330. [5] S.J. Green, Int. J. Press. Vessel. Pip. 25 (1986) 359–391. [6] P. Peyre, R. Fabbro, Opt. Quantum Electron. 27 (1995) 1213–1229. [7] I. Nikitin, I. Altenberger, Mater. Sci. Eng. A 465 (2007) 176–182. [8] R.K. Nalla, I. Altenberger, U. Noster, G.Y. Liu, B. Scholtes, R.O. Ritchie, Mater. Sci. Eng. A 355 (2003) 216–230. [9] W. Tsai, C. Chang, J. Lee, Corrosion 50 (1994) 98–105. [10] A. Telang, A.S. Gill, S. Teysseyre, S.R. Mannava, D. Qian, V.K. Vasudevan, Corros. Sci. 90 (2015) 434–444. [11] A. Telang, C. Ye, A. Gill, S. Teysseyre, S.R. Mannava, D. Qian, V.K. Vasudevan, 16th Int. Conf. Environ. Degrad. Mater. Nucl. Power Syst. React. Asheville, NC 90 (2013) 434–444. [12] P.J. Withers, H.K.D.H. Bhadeshia, Mater. Sci. Technol. 17 (2001) 355–365. [13] Z. Zhou, A.S. Gill, A. Telang, S.R. Mannava, K. Langer, V.K. Vasudevan, D. Qian, Exp. Mech. 54 (2014) 1597–1611.

162

9.0 Effect on Mechanical Loading and Temperature on Relaxation of Residual Stresses Induced by Surface Treatments in Alloy 600

Abstract Mechanical surface treatments like shot peening, laser shock peening, water jet peening, ultrasonic peening and others induce large compressive residual stresses in the near surface region. The stability of residual stresses is important to understand the mechanisms for improvement in properties including fatigue and stress corrosion cracking. In this study, effect of applied tensile stress and temperature on LSP induced residual stresses was investigated using in-situ neutron and high energy X-ray diffraction (XRD) and ex-situ high energy XRD. Under static loading conditions, residual stresses relaxed as the applied stress exceeded the yield strength of the treated material. Compressive residual stresses induced by LSP relaxed by ~ 15% on exposure to 350˚C for more than 2500 hours. Application of tensile stresses relaxed compressive residual stresses.

1. Introduction

Mechanical surface enhancement treatments like Shot Peening (SP), Laser Shock

Peening (LSP), Cavitation Jet Peening (CSP) and others introduce compressive residual stress.

Improvement in fatigue including high temperature fatigue and stress corrosion cracking have been co-related with the introduction of compressive residual stresses in the near surface region

[1-4]. The depth and magnitude of residual stresses introduced by each treatment is different [5].

LSP uses a Q-switched laser for ablating a thin coating (usually a vinyl tape) on a material surface, with a high energy (0.5- 8J), short duration (8-25 ns) laser pulse. The material is covered with a transparent confining medium like water. The plasma produced on ablation expands rapidly and finally blows off, introducing a shock wave into the material. The volume affected by the shockwave is plastically deformed during its propagation to a depth beyond which the

163 peak pressure pulse does not exceed the Hugoniot Elastic limit (HEL) of the material. The surrounding material in the sub-surface region is opposed to this lateral straining, resulting in a biaxial compressive stress near surface. A number of studies have demonstrated the beneficial effect of LSP in improvement of fatigue life [3,6-8]. The process is controlled by varying the energy of the laser, spot size and pulse width to create a patch as per requirement. A typical application of the process, utilizes a Q-switched Nd: Glass laser (λ=1.054 µm) or Nd:YAG laser

(λ=1.064 µm). A number of studies provide excellent description of the process and physics behind it [9].

Cavitation shotless peening is a peening process which makes use of impacts of cavitation bubbles to introduce plastic deformation in near surface regions. Cavitation is a phase change phenomenon (from liquid to gas phase) and refers to formation and implosion of bubbles in a liquid. On imploding, these bubbles generate shock waves which introduce plastic deformation in near surface regions of materials. Cavitation bubbles are well known to cause erosion in hydraulic equipment, but in cavitation peening the conditions are controlled to cause deformation without any removal of material.

In cavitation peening, liquid phase is turned into gas phase by a reduction in static pressure, till the saturated vapor pressure is reached, by increasing the flow velocity. The cavitation nuclei are tiny bubbles present in high speed region of jet. On increasing the static pressure (by decreasing flow velocity), the bubbles implode. On implosion, these bubbles produce micro jets of high velocity (~1500 m/s), which causes plastic deformation in the material. As the cavitation bubble shrinks, they rebound causing a shock wave, which also causes plastic deformation. The cavitation is normally produced by injecting a high-speed water jet in a water chamber (CSP in water), but now it is possible to produce cavitation in air (CSP in

164 air), by injecting a high speed water jet into a low speed water jet which is then injected into air as shown in [10,11]. This has widened the scope of application of cavitation peening as it does not require a water filled chamber for peening. The parameters used to control the process include pressure, traverse rate of jet, stand-off distance, exposure time, and types of jets used.

The mechanisms behind both the techniques i.e. CSP in air and water have been discussed by

Soyama [10,11].

The application of mechanical surface treatments is dependent on the stability of the residual stresses under environmental conditions. In power plants, this environment is high temperature (~288˚C to 360˚C) and components are subjected to varied operational stresses for a long time. Hence, it is essential to understand the effect of both temperature and applied stresses on the residual stresses induced by surface treatments to evaluate their suitability for an application. In this study, we investigate the effect of both applied static load as well as temperature on the residual stresses induced by LSP and CSP treatments. In-situ neutron, synchrotron X-ray diffraction and ex-situ conventional XRD were used to quantify residual stresses.

2. Experimental

Alloy 600 material with composition as shown in Table 8.1 was used in this study. Samples were treated with LSP and cavitation peening treatments with parameters as given in Table 8.2 and 8.3 respectively.

Table Error! No text of specified style in document..10. Chemical composition of the Inconel Alloy 600 used in this study.

C Mn Si S Cr Fe Co Cd Ti Cu P Al Ni

0.001 0.001 0.08 0.16 0.18 14.99 8.05 0.18 0.01 0.18 0.1 0.08 Bal. max. max.

165

Table Error! No text of specified style in document..11. Laser shock peening parameters.

Energy, J Pulse width, ns Spot Power density, Overlap diameter, mm GW/cm2 ratio 3 22 2 5.6 50 5 22 2 7.2 50

Table Error! No text of specified style in document..12. Cavitation peening parameters.

Type Injection Nozzle Standoff Scanning speed, pressure, MPa diameter, mm distance, mm s/mm Air 3 22 2 5.6

Residual stresses were also measured using Proto LXRD by the d vs. sin2ψ method. The

{311} plane with Bragg angle of 156˚, MnKα radiation and 2 mm spot size was used for strain

measurements. Samples were loaded to applied stresses of 100-600 MPa in steps of 100 MPa

with an extension rate of 10-3 mm/s using a servo-hydraulic tensile testing system. After each step, samples were unloaded and residual stresses were measured on the sample surface.

To quantify thermal relaxation of residual stresses, samples with dimensions of 15 mm x

15 mm x 2 mm were LSP and CSP treated with parameters as shown in Table 8.2 and 8.3, respectively. These samples were then wrapped in Ti foil to minimize oxidation and exposed to

350˚C in a laboratory furnace for different time intervals. Residual stresses were measured on the surface of the sample with the same procedure as mentioned previously after each interval.

3.0 Results and Discussion a) Effect of static load

It is known that LSP introduces large compressive residual stresses to a significant depth up to 0.5 mm to 1 mm depending on the material, process parameters and thickness of the material. The applicability of these surface treatments depends on the stability of these residual stresses under applied stress and temperature/environmental conditions. Figure 8.1 shows the

166

effect of applied stress on the residual stresses for 2 LSP conditions. A nominal applied stress of

36 MPa was considered as the as-treated condition. As the applied stress increases, residual

stresses relax from this initial value. With high energy neutrons, residual stress measurements

can be made at different depths. The resolution in these measurements is 0.2 mm and this results

in average values as compared with residual stress values determined by iterative layer removal

method. These measurements provide a good insight on relaxation of residual stresses at the

surface and in depth from the treated surface. The gradient nature of residual stresses induced by

repetitive shock loading during the LSP process is also seen in Figure 8.1. More significantly,

residual stresses values on the surface were significantly lower than the applied stress for both

LSP treatment conditions. For example, the measured residual stress on the surface was about -

50 MPa and -225 MPa when the applied stresses were 128 MPa and 330 MPa respectively. This is observed till the applied stresses were high enough to cause localized yielding, thereby

relaxing residual stresses on the surface and the interior.

Figure Error! No text of specified style in document..14. Effect of applied stress on measured residual stress in for a) LSP-3J and b) LSP 5J conditions. Data points correspond to stress MPa, strain values at different depths from the surface in the 2 mm thick sample.

167

In-situ relaxation of residual stresses induced by CP and LSP was also measured using

high energy X-rays and the results are as shown in Figure 8.2a and b respectively. In the sample

treated with CP, the measured residual stresses on the surface were lower than applied stress till

the applied stress exceeded 321 MPa. In case of the LSP treated sample, the residual stresses on

the surface were lower than the applied stress of ~600 MPa.

Figure Error! No text of specified style in document..15. Effect of applied static load on residual stress in for a) CP and b) LSP conditions.

168

To study the effect of quasi-static loading, LSP treated sample was loaded to

predetermined different stress values from 100-600 MPa. After each loading and unloading step,

residual stresses measured using XRD are as shown in Figure 8.3. Residual stresses were stable

for every step till the applied stress exceeded the yield stress of the LSP treated sample i.e ~ 550

MPa. Interestingly, the yield stress for the untreated Alloy 600 sample is ~325 MPa. It is known

that LSP induces deep compressive residual stresses and hardening in the near surface region

[9,136,148] which leads to the higher yield stress. This study indicates that residual stresses were

stable as long as the applied stress did not exceed the yield stress of the treated sample. In case of

the CP treated sample shown in Figure 8.4, residual stresses relaxed as the applied stress

exceeded the yield stress of the CP treated sample (~325 MPa), this stress being much higher

(~600 MPa) in the LSP treated case.

Figure 8.3. (a) Stress-strain curves from quasi-static loading on the LSP treated sample. (b) Plot of measured residual stress after each loading step.

169

Figure Error! No text of specified style in document..16. (a) Stress-strain curves from quasi- static loading on the CP treated sample. (b) Plot of measured residual stress after each loading step. b) Effect of temperature

Long term exposure to high temperatures is likely to cause relaxation of residual stresses induced by mechanical surface treatments like

LSP and CP. In this study, we periodically measured surface residual stresses on LSP and

CP treated Alloy 600 samples and the results are shown in Figure 8.5. Both samples show initial relaxation of close to 10% after exposure to 350˚C followed by very little relaxation up to 2000 hours. This indicated Figure 8.5 Effect of temperature on residual stresses that residual stresses would be stable and effective under the operating conditions for this alloy for a relatively long duration of time.

5.0 References [1] C. Ye, A. Telang, A.S. Gill, S. Suslov, Y. Idell, K. Zweiacker, J.M.K. Wiezorek, Z. Zhou, D. Qian, S. Ramaiah Mannava, V.K. Vasudevan, Mater. Sci. Eng. A 613 (2014) 274–288. [2] R.K. Nalla, I. Altenberger, U. Noster, G.Y. Liu, B. Scholtes, R.O. Ritchie, Mater. Sci. Eng.

