An Investigation Into Friction Stir Welding of Copper Niobium Nanolamellar Composites

Mississippi State University Scholars Junction

Theses and Dissertations Theses and Dissertations

5-1-2016

An Investigation into Friction Stir Welding of Copper Niobium Nanolamellar Composites

Josef Benjamin Cobb

Follow this and additional works at: https://scholarsjunction.msstate.edu/td

Recommended Citation Cobb, Josef Benjamin, "An Investigation into Friction Stir Welding of Copper Niobium Nanolamellar Composites" (2016). Theses and Dissertations. 569. https://scholarsjunction.msstate.edu/td/569

This Dissertation - Open Access is brought to you for free and open access by the Theses and Dissertations at Scholars Junction. It has been accepted for inclusion in Theses and Dissertations by an authorized administrator of Scholars Junction. For more information, please contact [email protected]. Template C v3.0 (beta): Created by J. Nail 06/2015

An investigation into friction stir welding of copper niobium nanolamellar composites

By TITLE PAGE Josef Benjamin Cobb

A Dissertation Submitted to the Faculty of Mississippi State University in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy in Engineering in the Department of Mechanical Engineering

Mississippi State, Mississippi

May 2016

2016

An investigation into friction stir welding of copper niobium nanolamellar composites

By APPROVAL PAGE Josef Benjamin Cobb

Approved:

______Judith A. Schneider (Major Professor)

______John S. Carpenter (Committee Member)

______Tonya W. Stone (Committee Member)

______Nathan A. Mara (Committee Member)

______Keith Walters (Graduate Coordinator)

______Jason M. Keith (Dean) Bagley College of Engineering

Name: Josef Benjamin Cobb ABSTRACT Date of Degree: May 6, 2016

Institution: Mississippi State University

Major Field: Engineering

Major Professor: Judith A. Schneider

Title of Study: An investigation into friction stir welding of copper niobium nanolamellar composites

Pages in Study 94

Candidate for Degree of Doctor of Philosophy

The workpiece materials used in this study are CuNb nano-layered composites

(NLC) which are produced in bulk form by accumulative roll bonding (ARB). CuNb

NLC panels are of interest because of their increase in strength and radiation damage tolerance when compared to either of their bulk constituents. These increased properties stem from the bi-metal interface, and the nanometer length-scale of the layers. However to be commercially viable, methods to successfully join the ARB NLC which retain the layered structure panels are needed. Friction stir welding is investigated in this study as a possible joining method that can join the material while maintaining its layered structure and hence its properties.

Mechanical properties of the weld were measured at a macro level using tensile testing, and at a local level via nano-indentation. The post weld layer structure was analyzed to provide insight into the flow paths. The grain orientation of the resulting weld nugget was also analyzed using electron backscatter diffraction and transmission

Kikuchi diffraction.

Results from this study show that the nano-layered structure can be maintained in the CuNb NLC by control of the friction stir welding parameters. The resulting microstructure is dependent on the strain experienced during the joining process. A variation in layer thickness reduction is correlated with increasing shear strain. Above a critical level of shear strain, the NLC microstructure was observed to fragment into equiaxed grains with a higher hardness than the NLC panels. Results from this study are also used to further the understanding of the material flow and hot working conditions experienced during the friction stir welding process.

DEDICATION

I would like to dedicate this dissertation to my family. The support of my wife,

April, throughout my graduate studies has been essential in my path to my graduation.

Her keeping our son during late nights at school and pushing me to work on research have kept me on schedule for graduation. Benjamin, you have been an excellent source of motivation to graduate. All of the extra events you have added during the course of my studies are worth it in exchange for having you in my life. The education provided by my parents has served me well thought my college life and I am sure it will continue to in future. Their loving care of my siblings and I engrained in me the character that has guided me throughout my college career.

ACKNOWLEDGEMENTS

I would like to thank my major professor Dr. Judy Schneider for all the help, guidance, insight, and knowledge she has provided over the course of my graduate studies. Her dedication to my progression as a researcher and a professional was invaluable. The information I obtained from her in our bi-weekly discussions improved not only my graduate studies, but will affect my lifelong career as a professional. Her teaching style, enthusiasm for material science, and genuine care for improving all of her graduate students created an environment that shaped my future in a positive way that would have only been possible under her guidance.