170

A 355 (2003) 216–230. [3] C. Ye, S. Suslov, B.J. Kim, E. a. Stach, G.J. Cheng, Acta Mater. 59 (2011) 1014–1025. [4] A. Telang, A.S. Gill, S. Teysseyre, S.R. Mannava, D. Qian, V.K. Vasudevan, Corros. Sci. 90 (2015) 434–444. [5] A. Gill, A. Telang, S.R. Mannava, D. Qian, Y.-S. Pyoun, H. Soyama, V.K. Vasudevan, Mater. Sci. Eng. A (2013). [6] I. Nikitin, B. Scholtes, H.J. Maier, I. Altenberger, Scr. Mater. 50 (2004) 1345–1350. [7] J.Z. Lu, L. Zhang, A.X. Feng, Y.F. Jiang, G.G. Cheng, Mater. Des. 30 (2009) 3673–3678. [8]] P. Peyre, R. Fabbro, P. Merrien, H.P. Lieurade, Mater. Sci. Eng. A Struct. Mater. Prop. Microstruct. Process. 210 (1996) 102–113. [9] R. Fabbro, P. Peyre, L. Berthe, X. Scherpereel, J. Laser Appl. 10 (1998) 265. [10] H. Soyama, T. Kusaka, M. Saka, J. Mater. Sci. Lett. 20 (2001) 1263–1265. [11] H. Soyama, JSME Int. Journal, Ser. A Solid Mech. Mater. Eng. 43 (2000) 173–178.

171

10.0 Gradient Nanostructure and Residual Stresses Induced by Ultrasonic Nano-crystal Surface Modification in 304 Austenitic Stainless Steel for High Strength and High Ductility

Chang Ye1*, Abhishek Telang2, Amrinder S. Gill2, Sergey Suslov3, Yaakov Idell4, Kai Zweiacker4, Jörg M.K. Wiezorek4, Zhong Zhou5, Dong Qian5, Seetha Ramaiah Mannava2, Vijay K. Vasudevan2 1Department of Mechanical Engineering, University of Akron, Akron, OH 45325 2Department of Mechanical and Materials Engineering, University of Cincinnati, Cincinnati, OH 45221-0072 3School of Materials Engineering, Purdue University, West Lafayette, IN 47906 4Department of Mechanical Engineering and Materials Science, University of Pittsburgh, Pittsburgh, PA 15261 5Department of Mechanical Engineering, University of Texas at Dallas, Richardson, TX 75080 Corresponding author: email: [email protected] phone: 330-972-4032 Fax: 330-972-6027

Published in Materials Science and Engineering, A613, 274-288 (2014); http://dx.doi.org/10.1016/j.msea.2014.06.114

Abstract

In this study, the effects of Ultrasonic Nano-crystal Surface Modification (UNSM) on residual stresses, microstructure changes and mechanical properties of austenitic stainless steel 304 were investigated. The dynamic impacts induced by UNSM leads to surface nanocrystallization, martensite formation, and the generation of high magnitude of surface compressive residual stresses (-1400 MPa) and hardening. Highly dense deformation twins were generated in material subsurface to a depth of 100 µm. These deformation twins significantly improve material work-hardening capacity by acting both as dislocation blockers and dislocation emission sources. Furthermore, the gradually changing martensite volume fraction ensures strong interfacial strength between the ductile interior and the two nanocrystalline surface layers and thus prevents early necking. The microstructure with two strong surface layers and a compliant interior embedded with dense nanoscale deformation twins and dislocations leads to both high strength and high ductility. The work-hardened surface layers (3.5 times the original hardness) and high magnitude of compressive residual stresses lead to significant improvement in fatigue performance; the fatigue endurance limit was increased by 100 MPa. The results have demonstrated that UNSM is a powerful surface engineering technique that can improve component mechanical properties and performance.

Keywords: Ultrasonic Nano-crystal Surface Modification (UNSM); ultrasonic peening; deformation twins; gradient microstructure; deformation-induced martensite; residual stresses; fatigue performance; precession electron diffraction (PED).

172

1. Introduction:

Generating a nanocrystalline layer at the material surfaces can lead to significant improvement in mechanical properties and thereby their performance. Many mechanical surface processing techniques have been reported to induce surface nanocrystallization, for example, by surface mechanical attrition treatment (SMAT) [1–3], shot peening (SP) [4], ultrasonic shot peening (USP) [5,6], laser shock peening (LSP) [7–14], etc. Surface properties (wear resistance

[15][16], corrosion resistance [17], biocompatibility [18], etc) and the bulk mechanical properties

(tensile, fracture, fatigue, etc) can be significantly improved through surface nanocrystallization.

Material strength improvement is typically accompanied by loss of ductility. For FCC (face centered cubic) materials, the ductility is strongly affected by the work-hardening capacity, i.e., the capacity to accumulate and store dislocations during tensile deformation. There have been many attempts in the past to produce materials with innovative microstructures that can lead to both high strength and high ductility by introducing, for example, nanoscale precipitates [19–23], highly dense stacking faults [24] and nanotwins [25–28], bimodal microstructure [29–31], hierarchical microstructure [32,33], etc. The idea of hierarchical microstructure originates from nature, where many biological structures possess high strength and high damage resistance [34–

37]. Recently, SMAT has been employed to fabricate hierarchical steels that possess both high strength and high ductility [3,32,38].

Ultrasonic Nanocrystal Surface Modification (UNSM) [39,40] is a recently developed technique that utilizes low amplitude ultrasonic frequency vibrations superimposed on static load to induce severe plastic deformation (SPD) that leads to surface nanocrystallization. During

UNSM, the sample surface is struck by a tungsten carbide ball attached to an ultrasonic device vibrating at high frequency (10 to 30 KHz). The repeated, high frequency strikes cause SPD,

173 which leads to nanocrystallization and compressive residual stresses at and below the material surface to a certain depth that depends on the amplitude, load and strike rate. Unlike the hand- held ultrasonic peening system, the UNSM system is typically integrated in a lathe that holds the part, so that the processing conditions can be precisely controlled for high-throughput industrial manufacturing. UNSM has been successfully used to process steel [40,41] and magnesium alloys

[42] for improved mechanical properties and performance. A recent study has compared UNSM with LSP and cavitation peening in IN718 SPF (superplastic forming) alloy [12]. However, in- depth investigations of how UNSM affects material microstructure and how this affects the mechanical behavior are still lacking.

Similar to SMAT, UNSM also utilizes mechanical strikes to generate plastic strain on material surface. Unlike SMAT, where neither the intensity nor the density of the strikes can be precisely controlled, both the strike intensity and density can be precisely manipulated for best performance in UNSM. This makes the UNSM process highly repeatable and thus more reliable for industrial applications. In addition, the UNSM system can be easily integrated into modern manufacturing system.

Stainless steels are widely used in a number of industries (aerospace, automotive, nuclear, biomedical, etc) and have been studied extensively. In this study, UNSM processing of austenitic stainless steel (SS) 304 was carried out and the treated samples were characterized by hardness testing, X-ray diffraction (XRD), scanning electron microscopy (SEM) with electron backscatter diffraction (EBSD) and transmission electron microscopy (TEM). The mechanical properties of the 304 stainless steel samples after UNSM processing were evaluated by tensile test and three-point bending fatigue test. The results are presented and discussed in the following.

2. Experimental details

174

2.1 Materials

Samples (20 mm X 20 mm X 1.8 mm) were cut by electric discharge machining (EDM) from a plate of AISI 304 stainless steel, the nominal composition of which is 0.08 C, 1.00 Si,

2.00 Mn, 0.045 P, 0.03 S, 18.0-20.0 Cr, 8.0-10.5 Ni, and balance Fe (all wt.%).

2.2 Ultrasonic Nano-crystal Surface Modification Experiments

In the UNSM process, a tungsten carbide ball (diameter 2.38 mm) attached to an ultrasonic device scans over the surface while striking it at high frequency (10 to 30 kHz).

During the strike, the depth that the tungsten carbide ball moves into the target material is called the amplitude, which typically ranges from 10 to 40 µm. At the same time, a static load

(typically 10 to 50 Newton) is applied to the ball against the material surface. The parameters in the UNSM process include: the static load, the amplitude of the strike, the scan speed, the intervals between neighboring scans, and the ultrasonic peening frequency. Detailed description of UNSM has been provided in [39,40].

In this study, the UNSM experiment was carried out by an LM-520 UNSM system with the conditions of static load of 20N, amplitude of 10 µm, frequency of 20 kHz, and scanning speed of 3000 mm/second. The interval between each pass was 10 µm.

2.3 Microstructural Characterization:

The changes brought about by the UNSM process were studied using a number of techniques as below.

Electron Backscatter Diffraction (EBSD/Orientation Imaging Microscopy (OIM):

Sample preparation: Cross-section EBSD samples were prepared by mechanical polishing, followed by electro-polishing with 87.5% methanol and 12.5% sulfuric acid at a voltage of 24 volts for 20 seconds. EBSD scans with a step size of 0.3 µm were carried out using a Genesis

175

4040 EDAX/TSL EDS/EBSD system in a XL-30 FEI SEM operating at 25 kV at a magnification of 500X.

Transmission Electron Microscopy (TEM): The TEM sample at the top surface was prepared by grinding from the side opposite to the UNSM-treated surface, followed by ion- milling using a Fischione Model 1010 system. The in-depth TEM samples were prepared by the

Focused Ion Beam (FIB) lift-out method in an FEI Nova-200 FIB system in Birck

Nanotechnology Center at Purdue University. TEM observations were carried out by an FEI

Titan TEM operated at 300 kV and a Philips/FEI CM20 TEM operated at 200 kV.

TEM-based Orientation Imaging Microscopy (OIM) Analysis: Nanoscale lateral spatial resolution OIM analysis has been performed by automated acquisition and indexing of precession electron diffraction (PED) patterns with a JEM2100F TEM equipped with the

DIGISTAR/ASTAR system from NanoMEGAS at the University of Pittsburgh. Precessed illumination, 0.6˚ precession angle, and electron beam focused to ~3nm in diameter at the TEM specimen section surface was scanned across a pre-selected area of interest with step-sizes as small as 2nm to obtain maps of PED patterns, which were indexed automatically by optimized matching to computer generated reciprocal lattice based templates of the austenite and martensite phases of interest here. The PED based TEM OIM data sets provide information akin to that available via EBSD based OIM in the SEM but with lateral resolution in areal maps being on the order of 1nm, limited essentially by the electron beam diameter used in the TEM instrument. The raw data sets of the PED based orientation indexed areal maps were processed and analyzed further with the TSL OIM Data Analysis software.

Phase Analysis: X’Pert MRD PRO X-ray powder diffraction system with Cu-Kα radiation source and a monochromator was used to analyze the phases in the sample after

176

UNSM. In-depth XRD analysis was carried out after electropolishing the material layer by layer

with the same solution as used for the preparation of the EBSD samples.

Residual Stress: Residual stresses were measured by a Proto LXRD system with Cr-Kα

radiation and {211} peak for martensite phase and Mn-Kα radiation and {311} peak for austenite

-6 -1 phase. The X-ray elastic constant S2/2 used was 6.04*10 MPa for the martensite phase and

7.18*10-6MPa-1 for the austenite phase. The diameter of the X-ray beam was 1 mm. In-depth residual stresses were measured by electropolishing the material layer by layer. The electropolishing solution was composed of 12.5 volume percent sulfuric acid and 87.5 volume percent methanol. The electropolishing was carried out in an Electro4 Met system from Buehler with a voltage of 24 volts. Strain gradient correction and layer removal corrections were carried out in accordance with SAE J784a standard (residual stress measurement by X-ray diffraction).