I would also like to thank Dr. John Carpenter. His mentoring during the time I spent at Los Alamos National Laboratories provided insight and experience that would not have been possible to obtain in a university setting. The knowledge and experience he contributed to my research was irreplaceable.

I would like to acknowledge Los Alamos National Laboratories for the LDRD that funded my two summer internships. Also, for the access to the material characterization equipment I used during these internships.

I would like to thank Donovan Leonard and acknowledge the Oak Ridge National

Laboratories SHARE facilities for the electron microscopy work performed.

iii

I would also like to thank my committee for all their time and effort put into my research. The valuable guidance they have provided in obtaining my Ph.D. is greatly appreciated.

TABLE OF CONTENTS

DEDICATION ...... ii

ACKNOWLEDGEMENTS ...... iii

LIST OF TABLES ...... vii

LIST OF FIGURES ...... viii

CHAPTER

I. INTRODUCTION ...... 1

1.1 Background ...... 1 1.1.1 CuNb nano-layered composites produced by ARB ...... 5 1.1.2 SPD processing ...... 10 1.2 Objective of study ...... 11 1.3 References ...... 13

II. LAYER STABILITY AND MATERIAL PROPERTIES OF FRICTION STIR WELDED COPPER NIOBIUM NANOLAMELLAR COMPOSITE PLATES1 ...... 17

2.1 Abstract ...... 17 2.2 Introduction ...... 17 2.3 Experimental Procedures ...... 19 2.4 Results and Discussion ...... 20 2.5 Conclusion ...... 26 2.6 Acknowledgments ...... 26 2.7 References ...... 27

III. MAINTAINING THE NANOLAMELLAR STRUCTURE OF ACCUMULATIVE ROLL BONDED PLATES DURING FRICTION STIR WELDING2 ...... 31

3.1 Abstract ...... 31 3.2 Introduction ...... 32 3.3 Experimental ...... 36 3.4 Results and Discussion ...... 39 3.5 Conclusions ...... 50 v

3.6 Acknowledgements ...... 51 3.7 References ...... 52

IV. TEXTURE ANALYSIS OF FRICTION STIR WELDED COPPER NIOBIUM NANOLAMELLAR COMPOSITES ...... 55

4.1 Introduction ...... 55 4.1.1 Energy dispersive X-ray spectroscopy (EDS) ...... 56 4.1.2 Conventional electron backscatter diffraction (cEBSD) ...... 56 4.1.3 Transmission Kikuchi Diffraction (TKD) ...... 60 4.2 Experimental Procedure ...... 61 4.2.1 EDS ...... 61 4.2.2 cEBSD ...... 61 4.2.3 TKD ...... 62 4.3 Results ...... 64 4.3.1 EDS ...... 64 4.3.2 cEBSD scans ...... 65 4.3.3 TKD scans ...... 67 4.4 Discussion ...... 69 4.4.1 cEBSD and TKD Scans ...... 69 4.4.2 Factors influencing indexing of TKD ...... 73 4.5 Conclusions ...... 75 4.6 Acknowledgements ...... 76 4.7 References ...... 78

V. CONCLUSIONS ...... 81

5.1 References ...... 83

VI. FUTURE WORK ...... 84

APPENDIX

A. SEM AND EBSD SAMPLE PREPERATION METHODS ...... 85

B. STRESS VS. STRAIN CURVES FOR P.M., SINGLE PASS FSW, AND DOUBLE PASS FSW ...... 87

C. COPYRIGHT PERMISSIONS ...... 89

LIST OF TABLES

1.1 Range of predicted strains and strain rates in FSW ...... 5

3.1 Tensile properties of PM, single pass FSW, and double pass FSW...... 39

4.1 SEM settings for TKD ...... 64

vii

LIST OF FIGURES

1.1 Friction stir welding overview with relevant terminology ...... 3

1.2 Phase diagram of Cu and Nb [25] ...... 6

1.3 ARB roll bonding schematic ...... 7

1.4 ARB processed material [29] ...... 9

1.5 Illustration of HPT for (a) unconstrained and (b) (c) constrained conditions [37] ...... 10

2.1 SEI of a transverse section of FSW with regions-of-interest marked...... 21

2.2 Details of FSW transverse seen in Figure 2.1...... 21

2.3 BFI TEM of WN ...... 22

2.4 Hardness v 1/h1/2 plot for ARB CuNb [5]. across length scales from the PM to the FSW WN...... 24

3.1 a) Optical microscopy image of an electron beam welded CuNb nanolamellar composite overview. (b) Higher magnification of the boxed region in (a) shows the edge of the weld melt zone where a dendritic Nb structure has formed within a Cu matrix...... 33

3.2 (a) FSW diagram showing the process parameters, workpiece material, tool, advancing and retreating sides. (b) A schematic of the transverse view of a FSW which contains the weld nugget, thermo- mechanically affected zone (TMAZ), heat affected zone (HAZ), and parent material (PM)...... 34

3.3 Schematic of plan view of FSW zone in which material enters in slip lines around the FSW tool [after 16]...... 35

3.4 Diagram showing the transverse view layout of the double pass weld...... 37

3.5 Tensile dogbone dimensions with panel transverse direction (TD) and rolling direction (RD) labeled...... 38

viii

3.6 DIC xx strain field maps showing large local strain on the AS for single (a) and double pass (b) FSWs...... 40

3.7 SEM secondary image of the center of the double pass FSW fracture surface showing ductile fracture features...... 41

3.8 SEM backscatter image of PM microstructure showing the nanolamellar structure...... 42

3.9 (a) Location of SEM image in FSW region. (b) SEM backscatter image of the single pass FSW nugget before removal of root side lack of penetration (LOP) region for the tensile test specimens...... 43

3.10 (a) Location of SEM image in FSW region for the single pass FSW. (b) SEM backscatter image of the AS of the FSW nugget showing the most extreme changes in direction and layer refinement...... 44

3.11 (a) Location of SEM image in FSW region for the double pass FSW. (b) SEM image of transverse section of the double pass FSW showing loss of visible layered structure, in the SEM, at the AS interface between the nugget and the TMAZ region...... 45

3.12 (a) Location of SEM image in FSW region for the double pass. (b) FSW Hardness map across the double pass FSW at the AS nugget/TMAZ interface. Hardness values are shown in GPa...... 46

3.13 Waviness swirls and vortices in the FSW region of a double pass weld in the CuNb nanolayers...... 48

3.14 Common velocity profile of FSW, HPT, and cyclic sliding...... 49

4.1 Typical EBSD setup showing the primary electron beam focused on the sample, and detector for collection of the electron backscatter patterns (EBSP)...... 57

4.2 Schematic illustrating an electron beam interaction volume for an EBSD setup for a material with a 30nm grain size [4]...... 59

4.3 Infrared image of a typical TKD setup in a SEM chamber [17] with the tilt angle α being a negative angle compared to a cEBSD setup...... 60

4.4 SEI SEM Image of the transverse section of the single pass FSW nugget...... 62

4.5 SEM secondary electron image of Sample C prepared by FIB process ...... 63

4.6 STEM - EDS maps of sample C from the AS of the FSW nugget which shows the distinct layers of Cu and Nb...... 65

4.7 Secondary electron SEM image with red arrows showing the locations of cEBSD scans in the single pass FSW nugget...... 66

4.8 a) Secondary electron SEM image with red arrows showing the locations of cEBSD scans in the single pass FSW nugget. b) Pole figures of the Nb just passed the centerline of the FSW on the AS. c) Pole figures of the Nb just before the centerline of the FSW on the RS. d) Pole figures of the Nb in the PM...... 67

4.9 a) Phase distribution map for sample B with the Cu phase shown in red and the Nb phase in green. b) Orientation map of Cu grains. c) Orientation map of Nb grains. d) A stereographic triangle relating the colors to orientations...... 68