2.4 Mechanical Property Tests:

Hardness: The hardness change of the samples was measured by a nano-indentation system (CSM Instruments) with a Berkovich indenter with a maximum load of 100 mN and 10 seconds holding time. An average of five measurements was used for each reported data point.

Tensile test: Tensile test samples were cut by an electro-discharge machine (EDM) from a sheet. Both of two opposite faces of the gauge area were processed by UNSM before tensile test. The tensile test was carried out at room temperature with a strain rate of 1.9 x 10-5 s-1.

Fatigue test: Three-point bending fatigue test was carried out using a MTS Bionix servo-

hydraulic fatigue test machine. The loading profile is a sine function and the frequency is 5 Hz

and the stress ratio is 0.1. The span of the bending fatigue test was 30 mm. All tests were carried

out at room temperature and in a laboratory environment. The sample thickness was 2.8 mm with

177

a gauge width of 10 mm and a gauge length of 40 mm. The fatigue samples were cut by EDM

from a large sample processed by UNSM on both sides.

3. Results and Analysis

3.1 Near-Surface and Through-the-depth Phase distribution after UNSM

Like many other mechanical surface processing techniques, the plastic deformation and

strain induced by UNSM assumes a gradient nature, with the top surface having the highest

plastic strain, followed by a gradual decrease in strain deeper into the material. Thus, the

martensite volume fraction also assumes a gradient with distance from the surface. To investigate

the in-depth martensite volume fraction, sequential layer removal by electrochemical polishing

and XRD analysis was performed. Figure 1a shows the XRD patterns recorded from the UNSM

sample with 10 µm interval at different depths. At the top surface and 5 µm below, the structure

is fully martensite, which is similar to the results reported in SS 304 processed by SMAT [2] and ball milling [43]. At 12 µm below the surface, the austenite phase peak at around 75o 2-theta appears. With increasing depth into the material, the martensite volume fraction decreases while the austenite phase volume fraction increases gradually. In addition, an ε-martensite peak at around 47.5o 2-theta appears from 12 µm below surface, and its intensity increases gradually

with depth into the material. At 79 µm below the surface, the microstructure is almost fully

martensite peak ׳austenitic except for the ε-martensite peak at around 47.5o 2-theta and a small α

at around 44.5o 2-theta. Transformation from austenite to ε-martensite at the subsurface has been

observed in SS 304 subjected to SMAT by Chen and co-workers [38]. Figure 1b shows the

martensite volume fraction at different depths, estimated by the direct comparison method by

calculating the contribution of different phases to the XRD peak intensities [44]. The martensite

178 volume fraction at the top surface is 100% and decreases sharply deeper into the material. At around 100 µm below the surface, the material is almost purely austenite.

(a) (b)

Figure 1. (a) XRD patterns at different depth of SS 304 after UNSM, (b) in-depth volume fraction of the martensite after UNSM.

3.2 Near-Surface and Through-the-Depth Microstructure Observed by Optical Microscopy and EBSD

The near-surface and through-the-depth microstructure was characterized by optical microscopy, SEM and EBSD. Figure 2 shows optical micrographs of the sample surface before and after UNSM processing. Before UNSM processing, the grain size is around 20 µm and sporadic annealing twins can be observed. After UNSM, a more refined microstructure with evidence for extensive plastic deformation can be observed on the surface. The cross-sectional view (Fig. 3a) of the sample shows a gradient microstructure with a SPD layer of around 50 µm.

Higher magnification optical (Fig. 3b) and SEM (Fig. 3c) images of the cross section reveals bent grains and a lot of deformation bands within grains as marked by the arrows.

EBSD has been widely used to characterize phase transformations induced by plastic deformation in stainless steels. Figure 4 shows the phase distribution from the EBSD map recorded from the cross-section of the UNSM treated sample. A layer of very refined grains can be observed at the very top surface. In Fig. 4a, the red color represents the martensite phase,

179 whereas the green color represents the austenite phase. The red color of the top layer means that fully martensite phase exists at the top surface, which corroborates well with the XRD result in

Fig. 1b. It should be noted that due to the limited resolution of the EBSD system, the fine features at the top surface could not be resolved. Thus, precession electron diffraction (PED) in the TEM has been used to further characterize this region and is presented later in this paper.

(a) (b)

Figure 2. Optical image at the top surface of the sample before (a) and after (b) UNSM

Surface

(b)

(c)

Figure 3. Optical (a,b) and SEM (c) images showing the cross section microstructure of the sample after UNSM; the arrows point to the deformation bands

180

Surface Surface

(f)

(d) (e)

Figure 4. Phase distribution from EBSD analysis in the cross section (all top edges correspond to the top surface of the samples) of the sample after UNSM: (a) phase map: red represents martensite phase, green represents austenite phase, (b) inverse pole figure (IPF) in the same region as in (a), (c) legend of IPF map in (b) and (e), (d) grain boundary map, (e) IPF map of the region in (d), (f) legend of the grain boundary map in (d)

Deeper into the material, the martensite volume fraction decreases and the austenite volume fraction increases. At the subsurface, small grains exist in-between big grains because most refined grains are generated near the grain boundaries. During plastic deformation, stress

181 concentration occurs at the grain boundaries, which is favorable for martensitic phase transformation. On the other hand, martensitic transformation also contributes to grain refinement [2,38]. This leads to the generation of martensite phase and fine grains along the grain boundaries. Figure 4d shows the grain boundary map, where the green colored boundaries represent the twin boundaries. In the very top surface, fine twin boundaries are intermixed with the refined grains. According to Fig. 4f, 7% in number and 23% in length of the boundaries are twin boundaries. Figure 4b and 4e show the inverse pole figure (IPF) maps of the grains in the regions in Fig. 4a and 4d, respectively. In both figures, random grain orientation can be observed.

3.3 Top-surface and Sub-surface microstructure observation by TEM and precession electron diffraction (PED)

Figure 5 shows the TEM images from the top surface (Fig. 5a) and 10 µm below surface

(Fig. 5b) of the SS 304 sample after UNSM. Nano-grains are observed at the top surface. The diffraction pattern recorded from the top surface (inset in Fig 5a) reveals many rings, indicating the existence of the nano-grains, and all the rings were indexed as arising from martensite phase, which corroborates well with the XRD result (Fig. 1b).

(a) (b)

Figure 5. Bright field TEM images from the top surface (a) and 10 µm below surface (b) of the sample after UNSM.

182

Due to the limitation of traditional TEM and EBSD, the grain size at the top surface cannot be precisely characterized. Thus, precession electron diffraction (PED) in a field-emission gun equipped high-resolution TEM (HR-TEM), which has been used successfully previously for the analysis of nano-scale refined and heavily plastically deformed microstructures in steels [45], was used here to accurately characterize the grains in the modified subsurface regions.

Figure 6 shows the PED images from the top surface of the sample after UNSM. Figure

6a shows the IPF map. Figure 6c shows the grain boundary map, from which, we can observe that 11.2% of the grain boundaries are twin boundaries. Figure 6e shows the grain size histogram, where we can observe that the majority of the grains are less than 10 nm, while the average grain size at the top surface is 4.1 nm. Figure 6f shows the phase distribution at the top surface. As expected, fully martensitic phase has been observed. This corroborates well with the

XRD (Fig. 1), the EBSD (Fig. 4a) and conventional TEM (Fig. 5a) observations.

(a) (b)

183

(d)

(c)

(e)

(g)

(f)

Figure 6. PED images from the top surface: (a) IPF and the legend (b), (c) grain boundaries map and the legend (d), (e) histogram map of the grains, (f) phase map and the legend (g)

184

Figure 7 shows the PED images from 10 µm below the surface of the sample after

UNSM. Figure 7a shows the IPF map. Figure 7c shows the grain boundary map, where we can observe that 4.5% of the grain boundaries are twin boundaries. Figure 7e shows the grain size histogram, where we can observe that the majority of the grains are less than 50 nm, while the average grain size at 10 µm below the surface is 18 nm. Figure 7f shows the phase distribution at

10 µm below the surface. Less than 2% austenitic phase has been observed, which corroborates well with the results of XRD (Fig. 1b) and the EBSD (Fig. 4a).

(b)

(a)

(d) (c)

185

(e)

(g)

(f)

Figure 7. PED images from 10 µm below the top surface: (a) IPF map and the legend (b), (c) grain boundaries map and the legend (d), (e) histogram map of the grains, (f) phase map and the legend (g)

3.4 Sub-surface Deformation Twins Observed by TEM

Due to the gradient nature of the plastic strain induced by UNSM, the microstructure also assumes a gradient nature. Figure 8 shows the TEM images from 20 µm below the surface of the

UNSM sample. Highly dense deformation twins with different orientations can be observed.

Note that deformation twins are present throughout the sample (Fig 8a). Figure 8b shows a dark field image of the deformation twins. Figure 8c and 8d show the deformation twins in another

186 region at different magnifications. The presence of the deformation twins can also be confirmed by the diffraction pattern (Fig. 8e) taken from the region marked by the green ring in Fig. 8d. At this depth (20 µm), the accumulation of highly dense dislocations from the high magnitude of the plastic strain leads to stress concentration, which is favorable for deformation twins to nucleate.

It should be pointed out that some twins are curved, which could be caused by further straining after the twins are formed. The high magnification image in Fig. 8f reveals that the twin thickness is only a few nanometers, which is due to the high strain rate and high flow stress in the region caused by the dynamic impact loading. As pointed out by Chen and co-workers [38], the higher the flow stress, the thinner the deformation twins. We can also observe the formation of Rhombic blocks (marked in Fig. 8f), which has also been observed in other studies [2,8].

(a) (b)

187

(e) (f)

Figure 8. TEM images of the sample after UNSM at 20 µm below the surface showing highly dense deformation twins: (a) bright field image, (b) dark field image, (c) bright field image of another region, (d) high magnification bright field image, (e) diffraction pattern taken from the ring in (d) revealing reflections at 1/3{110} positions from the deformation twins, (f) bright field image at high magnification showing the rhombic blocks

Deeper into the material, the twin density decreases gradually. Figures 9a and Fig. 9b show the TEM images from 50 µm below the surface of the UNSM sample. We can still observe some deformation twins, but the density is lower than that in the 20 µm sample (Fig. 8). This is because deeper into the material, both the plastic strain and the plastic strain rate are much lower

188 and thus less favorable for deformation twins to form. Similar to the deformation twinning structure in Fig. 8a, twins can also be observed throughout the sample in Fig. 9a.

(a) (b)

Figure 9. TEM images of the sample after UNSM at 50 µm below the surface: (a) bright field image, (b) dark field image; (c) dark field TEM images of the sample after UNSM at 100 µm below the surface: (d) High magnification dark field image of (c)

Figure 9c and Fig. 9d show the microstructure of the TEM sample at 100 µm below the surface. Two grains can be observed in the TEM sample, the first at the upper left region, the other at the lower right region. In the first grain, very few deformation twins can be observed,

189 whereas those in the second grain are very dense. The dramatic different deformation twinning behavior stems from the different orientations of the grains relative to the loading direction, which results in different Schmid factors and thus affecting the stress applied to the grain during

UNSM striking. It should be noted that the deformation twins appear to originate at the grain boundaries. This is because the grain boundaries are locations where stress concentrations occur and thus are favorable for the nucleation of the deformation twins.

3.5 Near-Surface and Through the Depth Residual Stresses after UNSM

The residual stress profile is of interest because it affects component fatigue performance.

In this study, XRD was used to measure the residual stresses induced by UNSM. Due to the phase transformation induced by UNSM, two phases (austenite and martensite) exist near the surface. Thus, residual stresses in both the two phases were evaluated. Sequential layer removal by electrochemical polishing followed by XRD residual stress measurement in both martensite and austenite phase were carried out till a depth of 47 µm. Beyond that only residual stresses in austenite phase were measured because minimal martensitic phase existed. Figure 10a shows the residual stresses in austenite phase in both the X and Y directions. The X direction was defined as the traveling direction of the UNSM tip. The Y direction is perpendicular to the X direction.

Note that the residual stress in the austenite phase on the top surface (0-5 µm) could not be measured because only the martensite phase existed there. In Fig. 10a, the residual stresses in X direction assume a typical distribution seen in most surface processing techniques, with a high magnitude of compression at the surface and a gradual decrease deeper from the surface. The depth of the compressive residual stresses generated in SS 304 by UNSM with the parameters in this study is 0.276 mm, close to that of SP and lower than that of LSP. It should be noted that the residual stresses in austenite in the Y direction have a sharp tensile region at around 25 µm below the surface. This has not been observed in most surface processing techniques, such as SP,

190

LSP, SMAT, etc. It is hypothesized the tensile residual stresses region is related to the

deformation-induced austenite to martensite phase transformation. Further study is needed to

clarify this point.

Figure 10b shows the residual stresses measured in the martensite phase in both the X

and the Y directions. It can be clearly observed the residual stresses magnitudes are dramatically

different in the X and the Y direction. Also, the residual stresses magnitudes in both the two

directions in the martensite phase are much higher than those measured in the austenite phase.

For example at 5 µm below the surface, the residual stresses measured in the martensite phase in

the X and the Y direction are -748 and -952 MPa, respectively; at the same depth, the residual

stresses measured in the austenite phase in the X and the Y directions are both around -400 MPa.

It should be noted that the residual stresses measured by both martensite phase and austenite

phase are not equi-biaxial.

(a) (b)

Figure 10. In-depth residual stresses measured by (a) martensite and (b) austenite phases.

3.6 Mechanical Behavior after UNSM

In-depth Hardness after UNSM To investigate the effects of martensite phase transformation and grain refinement on material hardness, nano-indentation test on cross-section of the sample was carried out. Figure

191

11a shows representative load-displacement curves at different depths from the UNSM-treated

sample. The through-the-depth change in hardness is shown in Fig. 11b. At the very top surface,

the hardness is as high as 7 GPa. The nanoscale grains, the high volume fraction of the

martensite phase and the high dislocation density contribute to the high hardness. With

increasing distance from the treated surface, the hardness drops quickly. At around 400 µm below the surface, the hardness is around 2.2 GPa, slightly higher than that of the base material.

(a) (b)

Figure 11. (a) representative load-displacement curves from nano-indentation, (b) in-depth indentation hardness from nano-indentation

Stress-strain curves

Material strength and ductility are of interest because they affect component performance in structural applications. Figure 12 compares the engineering stress-strain curves before and after UNSM. The materials yield stress increases from 340 MPa before UNSM to 630 MPa after

UNSM, whereas uniform elongation engineering strain correspondingly decreased from 82% before UNSM to 70% after UNSM. This means that the strength of the 304 SS increases significantly after UNSM without too much loss in ductility. It is known that the strength and ductility of materials are two mutually-exclusive properties: materials with high strength typically have low ductility. In this study, the 304 SS after UNSM processing has yield strength

192 of 630 MPa while still preserving high ductility (uniform elongation of 70%). This is superior to most high strength steels used in engineering.

Figure 12. Engineering stress-strain curves of the samples before and after UNSM treatment

Fatigue performance:

Component fatigue performance affects its long-term stability in structural applications.

As can be seen in Figure 13, the fatigue performance increased significantly after UNSM processing of the samples. For example, with a maximum bending stress of 850 MPa, the fatigue lives of the samples before and after UNSM are 156,312 and 1,200,000 (run out) cycles respectively, which corresponds to a seven times improvement in fatigue life. The fatigue performance improvement is the comprehensive effect of surface compressive residual stress, surface hardening and microstructural changes. The plastically deformed surface (Fig. 3a) results in high surface hardness (Fig. 11b), which presents a higher resistance to crack initiation. In addition, the high magnitude of surface compressive residual stress, Fig. 10 (400 MPa in austenite phase and 850 MPa in martensite phase) also contributes to the improvement. During the fatigue test, the surface compressive residual stress can decrease the magnitude of the effective tensile stress and thus lead to improved fatigue performance. Thus, the fatigue

193 performance improvement in SS 304 by UNSM is a synergistic effect of surface work hardening and surface compressive residual stresses. Enhanced fatigue resistance in Cu with a gradient nanostructure has been reported by Yang and co-workers [46].

Figure 13. S-N curves from 3-point bending fatigue test of the samples before and after UNSM.

4. Discussion

The present study has led to a number of new and interesting findings on the effects of

UNSM processing on the near and sub-surface residual stress distributions and microstructural changes and the effects of these on local and bulk mechanical properties, such as hardness, strength, ductility and fatigue properties of 304 austenitic stainless steel. These results are discussed in detail in the following.

4.1 Generation of deformation twins

During the plastic deformation of stainless steel, dislocation slip and deformation twinning are two competing process. A critical twinning stress (CTS) must be reached for deformation twinning to occur [47]. According to Chen and co-workers [38], the critical equivalent stress (σT) for twinning to occur can be described by:

194

σT = 6.13γ SFE /bp (1)

where γSFE is the stacking fault energy and bp is the Burgers vector of a Shockley partial

€ dislocation. Thus, the critical twinning stress can be calculated to be 584 MPa for 304 stainless

-2 steel (γSFE = 16 mJm , and bp = 0.147 nm). This means that when the critical resolved shear

stress in certain grain reaches 584 MPa, deformation twinning would initiate. Theoretically, the

top surface should have the highest deformation twin density; the twin density should decrease as

it goes deeper into the material. No de formation twinning would occur at the depth at which the

critical resolve shear stress goes below the critical twinning stress.

According to the TEM observation, deformation twins have been observed from 20 µm

to 100 µm below surface, with the 20 µm showing the highest twin density (Fig. 8). However, at

the top surface and 10 µm below surface, very few deformation twins have been observed in the

TEM images, even though very short twin boundaries (the boundaries with 60o misorientation)

have been observed by PED images (Fig. 6c, Fig. 7c). During the UNSM process, deformation

twins may have been generated in the top surface at some intermediate stage (Fig. 14). As plastic

strain accumulates, the dislocations within the twin-matrix (TM) lamellae increase in density,

which results in stress concentration in the TM lamellae, leading to the subdivision of the grains

and thus making the twins disappear. This phenomena has also been observed by Lu [8] and

Zhang [2]. It has also been suggested in quite a few studies [2,8,48] that, in metallic materials

with relatively low SFE, deformation twinning plays an important role in the nanocrystallization

process caused by severe plastic strain. During this process, the original grains were first sub-

divided by mechanical twins; the dislocations generated in-between the TM structure further

subdivided the twin lamellae structure into equiaxed nano-crystallites. This finally leads to the

195 nanocrystalline martensitic microstructure in the top surface to 10 µm below the surface, which makes long deformation twins hard to observe.

At the subsurface, for example 20 µm below the surface, however, the dislocation density generated by plastic strain has not reached the critical point to break the twin boundaries, thus leaving highly dense deformation twins at this depth. Deeper into the material (50 to 100 µm), the strain rate and plastic strain both decrease, leading to relatively lower density of the deformation twins. Thus, the above mentioned mechanisms lead to the occurrence of the maximum twin density at a depth of around 20 µm.

3.1 Deformation-induced surface nanocrystallization

As described in the foregoing, the UNSM process is observed to lead to the formation of nanoscale crystallites at the surface and near-surface regions of the 304 SS material.

Deformation-induced nanocrystallization by bulk SPD as well as by other surface treatment processes has been widely reported in the past [49]. The grain refining mechanisms in SS 304 by plastic deformation have been well discussed by Lu et al. [8], Zhang et al. [2] and Chen et al.

[38]. Deformation-induced grain refinement in SS 304 originates from dislocation activity [50], deformation twinning [51,52] and martensitic phase transformation [53].

During plastic deformation of stainless steel, dislocation slip and deformation twinning are two competing process. Due to the low stacking fault energy (16 mJ/m2) of stainless steel

304, it is hard to form dislocation cells through dislocation cross-slip. Rather, planar dislocation arrays, stacking faults with widely separated partial dislocations and twins are more easily formed on the {111} slip planes [1]. As more plastic strain accumulates, deformation twins in different directions divide the grains into smaller sections and thus generate some rhombic blocks (Fig. 8f), which eventually leads to nanocrystallization through dynamic recrystallization

196

[8]. Further straining induced by the mechanical strikes generates more and more sub-grain boundaries. As more mechanical strikes are imposed during the UNSM process, the refined grains are rotated, generating randomly distributed nano-grains, as manifested by the ring diffraction pattern in Fig. 5. It should be noted that, even after severe mechanical strikes, some localized deformation twins are still preserved, which are manifested by the nanoscale twin boundaries as presented in Fig. 6c and Fig. 7c, where the green lines represents boundaries with

60o misorientation, i.e., twin boundaries. It should be noted that this nanocrystallization only occurs at the top surface and to a depth of 10 µm. Deeper into the material, the magnitude of the plastic deformation decreases to the extent that not enough plastic strain was accumulated to refine the grains to the 10 nm range.

Martensitic phase transformation also contributes to the nanocrystallization process.

Martensitic phase transformation in SS by plastic deformation has been widely studied and reported in the past [2,32,38,54,55]. Martensite phase transformation is a major method to accommodate plastic strain when stainless steels are subjected to SPD. Martensite phase transformation occurs in this sequence: first, plastic deformation introduces defects in the material in the form of dislocations, stacking faults and twins; second, the intersections of the stacking faults and twins generates plastic strain concentration, and thus serve as embryos for martensite formation; finally, the martensite embryos grow to ultrafine crystallites when the material is subjected to further plastic strain. Typically, the martensite phase formed in this way has very small grain sizes. In the present case, martensite with an average grain size of less than

10 nm was observed in the near surface region (Fig. 6e). This mechanism explains the presence of large volume fraction of martensite in the near surface region (to ~20 µm). As the amount of plastic strain decreases at greater depths (50-100 µm), martensite volume fraction also decreases

197 as observed in the XRD (Fig. 1b) and EBSD (Fig. 4) results presented earlier. It should be noted that martensite phase transformation should not be separated with deformation twinning when explaining the nanocrystallization process, instead they work in synergy to produce surface nanocrystallization.

3.2 Strength-Ductility-Microstructure Relationship

Strength:

The large improvement in material strength after UNSM observed in this study (Fig. 12) could be explained as follows. UNSM involves repeated multidirectional mechanical strikes at high speed onto the material surface. This generates a gradient in plastic deformation with distance from the treated surface, and thus a gradient microstructure (Fig. 14). At the very top surface, large and rapid plastic deformation results in extreme grain refinement to the nanoscale.

In the subsurface layer, highly dense deformation twins are generated. In the interior, little plastic deformation occurs and the grain size is largely unaffected. Theoretically, materials strength can be predicted by adding contributions from the reduction in grain size and increase in dislocation density with plastic deformation [37-39]:

(2)

where σf is the strength, σ0 is a friction stress, k is the Hall-Petch constant, dfp is the mean free path for dislocations, α is a constant and G is the shear modulus, b is the burgers vector and

ρ is the dislocation density. We can tell from Eq. 2 that as the dislocation mean free path decreases and the dislocation density increases, the material hardness increases. In this section, the strengthening mechanisms at different layers will be systematically discussed.