4.10 a) Phase distribution map for sample C with the Cu phase shown in red and the Nb phase in green. b) Orientation map of Cu grains. c) Orientation map of Nb grains. d) A stereographic triangle relating the colors to orientations...... 68

4.11 a) Phase distribution map for sample D with the Cu phase shown in red and the Nb phase in green. b) Orientation map of Cu grains. c) Orientation map of Nb grains. d) A stereographic triangle relating the colors to orientations...... 69

4.12 Inverted x-ray radiograph images from the tungsten wire tracer study showing a) the plan view of the wire entering on the RS exiting in an orderly fashion b) a plan view of the wire entering the AS exiting in a scattered fashion in the x and y coordinates c) a transverse view of the wire from (b) showing scattered z coordinates [23]...... 71

4.13 cEBSD results from h=86 nm ARB CuNb sample...... 73

A.1 Stress vs. strain curve for PM, single pass FSW, and double pass FSW...... 88

CHAPTER I

INTRODUCTION

1.1 Background

Friction stir welding (FSW) is a solid state process used to join materials by inserting a rotating tool pin through the workpiece thickness, which imparts a shearing strain into the workpiece material as the rotating tool traverses the length of the weld.

The process is illustrated in Figure 1.1 which shows the shoulder riding on the crown surface of the panels in a butt weld configuration with the pin extending through the panel thickness. This non-symmetrical process is characterized by a retreating side (RS), where the velocity and travel vectors are opposed, and an advancing side (AS), where the vectors are aligned. FSW was patented in 1991 for the joining of like aluminum alloy

materials [1], later progressing to joining of dissimilar aluminum alloys [2] [3], and, then,

other higher melting temperature materials [4]. FSW alleviated many of problems

encountered in fusion welding of heavily alloyed materials such as the tendency for

liquation and hot cracking [4]. FSW has also shown promise in joining families of very

dissimilar materials, such as Al to steel, without forming detrimental phases [5] [6] [7]

[8].

Fundamental to developing predictive models for the FSW process is an

understanding of how the material flows as influenced by the tool geometry and process

parameters. Embedded tracers in FSWs have been used in an attempt to trace or mark the

material flow paths during the process [9] [10] [11] [12]. Post weld inspection, such as x- ray radiography was used to document the post weld placement of the marker. Colligan used 380 µm diameter steel shot embedded along a seam in the faying surface to first demonstrate that the material had an orderly flow [9]. Seidel used AA5454-H3 markers inserted at set intervals across the length and thickness of the weld which documented the flow of material flow around the tool which moved past its entrance point [10].

Schneider inserted 25 µm diameter longitudinal tungsten wires along the faying surfaces at different material thickness locations, in addition to tool offsets, to understand the flow as a function of the processing parameters. The wires fractured into segments around the tool which helped visualize the material flow patterns [11]. In another study, a 250 µm diameter lead wire was placed along the faying surface of an AA2195 PM which became molten at the FSW temperatures providing a continuous trace of the material flow [11].

This study documented changing stick-slip conditions at the shoulder-workpiece interface

[24]. Thus the use of tracers has improved our understanding of the material flow during

FSW and how process parameters affect the material flow. However, tracer size, discontinuity, and dissimilarity to the welded material leave questions as to how completely they capture the finer nuances of the flow paths.

A recent experimental approach toward measuring strain rate utilized in-situ x-ray tracking of a tungsten spheres inside a FSW [12]. Morisada embedded 300 µm tungsten

balls along the faying surface and tracked their position in real time using x-ray video

cameras [12]. Using a rotational speed of 1000 rpm and a travel rate of 400 mm/min, a

maximum strain rate of 13.4 s-1 was measured.

Figure 1.1 Friction stir welding overview with relevant terminology

Marker studies have improved the understanding of material flow during the FSW

process and have been used as the basis for modeling approaches. These approaches

however rely on knowing the hot working conditions of the FSW process which are not

completely understood. Typically hot working conditions for a metal are defined by the

strain, strain rate and temperature of the process. Various approaches have been

attempted to define these parameters for the FSW process including analytical modeling,

numerical modeling, and experimentation.