Top layer: strengthening by nanoscale martensitic grains. It is widely known that nanoscale grains can significantly improve the hardness and strength of metallic materials. In the

198

SS 304 samples processed by UNSM, nanoscale grains extend from the top surface to 10 µm below the surface. The sizes of the grains are in the range of a few to 10s of nanometers. Due to the existence of high density of grain boundaries, the mean free path for dislocations is significantly reduced and thus the alloy strength is improved significantly (Eq. 3). In addition, the martensite phase, which extends from the top surface to 50 µm below surface and has high strength, also improves strength.

Subsurface: Strengthening by heavily dislocated twin boundaries. In the material’s subsurface regions, highly dense deformation twins exist (Fig. 8 and Fig. 9). For polycrystalline materials without twin boundaries, the mean free path dfp is equal to mean grain size d. For materials with twin boundaries, the mean free path dfp is smaller than the mean grain size due to

the interaction between dislocations and twin boundaries, i.e., dislocations will be blocked by the

discontinuity in the slip systems at the two sides of the symmetrical twin boundaries. This means

the twin boundaries, like the grain boundaries, can also block dislocation movement and thus

improve material strength. As a result, the mean free path of dislocations will be significantly

decreased by the presence of the twin boundaries. The effective free path considering both the

grain boundaries and the twin boundaries can be expressed by [40]:

(3) where dtwin and dsub are the average twin and subgrain boundary distances, respectively. From

Eq. 3, it can be concluded that the existence of very fine scale twins and numerous twin

boundaries in the subsurface regions generated by UNSM can effectively decrease the mean free

path of the dislocations through dislocation-twin interaction and thus improve material strength.

Ductility:

199

Structural applications require material to possess both high strength and high ductility.

The ductility is related to the inhibition of necking and strain localization, which is affected by

material work hardening capacity. Work hardening can effectively prevent localized deformation

and thus delay necking during tensile test [28]. According to the Considère criterion [56,57], the

uniform elongation holds in a tensile test until the onset of the localized deformation, which is

governed by

⎛ ∂σ ⎞ ⎜ ⎟ = σ (4) ⎝ ∂ε ⎠ε ˙

where σ is true stress, ε is true strain and � is the strain rate. From this equation, we can tell that € ductility of the material is essentially determined by its strain hardening capacity, which is the

capacity to accumulate dislocations generated during plastic deformation. This requires, firstly,

dislocation emission sources, and secondly that the dislocations generated can be preserved.

During plastic deformation, dislocation multiplication and annihilation occurs simultaneously. In

metallic materials with coarse grains, dislocation accumulates through lattice dislocation storage.

For example, a classic Frank-Read source can effectively generate dislocations and thus sustain

the plastic deformation [58]. For nano-grained materials, however, most dislocation sources are

not operative [26], and thus cannot supply enough dislocations to sustain the plastic deformation.

In addition, dislocations cannot be properly stored because they tend to disappear at the

nanoscale grain boundaries, which often serve as sinks for dislocation annihilation [30,59,60].

This often leads to low ductility of the nanostructured materials.

However, in the SS 304 processed by UNSM in this study, the strength has been

increased from 340 MPa to 630 MPa while still preserving significant amount of ductility (Fig.

12). To understand why the SS 304 processed by UNSM possesses both high strength and high

ductility, it is first necessary to study its overall microstructure. As presented in Fig. 14a, SS 304

200 after UNSM processing on both sides features a multi-layer (yet integral) metallic component with two strong nanocrystalline surface layers connected by a ductile interior with gradually decreasing martensite volume fraction (Fig. 1b) and increasing grain size. The two nanocrystalline surface layers harden the material and thus significantly improve material strength while the unaffected inner layer provides a ductile interior for dislocation accumulation during tensile deformation. In addition, the highly dense twin boundaries in the subsurface layers can act both as dislocation barriers and dislocation emission sources. It has been reported that the twin boundaries can effectively toughen the materials by providing dislocation nucleation sites through dislocation-TBs interaction [26,61–63]. The interaction between dislocations and the twin boundaries has been intensively studied in the literature [58,64–66].

When a gliding dislocation encounters a twin boundary, stress concentration leads to the generation of new dislocations on the other side of the boundary (Fig. 14b) [26]. This makes the twin boundaries effective sources for dislocation generation, and thus provide efficient dislocation emission sources, toughening the materials [67]. In this way, the deformation twins at material subsurface can serve both as dislocation movement blockers and dislocation emission sources. This means that the twin boundaries first act as dislocation barriers and thus strengthen the materials; at the same time, the twin boundaries can serve as dislocation emission sources and thus provide more mobile dislocations (Fig. 14b) to sustain the plastic deformation, improving material ductility [68].

In addition, in the center of the material (Fig. 14a), the microstructure is essentially unaffected. This way the dislocation accumulation capacity in material interior is preserved. The unaffected interior on the one hand provides the work-hardening capacity, and on the other hand organically bonds the two surface nanocrystalline surface layer and thus prevents premature

201 necking [69,70], leading to both high strength and high ductility. In addition, the gradually changing martensite distribution ensures strong interfacial strength between the ductile interior and the two nanocrystalline surface layers [7]. Finally, the high magnitude of compressive residual stress can also delay crack propagation [71,72] and thus contributes favorably to material ductility. The comprehensive effects of all these factors lead to both high strength and high ductility in the UNSM samples.

(a) (b)

Figure 14. (a) Schematic representation of (a) the sandwiched/gradient microstructure in SS 304 processed by UNSM (redrawn with permission from [32]) and (b) dislocation-twinning interaction (redrawn with permission from [26]).

5.0 Conclusions:

UNSM processing of SS 304 was carried out in this study. The results have shown that this process leads to high levels of near-surface compressive residual stresses and dramatic changes in microstructure and hence local and bulk mechanical properties. After UNSM, the material assumes a multi-layer microstructure with two strong nanocrystalline surface layers and a ductile interior with gradually changing martensite volume fraction. The two nanocrystalline surface layers with high martensite content provide strong resistance to plastic flow and thus lead

202 to high yield strength while the unaffected interior provides strain hardening capacity and thus preserves the high ductility. In the subsurface regions of the material, highly dense deformation twins are generated. These deformation twins significantly improve material hardening capacity by acting both as dislocation blockers and dislocation emission sources, and thus strengthening the material while providing enough strain hardening capacity. Furthermore, the gradually changing martensite volume fraction ensures strong interfacial strength between the ductile interior and the two nanocrystalline surface layers and thus prevents early necking. The unique microstructure with a compliant interior sandwiched by two strong nanocrystalline layers leads to both high material strength and high ductility. In addition, the work-hardened surface layer and the high magnitude of compressive residual stress lead to significantly improvement in fatigue properties. It can be concluded that UNSM is a superior surface processing technique that can generate unique microstructure for improved properties and performance. Finally,

UNSM can be easily integrated into modern manufacturing system and has great potential in the industry.

6.0 Acknowledgements

The authors are grateful for financial support of this research by the Nuclear Energy

Frontier Commission, which provided funding in support of “Ohio Center for Laser Shock

Processing for Advanced Materials and Devices” and the experimental and computational equipment in the Center that was used in this work. The work performed at the University of

Pittsburgh received part support from the National Science Foundation (NSF-DMR-11005757)

203 and the Nuclear Regulatory Commission (NRC-38-09-935). Any opinions, findings, conclusions, or recommendations expressed in these documents are those of the author(s) and do not necessarily reflect the views of the DOE, State of Ohio, Department of Development, the

National Science Foundation and the Nuclear Regulatory Commission.

7.0 References

[1] K. Lu, J. Lu, Mater. Sci. Eng. A 375-377 (2004) 38.

[2] H.W. Zhang, Z.K. Hei, G. Liu, J. Lu, K. Lu, Acta Mater. 51 (2003) 1871.

[3] H.L. Chan, H.H. Ruan, a. Y. Chen, J. Lu, Acta Mater. 58 (2010) 5086.

[4] G. Liu, S.. Wang, X.. Lou, J. Lu, K. Lu, Scr. Mater. 44 (2001) 1791.

[5] G. Liu, J. Lu, K. Lu, Mater. Sci. Eng. A 286 (2000) 91.

[6] N.R. Tao, M.L. Sui, J. Lu, K. Lua, Nanostructured Mater. 11 (1999) 433.

[7] J.Z. Lu, K.Y. Luo, Y.K. Zhang, C.Y. Cui, G.F. Sun, J.Z. Zhou, L. Zhang, J. You, K.M. Chen, J.W. Zhong, Acta Mater. 58 (2010) 3984.

[8] J.Z. Lu, K.Y. Luo, Y.K. Zhang, G.F. Sun, Y.Y. Gu, J.Z. Zhou, X.D. Ren, X.C. Zhang, L.F. Zhang, K.M. Chen, C.Y. Cui, Y.F. Jiang, a. X. Feng, L. Zhang, Acta Mater. 58 (2010) 5354.

[9] C. Ye, Y. Liao, G.J. Cheng, Adv. Eng. Mater. 14 (2010) 291.

[10] C. Ye, S. Suslov, B.J. Kim, E. a. Stach, G.J. Cheng, Acta Mater. 59 (2011) 1014.

[11] A.S. Gill, Z. Zhou, U. Lienert, J. Almer, D.F. Lahrman, S.R. Mannava, D. Qian, V.K. Vasudevan, J. Appl. Phys. 111 (2012) 084904.

[12] A. Gill, A. Telang, S.R. Mannava, D. Qian, Y.-S. Pyoun, H. Soyama, V.K. Vasudevan, Mater. Sci. Eng. A 576 (2013) 346.

[13] J.Z. Lu, H. Qi, K.Y. Luo, M. Luo, X.N. Cheng, Corros. Sci. 80 (2014) 53.

[14] K.Y. Luo, J.Z. Lu, Q.W. Wang, M. Luo, H. Qi, J.Z. Zhou, Appl. Surf. Sci. 285 (2013) 607.

[15] Z.B. Wang, J. Lu, K. Lu, Surf. Coatings Technol. 201 (2006) 2796.

204

[16] J.Z. Lu, K.Y. Luo, F.Z. Dai, J.W. Zhong, L.Z. Xu, C.J. Yang, L. Zhang, Q.W. Wang, J.S. Zhong, D.K. Yang, Y.K. Zhang, Mater. Sci. Eng. A 536 (2012) 57.

[17] S. Jelliti, C. Richard, D. Retraint, T. Roland, M. Chemkhi, C. Demangel, Surf. Coatings Technol. 224 (2013) 82.

[18] S. Faghihi, a. P. Zhilyaev, J. a. Szpunar, F. Azari, H. Vali, M. Tabrizian, Adv. Mater. 19 (2007) 1069.

[19] Y.H. Zhao, X.Z. Liao, S. Cheng, E. Ma, Y.T. Zhu, Adv. Mater. 18 (2006) 2280.

[20] R. Song, D. Ponge, D. Raabe, Acta Mater. 53 (2005) 4881.

[21] R. Song, D. Ponge, D. Raabe, Scr. Mater. 52 (2005) 1075.

[22] Z. Horita, K. Ohashi, T. Fujita, K. Kaneko, T.G. Langdon, Adv. Mater. 17 (2005) 1599.

[23] S. Cheng, Y.H. Zhao, Y.T. Zhu, E. Ma, Acta Mater. 55 (2007) 5822.

[24] W.W. Jian, G.M. Cheng, W.Z. Xu, H. Yuan, M.H. Tsai, Q.D. Wang, C.C. Koch, Y.T. Zhu, S.N. Mathaudhu, Mater. Res. … 1 (2013) 61.

[25] H.T. Wang, N.R. Tao, K. Lu, Scr. Mater. (2012) 1.

[26] K. Lu, L. Lu, S. Suresh, Sci. 324 (2009) 349.

[27] L. Lu, X. Chen, X. Huang, K. Lu, Sci. 323 (2009) 607.

[28] T.H. Fang, W.L. Li, N.R. Tao, K. Lu, Science 331 (2011) 1587.

[29] Y. Wang, M. Chen, F. Zhou, E. Ma, Nature 419 (2002) 912.

[30] Y.M. Wang, E. Ma, Acta Mater. 52 (2004) 1699.

[31] K. Edalati, S. Toh, H. Iwaoka, M. Watanabe, Z. Horita, D. Kashioka, K. Kishida, H. Inui, Scr. Mater. 67 (2012) 814.

[32] A.Y. Chen, H.H. Ruan, J.B. Zhang, X.R. Liu, J. Lu, Mater. Chem. Phys. 129 (2011) 1096.

[33] P. V Liddicoat, X.-Z. Liao, Y. Zhao, Y. Zhu, M.Y. Murashkin, E.J. Lavernia, R.Z. Valiev, S.P. Ringer, Nat. Commun. 1 (2010) 63.

[34] K. Tai, M. Dao, S. Suresh, A. Palazoglu, C. Ortiz, Nat. Mater. 6 (2007) 454.

[35] R.K. Nalla, J.H. Kinney, R.O. Ritchie, Nat. Mater. 2 (2003) 164.

205

[36] S. Kamat, X. Su, R. Ballarini, A.H. Heuer, Nature 405 (2000) 1036.

[37] B.J.F. Bruet, J. Song, M.C. Boyce, C. Ortiz, Nat. Mater. 7 (2008) 748.

[38] A.Y. Chen, H.H. Ruan, J. Wang, H.L. Chan, Q. Wang, Q. Li, J. Lu, Acta Mater. 59 (2011) 3697.

[39] Y. Pyoun, I. Cho, C. Suh, J. Park, Shotpeener.com (n.d.) 2.

[40] A. Cherif, Y. Pyoun, B. Scholtes, J. Mater. Eng. Perform. 19 (2009) 282.

[41] X.J. Cao, Y.S. Pyoun, R. Murakami, Appl. Surf. Sci. 256 (2010) 6297.

[42] A. Amanov, O. V. Penkov, Y.-S. Pyun, D.-E. Kim, Tribol. Int. 54 (2012) 106.

[43] H. Huang, J. Ding, P.G. McCormick, Mater. Sci. Eng. A 216 (1996) 178.

[44] B.D. Cullity, Elements of X-Ray Diffraction, 2nd ed., Addison-Wesley Publishing Company, Inc., Reading, MA, 1978.

[45] Y. Idell, G. Facco, a. Kulovits, M.R. Shankar, J.M.K. Wiezorek, Scr. Mater. 68 (2013) 667.

[46] L. Yang, N.R. Tao, K. Lu, L. Lu, Scr. Mater. 68 (2013) 801.

[47] J.W. Christian, S. Mahajan, Prog. Mater. Sci. 39 (1995) 1.

[48] K. Wang, N.R. Tao, G. Liu, J. Lu, K. Lu, Acta Mater. 54 (2006) 5281.

[49] R. Valiev, Nat Mater 3 (2004) 511.

[50] Y. Iwahashi, Z. Horita, M. Nemoto, T.G. Langdon, Acta Mater. 46 (1998) 3317.

[51] S. Qu, X.H. An, H.J. Yang, C.X. Huang, G. Yang, Q.S. Zang, Z.G. Wang, S.D. Wu, Z.F. Zhang, Acta Mater. 57 (2009) 1586.

[52] Y.S. Li, N.R. Tao, K. Lu, Acta Mater. 56 (2008) 230.

[53] C.X. Huang, Y.L. Gao, G. Yang, S.D. Wu, G.Y. Li, S.X. Li, J. Mater. Res. 21 (2006) 1687.

[54] C. Ye, S. Suslov, D. Lin, G.J. Cheng, Philos. Mag. 92 (2012) 1369.

[55] J. Fan, T. Fu, Mater. Sci. Eng. A 552 (2012) 359.

[56] G. Dieter, Mechanical Metallurgy, 3rd ed., McGraw-Hill Science, New York, 1986.

206

[57] Y.M. Wang, E. Ma, Mater. Sci. Eng. A 375–377 (2004) 46.

[58] Z.X. Wu, Y.W. Zhang, D.J. Srolovitz, Acta Mater. 57 (2009) 4508.

[59] M.A. Meyers, A. Mishra, D.J. Benson, Prog. Mater. Sci. 51 (2006) 427.

[60] E. Ma, Scr. Mater. 49 (2003) 663.

[61] M. Dao, L. Lu, Y.F. Shen, S. Suresh, Acta Mater. 54 (2006) 5421.

[62] X. Li, Y. Wei, L. Lu, K. Lu, H. Gao, Nature 464 (2010) 877.

[63] H. Zhou, S. Qu, W. Yang, Model. Simul. Mater. Sci. Eng. 18 (2010) 65002.

[64] Y.T. Zhu, X.L. Wu, X.Z. Liao, J. Narayan, L.J. Kecskés, S.N. Mathaudhu, Acta Mater. 59 (2011) 812.

[65] Z.-H. Jin, P. Gumbsch, K. Albe, E. Ma, K. Lu, H. Gleiter, H. Hahn, Acta Mater. 56 (2008) 1126.

[66] Z.-H. Jin, P. Gumbsch, E. Ma, K. Albe, K. Lu, H. Hahn, H. Gleiter, Scr. Mater. 54 (2006) 1163.

[67] R.J. Asaro, S. Suresh, Acta Mater. 53 (2005) 3369.

[68] Y.B. Wang, M.L. Sui, Appl. Phys. Lett. 94 (2009) 021909.

[69] N. Lu, X. Wang, Z. Suo, J. Vlassak, J. Mater. Res. 24 (2009) 379.

[70] Y. Xiang, T. Li, Z. Suo, J.J. Vlassak, Appl. Phys. Lett. 87 (2005).

[71] D.J. Green, R. Tandon, V.M. Sglavo, Sci. 283 (1999) 1295.

[72] M.P. Rao, A.J. Sánchez-Herencia, G.E. Beltz, R.M. McMeeking, F.F. Lange, Sci. 286 (1999) 102.

207

11.0 Effects of Ultrasonic Nanocrystal Surface Modification on the Residual Stress, Microstructure and Stress Corrosion Cracking Behavior of SS304 Welds

Chang Ye1*, Abhishek Telang2, Amrinder Gill2, S. R. Mannava2, D. Qian3, Vijay K. Vasudevan2 1Department of Mechanical Engineering, University of Akron, Akron OH 44325-3903 2Department of Mechanical Engineering and Material Science, University of Cincinnati, Cincinnati OH 45221-0072 3Department of Mechanical Engineering, University of Texas, Richardson, TX 75080-3021 Published in the Proceedings of the 16th International Conference on Environmental Degradation of Materials in Nuclear Power Systems- Water Reactors, Asheville, NC, 2013. Abstract In this study, Ultrasonic Nano-crystal Surface Modification (UMSM) of welded stainless steel 304 samples was carried out. On the top surface, UNSM effectively eliminates the tensile stress generated by the welding process and imparts beneficial compressive residual stress. In addition, UNSM can effectively refine the grains and increase hardness in the near surface region. SCC testing in boiling MgCl2 solution shows that UNSM can significantly improve resistance to SCC due to the compressive residual stresses and changes in the near surface microstructure. Keywords: Ultrasonic nanocrystal surface modifications, stainless steel, welds, residual stress, stress corrosion cracking. 1. Introduction

Austenitic stainless steels (SS) and their welds are widely used due to a good combination of mechanical properties, weldability and corrosion resistance. However, SS 304 and its weldments are known to be susceptible to chloride stress corrosion cracking (SCC) [1],

[2]. SCC occurs when the following three factors are present: (1) a corrosive environment, (2) a susceptible material and (3) tensile stress. Austenitic stainless steels are susceptible to SCC failure. A number of studies in the past few years have been dedicated to understand the effect of environment [3], surface finish and residual stresses [1] on the SCC failure of austenitic stainless steels.

1 Corresponding author email: [email protected]

208

Welding induces tensile stresses in materials. These stresses can be deleterious to the corrosion and SCC resistance as observed in the literature. Advanced mechanical surface treatment techniques have been investigated and used to modify the nature of the stresses.

Compressive residual stresses can be induced in the near surface region by shot peening (SP) [4], laser shock peening (LSP) [5]–[8], low plasticity burnishing (LPB) [9], etc. Thus, eliminating tensile stress or reducing the stress level to below the threshold for SCC would be effective in preventing SCC [10].

The ultrasonic nanocrystal surface modification (UNSM) system [11], [12] utilizes ultrasonic vibration energy and mechanical strikes to induce severe plastic deformation (SPD) that imparts compressive residual stress to material surface. UNSM system has been successfully used to process carbon steel [12], [13], stainless steel [14], NiTi [15], magnesium alloys [16] for improved properties and performance. Considering that UNSM can generate beneficial compressive residual stress, which can potentially increase SCC resistance, the aim of this study is to investigate the effect of UNSM on the microstructure, residual stresses and SCC resistance of welded SS304.

In this study, stainless steel (SS) 304 welds were treated by UNSM. Residual stresses and changes in the near surface microstructure after UNSM were characterized. In addition, we investigated the effects of UNSM on the SCC resistance of the SS304 welds in boiling MgCl2 solution.

2. Experimental methods

2.1 Materials

Samples were cut by electric discharge machining (EDM) from two plates (thickness 1.8 mm and 2.8 mm) of AISI 304 stainless steel, the composition of which is 0.08 C, 1.00 Si, 2.00

209

Mn, 0.045 P, 0.03 S, 18.0-20.0 Cr, 8.0-10.5 Ni, and balance Fe (all wt.%). SS 304 plates were welded with a single pass butt joint by using gas tungsten arc welding (GTAW) by a commercial sheet metal welding company. SS 308L was used as filler material.

2.2 UNSM treatment

In the UNSM system, a tungsten carbide ball attached to an ultrasonic device scans over material surface while striking the material surface at high frequency (10K to 30K Hz). During the strike, the depth that the tungsten carbide ball moves into the target material is called the amplitude, typically ranging from 10 microns to 40 microns. At the same time, a static load

(typically 10 to 50 N) is applied to the ball against the material surface. The parameters in a

UNSM process include: the static load, the amplitude of the strike, the scan speed, the intervals between neighboring scans (Fig. 1b), and the ultrasonic peening frequency. In this study, the

UNSM experiment was carried out by an LM 520 UNSM system. The static load was 20 N, the horn tip diameter was 2.38 mm, the amplitude was 10 microns, the frequency was 20 kHz, the interval was 70 µm and the scanning speed was 3000 mm/minute.

Figure 1. (a) Schematic setup of the UNSM experiment, (b) Scan direction of the UNSM process, (the intervals were exaggerated for illustration purpose)

210

2.3 Characterizations:

A number of characterization methods were used to characterize the material properties

after UNSM processing as below.

Micro-hardness: The hardness change of the samples was measured by a micro-

indentation system with a Berkovich indenter with a maximal load of 500 mN and 10 seconds

holding time. An average of five measurements was used for each reported data point.

Microstructure observation: to observe the microstructure after processing, the material

cross-section was etched using 25 ml HCl + 20 ml CH3OH + 15 ml HNO3 solution. Optical observation was carried out using a Keyence VX-600 digital optical microscope. A XL-30 scanning electron microscope (SEM) was also used to observe the microstructure. Electron back scattered diffraction (EBSD) was performed using the TSL OIM system with a Hikari camera.

Residual Stress: Residual stresses were characterized by peak shift using a Proto LXRD system using MnKα (311) for austenite phase. The X-ray elastic constant S2/2 used was

7.18*10-6 MPa-1 for the austenite phase. The size of the X-ray beam is 1 mm. Before the measurement, the Proto LXRD system was calibrated by standard samples. In-depth residual stress was measured by electropolishing the material layer by layer. The electropolishing solution was composed of 12.5 percent (volume) sulfuric acid and 87.5 percent methanol.

TEM: The TEM samples were prepared from the top surface of the base material by grinding and ion-milling. TEM work was carried out by a Philips/FEI CM20 operated at 200 kV.

2.4 SCC tests

Samples with a dimension of 50 mm x 10 mm were sectioned from the weld plate using wire EDM. Samples were then exposed to boiling MgCl2 solution for 120 hours or longer as per

ASTM G44 standard. A few samples were UNSM treated and then tested in the same

211 environment to investigate the effect of UNSM on the SCC behavior. After 120 hours or failure

(visual inspection with optical microscope), samples were characterized using SEM and EBSD.

3. Results and Discussion

In this section, we present the effect of UNSM on the microstructure, residual stresses and SCC behavior. Further, we discuss these results in the context of use of mechanical surface treatments to mitigate SCC in SS and its weldments.

3.1 Microstructure evolution across the welding zone before and after UNSM

In this study, optical microscope and SEM were used to characterize the microstructure across the welding zone before and after UNSM. Fig. 2 compares the microstructure of the SS

304 welds before and after UNSM. Fig. 2a and 2b show the microstructure of the welds before

UNSM. A transition area between the weld metal and the base metal can be clearly observed.

Figs. 2c, 2d and 2e show the microstructure across the welding zone after UNSM. In the base material after UNSM, deformation bands can be observed. Deformation bands can also be observed in the welding zone, as pointed by the green arrow in Fig. 2e. Because of the heterogeneous microstructure in the welding zone, the deformation band morphology is not as clear as that in the base material. Fig. 3 shows the inverse pole figure map of the weld zone and base metal. In both the weld zone and the base metals, random orientation can be observed.

Fig. 4 shows the microstructure observed by TEM in the base material after UNSM.

Nanoscale grains can be observed. According to the optical images in Fig. 2a, the initial grain size is around 30 µm. Thus, significant grain refinement has been achieved through UNSM.

During SCC, the cracks propagation can be effectively arrested by triple junctions in grain [17]–

[20]. Grain refinement can significantly increase the number of grain boundaries and thus the number of triple junctions, making the propagation of the SCC cracks harder, leading to higher

212

SCC resistance. On the other hand, the dense grain boundaries and triple junctions in the nanocrystalline material can also serve as diffusion channels for corrosion media and thus decrease the corrosion resistance [21]. Further study is needed to investigate the effect of the nanocrystalline surface layer generated by UNSM on material corrosion behavior.

Weld toe

Weld metal

Base metal (a) (b)

Weld toe

Base metal Weld metal

(e)

Figure 2. Microstructure across the welding zone before (a and b) and after (c, d and e) UNSM.

213

Figure 3. Inverse pole figure map showing the weld zone and base metal.

Figure 4. Microstructure observed by TEM in the base material after UNSM

3.2 Residual stresses

The residual stresses before and after UNSM were measured by XRD using the sin2ψ

method. Fig. 5a shows the surface residual stress across the welding zone before and after

UNSM. Tensile stresses exist across the welding zone and were generated during the welding

process. It can be clearly observed that the residual stress changed from 100 to 500 MPa in

214

tension before UNSM to 700 to 1300 MPa in compression after UNSM. Also, it should be noted

that the residual stresses before and after UNSM assume a very similar pattern, i.e., the regions

have high tensile stress before UNSM also have low compressive stress after UNSM; the regions

have low tensile stress before UNSM have high compressive stress after UNSM. The highest

magnitude of compressive residual stress in the sample after UNSM exists at around 20 mm

away from the center of the welding zone. This is because there exists lowest magnitude of tensile residual stress at 20 mm away from the center of the welding zone in the as-weld sample.

Due to the symmetry of the residual stress field across the welding zone (Fig. 5a), the in- depth residual stresses were measured only on one side. A three-dimensional (3D) representation of the in-depth residual stress across the welding zone is shown in Fig. 5b. As we can clearly observe, the magnitude of the compressive residual stress across the welding zone decreases gradually as it goes deeper into the materials. It should be noted that at even 250 µm below the surface, the residual stress is compressive at around 200 to 300 MPa. The existence of compressive residual stress can potentially improve the SCC resistance of the material.

(a)

215

(b)

Figure 5. (a) Surface residual stress across the welding zone before and after UNSM; (b) 3D residual stress distribution in the welding zone after UNSM 3.3 Hardness across the welding zone before and after UNSM

Fig. 6 compares the in-depth hardness in the weld before and after UNSM. The surface hardness increases from around 210 HV to 400 HV. The hardness decreases gradually as it goes deeper into the welding zone, and gets close to the as-weld sample at around 400 µm below surface. The hardness improvement is caused by the grain refinement and the introduction of dislocations through plastic deformation.

Figure 6. In-depth micro-hardness of the welding zone before and after UNSM

216

3.4 SCC tests

SCC tests were performed as per ASTM G44 standard to evaluate the effects of UNSM

on the SCC resistance. In the as-welded condition, cracks were observed on the sample surface

after 72 hours and the tests were then terminated. Fig. 7 (a-d) show cracks on the sample after

testing in boiling MgCl2 solution for 72 hours. Cracks were observed to initiate from the surface close to the weld toe and propagate through the sample thickness in 72 hours. This indicates that the welds were susceptible to chloride SCC. Fig. 8 shows an SEM image of the crack and the corresponding IPF map. From the IPF map, we can clearly observe the transgranular nature of cracks in the as-welded SS 304 sample. Similar tests were also performed on SS304 weld samples after UNSM treatment. These samples did not fail even after 120 hours of testing.

Figure 7. (a-d) SEM micrographs showing locations of cracks on the as-welded SS304 sample after testing in boiling MgCl2 solution

217

Figure 8. SEM micrograph (a) and corresponding IPF map (b) showing the transgranular nature of the crack in the as-welded SS 304 sample after corrosion test, grain boundaries are colored black 4. Conclusions UNSM processing of SS 304 welds was carried out in this study. It has been demonstrated that UNSM can effectively introduce compressive residual stress in the weld region. Also, fine grained microstructure can be observed in the surface region after UNSM.

After UNSM, material hardness has been significantly improved. SCC test results demonstrate that UNSM processing significantly improved the SCC resistance of the SS 304 welds.

5. Acknowledgements The authors are grateful for financial support of this research by the Nuclear Energy

Frontier Commission, which provided funding in support of “Ohio Center for Laser Shock

Processing for Advanced Materials and Devices” and the experimental and computational equipment in the Center that was used in this work. Any opinions, findings, conclusions, or

218 recommendations expressed in these documents are those of the author(s) and do not necessarily reflect the views of the DOE, State of Ohio, Department of Development.

6. References [1] S. Ghosh and V. Kain, “Microstructural changes in AISI 304L stainless steel due to surface machining: Effect on its susceptibility to chloride stress corrosion cracking,” J. Nucl. Mater., vol. 403, no. 1–3, pp. 62–67, Aug. 2010. [2] O. M. Alyousif and R. Nishimura, “The stress corrosion cracking behavior of austenitic stainless steels in boiling magnesium chloride solutions,” Corros. Sci., vol. 49, no. 7, pp. 3040–3051, 2007. [3] O. M. Alyousif and R. Nishimura, “The effect of test temperature on SCC behavior of austenitic stainless steels in boiling saturated magnesium chloride solution,” Corros. Sci., vol. 48, no. 12, pp. 4283–4293, 2006. [4] U. Zupanc and J. Grum, “Effect of pitting corrosion on fatigue performance of shot-peened aluminium alloy 7075-T651,” J. Mater. Process. Technol., vol. 210, no. 9, pp. 1197–1202, 2010. [5] P. Peyre, C. Braham, J. Ledion, L. Berthe, and R. Fabbro, “Corrosion Reactivity of Laser- Peened Steel Surfaces,” J. Mater. Eng. P, vol. 9, no. December, pp. 656–662, 2000. [6] J. Z. Lu, K. Y. Luo, D. K. Yang, X. N. Cheng, J. L. Hu, F. Z. Dai, H. Qi, L. Zhang, J. S. Zhong, Q. W. Wang, and Y. K. Zhang, “Effects of laser peening on stress corrosion cracking (SCC) of ANSI 304 austenitic stainless steel,” Corros. Sci., vol. 60, pp. 145–152, Jul. 2012. [7] C. Ye, S. Suslov, B. J. Kim, E. a. Stach, and G. J. Cheng, “Fatigue performance improvement in AISI 4140 steel by dynamic strain aging and dynamic precipitation during warm laser shock peening,” Acta Mater., vol. 59, no. 3, pp. 1014–1025, Feb. 2011. [8] C. Ye, Y. Liao, S. Suslov, D. Lin, and G. J. Cheng, “Ultrahigh dense and gradient nano- precipitates generated by warm laser shock peening for combination of high strength and ductility,” Mater. Sci. Eng. A, vol. 609, no. 0, pp. 195–203, Jul. 2014. [9] P. S. Prevéy and J. T. Cammett, “The influence of surface enhancement by low plasticity burnishing on the corrosion fatigue performance of AA7075-T6,” Int. J. Fatigue, vol. 26, no. 9, pp. 975–982, 2004. [10] C. Cheung, U. Erb, and G. Palumbo, “Application of grain boundary engineering concepts to alleviate intergranular cracking in Alloys 600 and 690,” Mater. Sci. Eng. A, vol. 185, no. 1–2, pp. 39–43, 1994. [11] Y. Pyoun, I. Cho, C. Suh, and J. Park, “Nanocrystal Surface Modification) Technology for Prolonging the Service life of AISI 1045 Shear Pin in the Flange Yoke Assembly of Stainless Hot Rolling Mill,” shotpeener.com, pp. 2–6. [12] A. Cherif, Y. S. Pyoun, and B. Scholtes, “Effects of Ultrasonic Nanocrystal Surface Modification (UNSM) on Residual Stress State and Fatigue Strength of AISI 304,” J. Mater. Eng. Perform., vol. 19, no. 2, pp. 282–286, May 2009.

219

[13] X. J. Cao, Y. S. Pyoun, and R. Murakami, “Fatigue properties of a S45C steel subjected to ultrasonic nanocrystal surface modification,” Appl. Surf. Sci., vol. 256, no. 21, pp. 6297– 6303, Aug. 2010. [14] C. Ye, A. Telang, A. S. Gill, S. Suslov, Y. Idell, K. Zweiacker, J. M. K. K. Wiezorek, Z. Zhou, D. Qian, S. R. Mannava, V. K. Vasudevan, and S. Ramaiah Mannava, “Gradient nanostructure and residual stresses induced by Ultrasonic Nano-crystal Surface Modification in 304 austenitic stainless steel for high strength and high ductility,” Mater. Sci. Eng. A, vol. 613, pp. 274–288, Jul. 2014. [15] C. Ye, X. Zhou, A. Telang, H. Gao, Z. Ren, H. Qin, S. Suslov, A. S. Gill, S. R. Mannava, D. Qian, G. L. Doll, A. Martini, N. Sahai, and V. K. Vasudevan, “Surface amorphization of NiTi alloy induced by Ultrasonic Nanocrystal Surface Modification for improved mechanical properties,” J. Mech. Behav. Biomed. Mater., vol. 53, pp. 455–462, Jan. 2016. [16] A. Amanov, O. V. Penkov, Y.-S. Pyun, and D.-E. Kim, “Effects of ultrasonic nanocrystalline surface modification on the tribological properties of AZ91D magnesium alloy,” Tribol. Int., vol. 54, pp. 106–113, Oct. 2012. [17] R. Jones and V. Randle, “Sensitisation behaviour of grain boundary engineered austenitic stainless steel,” Mater. Sci. Eng. A, vol. 527, no. 16–17, pp. 4275–4280, Jun. 2010. [18] H. Kokawa, M. Shimada, M. Michiuchi, Z. J. Wang, and Y. S. Sato, “Arrest of weld-decay in 304 austenitic stainless steel by twin-induced grain boundary engineering,” Acta Mater., vol. 55, no. 16, pp. 5401–5407, Sep. 2007. [19] M. Shimada, H. Kokawa, Z. J. Wang, Y. S. Sato, and I. Karibe, “Optimization of grain boundary character distribution for intergranular corrosion resistant 304 stainless steel by twin-induced grain boundary engineering,” Acta Mater., vol. 50, no. 9, pp. 2331–2341, May 2002. [20] M. Michiuchi, H. Kokawa, Z. J. Wang, Y. S. Sato, and K. Sakai, “Twin-induced grain boundary engineering for 316 austenitic stainless steel,” Acta Mater., vol. 54, no. 19, pp. 5179–5184, Nov. 2006. [21] A. Q. Lü, Y. Zhang, Y. Li, G. Liu, Q. H. Zang, and C. M. Liu, “Effect of nanocrysatlline and twin boundary on corrosion behavior os 316L stainless steel using SMAT,” Acta Metall. Sin. (English Lett., vol. 19, no. 3, pp. 183–189, 2006.

220

12.0 Directions for Future Work

Iterative UNSM and strain annealing has been shown to modify the GBCD in Alloy 600.

This surface GBE treatment was shown to have improved the SCC resistance of Alloy 600 in tetrathionate solution. Currently, U-bend SCC tests are being conducted in simulated PWR/BWR environments to evaluate the SCC resistance of Surface GBE alloy 600. Some other studies that may be of interest are listed below:

• GBE has also been shown to improve resistance to sensitization and hence the effect of

GBE on weld induced sensitization and cracking would be interesting to evaluate.

• Application of SGBE and conventional GBE treatments to improve high temperature

creep and fatigue resistance of low stacking fault energy alloys.

• Studies to evaluate the effectiveness of LSP, UNSM and GBE in Alloy 600, 690, SS316L

on improving oxidation and irradiation resistance and in mitigation of irradiation assisted

SCC.

• Investigations on the effectiveness of LSP, UNSM and SGBE as post weld treatment to

mitigate SCC in austenitic SS316L and Alloy 600, 690, as well as dissimilar alloy

weldments.

221

13.0 Publications

Published

1. A. Telang, A. S. Gill, M. Kumar, S. Teysseyre, D. Qian S R. Mannava and V. K. Vasudevan, “Iterative Thermomechanical Processing of Alloy 600 for Improved Corrosion and Stress Corrosion Cracking Resistance,” Acta Mater., 113, 180-193 (2016); http://dx.doi.org/10.1016/ j.actamat.2016.05.009 2. A. Telang, A. S. Gill, D. Tammana, X. Wen, M. Kumar, S. Teysseyre, S. R. Mannava, D. Qian and V. K. Vasudevan, “Surface Grain Boundary Engineering of Alloy 600 for Improved Resistance to Stress Corrosion Cracking,” Mater. Sci. Engr., A648, 280–288 (2015); http://dx.doi.org/10.1016 /j.msea.2015.09.074 3. A. S. Gill, A. Telang and V. K. Vasudevan, “Characteristics of Surface Layers Formed on Inconel 718 by Laser Shock Peening With and Without a Protective Coating,” J. Mater. Proc. Tech., 225, 463-472 (2015); http://dx.doi.org/10.1016/j.jmatprotec.2015.06.026 4. A. Telang, A. S. Gill, S. Teysseyre, S. R. Mannava, D. Qian and V. K. Vasudevan,” Effects of Laser Shock Peening on SCC Behavior of Alloy 600 in Tetrathionate Solution, Corrosion Science, 90, 434-444 (2015); http://dx.doi.org/10.1016/j.corsci.2014.10.045. 5. C. Ye, A. Telang, A. S. Gill, S. R. Mannava, D. Qian and V. K. Vasudevan, “Gradient Nanostructure by Ultrasonic Nanocrystal Surface Modification for Enhanced Strength, Ductility and Fatigue Resistance of 304 Austenitic Stainless Steel,” Mater. Sci. Engr., A673, 274-288 (2014); http://dx.doi.org/10.1016/ j.msea.2014.06.114 6. A. Telang, C. Ye, A. S. Gill, S. Teysseyre, S. R. Mannava, D. Qian and V. K. Vasudevan “Effects of Laser Shock Peening on SCC Behavior of Alloy 600,” in: Procs. of 16th International Conference on Environmental Degradation of Materials, Wiley (2014). 7. C. Ye, A. S. Gill, A. Telang, Z. Zhou, D. Qian, S. R. Mannava and V. K. Vasudevan “Effect of Ultrasonic Nanocrystal Surface Modification on Residual Stress, Microstructue and Local Properties of 304 Stainless Steel Welds,” in: Procs. of 16th International Conference on Environmental Degradation of Materials, Wiley (2014).

Under Review/In Preparation

8. A. Telang, A. S. Gill, K. Zweiacker, C. Liu, J. M. K. Wiezorek, M. Kumar, S. Teysseyre, S. R. Mannava, D. Qian, and V. K. Vasudevan, “Effect of Thermomechanical Processing on Sensitization and Corrosion of Alloy 600 Studied by SEM and TEM-Based Diffraction and Orientation Imaging Techniques,” J. Nuclear Mater., Submitted (2016). 9. C. Ye, A. Telang, A. Gill, X. Wen, S. R. Mannava, D. Qian and V. K. Vasudevan, “Effects of Ultrasonic Nanocrystal Surface Modification on Residual Stress, Microstructure and Corrosion Resistance of 304 Stainless Steel Welds,” Surfaces & Coatings, Submitted (2017).

222

10. A. Telang, S. Teysseyre, S. R. Mannava, D. Qian, and V. K. Vasudevan, “Effect of Laser Shock Peening on Stress Corrosion Cracking Behavior of Alloy 600 in Simulated High Temperature Water,” Corrosion Science, In Preparation (2017). 11. A. Telang, S. Teysseyre, S. R. Mannava, D. Qian, and V. K. Vasudevan, “In situ Synchrotron X-ray Diffraction Study of the Effects of Mechanical and Thermal Loading on Residual Stress Relaxation Behavior in Surface-Treated Alloy 600,” J. Appl. Phys., In Preparation (2017).

14.0 Presentations

1. A. Telang, A. S. Gill, M. Kumar, S. Teysseyre, D. Qian S R. Mannava and V. K. Vasudevan, “Bulk and Surface Grain Boundary Engineering for Improved Resistance to Corrosion and Stress Corrosion Cracking Resistance of Nuclear Alloys,” Conference held in honor of Professor Ranganathan’s 70th birthday, Indian Institute of Science, Bangalore, India, June 20-22 (2016). Invited

2. A. Telang, A. S. Gill, M. Kumar, S. Teysseyre, D. Qian S R. Mannava and V. K. Vasudevan “Bulk and Surface Grain Boundary Engineering for Improved Resistance to Corrosion and Stress Corrosion Cracking Resistance of Nuclear Alloys Studied by Electron Microscopy,” EMSI 2016: Electron Microscopy Society of India Conference, Banaras Hindu University, Banaras, India, June 2-4 (2016). Invited.

3. A. S. Gill, A. Telang, C. Ye, Z. Zhou, S. R. Mannava, D. Qian and V. K. Vasudevan, “Microstructure, Residual Stress and Property Changes in Metallic Alloys Induced by Advanced Mechanical Surface Treatments,” 5th International Conference on Laser Peening and Related Phenomena, University of Cincinnati, Cincinnati, OH, May 10-15 (2015).

4. A. Telang, S. Teysseyre, X. Wen, S. R. Mannava, D. Qian and V. K. Vasudevan, “Effects of Laser Shock Peening on SCC Behavior of Alloy 600 in Tetrathionate Solution and High Temperature Water,” 5th International Conference on Laser Peening and Related Phenomena, University of Cincinnati, Cincinnati, OH, May 10-15 (2015).

5. A. Telang, A. S. Gill, S. Teysseyre, S. R. Mannava, D. Qian and V. K. Vasudevan, “Effects of Laser Shock Peening on SCC Behavior of Alloy 600 in Tetrathionate Solution, MS&T, Pittsburgh, Oct (2014).

6. A. S. Gill, A. Telang, M. Miller, T. Ungar, G. Eggeler, H. Soyama, Y-S. Pyun, S. R. Mannava, D. Qian and V. K. Vasudevan, “Characterization of Near-Surface Microstructure of Surface-Treated IN718 Superalloy by X-Ray Diffraction and TEM ,” Symposium on Advanced Techniques for Characterization of Plasticity, TMS Annual Meeting, San Diego, CA, Feb (2014).

7. A. S. Gill, A. Telang, C. Ye, Z. Zhou, G. Ramakrishnan, Y. Zhao, S. Bhamare, H. Soyama, Y-S. Pyun, S. R. Mannava, D. Qian and V. K. Vasudevan, “Microstructure and Property Changes in Metallic Alloys Induced by Advanced Mechanical Surface Treatments,” Symposium on Phase Transformations, TMS Annual Meeting, San Diego, CA, Feb (2014).

223

8. A. Telang, C. Ye, A. S. Gill, S. Teysseyre, S. R. Mannava, D. Qian and V. K. Vasudevan “Effects of Laser Shock Peening on SCC Behavior of Alloy 600,” 16th International Conference on Environmental Degradation of Materials, Asheville, NC (2013).

9. C. Ye, A. S. Gill, A. Telang, Z. Zhou, D. Qian, S. R. Mannava and V. K. Vasudevan “Effect of Ultrasonic Nanocrystal Surface Modification on Residual Stress, Microstructue and Local Properties of 304 Stainless Steel Welds,” 16th International Conference on Environmental Degradation of Materials, Orlando, FL, March (2013).

10. A. Telang, A. S. Gill, S. Teysseyre, J. Jackson, J. Guenes, S. R. Mannava, D. Qian and V. K. Vasudevan “Effects of Laser Shock Peening on Residual Stress, Microstructure and Corrosion Behavior of Alloy 600,” TMS Annual Meeting Symposium on Materials and Fuels for Current and Future Nuclear Reactor, Asheville, NC (2013).

224