Various studies have embedded thermocouples in the weld panels to measure the

bulk temperature during a FSW. Although the thermocouples are usually consumed

during the weld, this approach has shown bulk temperatures around the FSW tool to be in

the range of 0.5 to 0.9 times the homologous temperature [13] [14] [15].

Various modeling approaches have also been undertaken to determine the strain and strain rate the material experiences as summarized in Table 1. These include numerical models [16] [17] [18] [19] [20] and analytical models [21] [22]. These

methods, predict strain in the FSW process, utilizing either material dependent or material independent models. They are also based on assumptions regarding the material flow path which has been obtained in various marker studies. Material dependent estimations have been made using finite element models (FEM) [18] [16] with strains <

10 reported. The FSW strain rates predicted by FEM are on the order of 101 s-1 . As

Table 1 shows, these values are the lowest as compared to other techniques which may be due to the inability to refine the FEA mesh in the localized shear zone regions of a FSW.

Computational fluid dynamic (CFD) models are able to give a depiction of the material flow, but because they assume a metal viscosity, they do not estimate localized strain.

CFD approaches estimate strain rates in the range of 102 to 103 s-1, which is higher than that predicted using FEM.

A material independent, kinematic model has been developed to estimate strains and strain-rates during FSW [21]. This model is based on slip-line theory used in metal forming processes and is used to describe the material flow paths. Based on these predicted flow paths, strain and strain rates can be calculated based on the tool geometry and process parameters [23]. The equations for strain and strain-rate are shown in

Equation 1.1 and Equation 1.2 respectively where R is the pin radius, ω is the tool rotation, V is the travel rate, and δ is the estimated shear zone thickness.

� × � � = (1.1) �

� × � �′ = (1.2) �

The kinematic model assumes full adhesion of the material to the tool during in- plane material movement, and does not take into account the out-of-plane material flow that occurs in FSW [11] [10] [9]. Some validation of the predicted flow fields or slip lines have been made using wire tracer experiments [24]. The differences in strains and strain rates summarized in Table I demonstrate the need to explore other experimental techniques to quantify the strain and strain rate conditions that occur during FSW. This is ultimately needed to develop predictive tools to improve the process and reduce costs associated with implementation.

Table 1.1 Range of predicted strains and strain rates in FSW

Modeling Strain-rate Strain Reference Method

FEM 101 s -1 101 [16] [18]

CFD 102 – 103 s -1 Not reported [19] [20]

Kinematic 103 – 106 s -1 5*101 – 102 [21], [22]

Experimental 13.4 s-1 Not reported [12]

1.1.1 CuNb nano-layered composites produced by ARB

The workpiece materials used in this study are CuNb nano-layered composites

(NLC) which are produced in bulk form by accumulative roll bonding (ARB). This

process produces panels composed of alternating layers of Cu and Nb which have sub-

micron individual layer thicknesses. The phase diagram shown in Figure 1.2 shows that

Cu and Nb are thermodynamically immiscible at room temperature [25].

Figure 1.2 Phase diagram of Cu and Nb [25]

The immiscibility of the two metals allows for elementally distinct layers to be maintained during the accumulative roll bonding (ARB) process [26] [27]. The ARB process is illustrated in Figure 1.3 and described in more detail in [26] [27]. This process consists of stacking individual sheets on Cu and Nb on top of one another and putting the stacked sheets through a rolling mill. This deformation bonds the panels which are then cut, restacked, and the process is repeated until the layers are the final thickness desired.

No process annealing was done on the ARB panels used in this study.

Figure 1.3 ARB roll bonding schematic

CuNB NLC panels exhibit increased strength [26], [27], reduction in damage

under shock loading [28], and improved radiation damage tolerance[29] when compared

to either of their bulk constituents. The ultimate strength of CuNb NLCs is highly

dependent on the layer thickness of the material with the strength increasing as the layer

thickness decreases. This effectively limits the grain size in the direction perpendicular to

the interface plane. Ultimate tensile strengths in excess of 1 GPa can be achieved by

reducing the layer thickness below 135 nm [27]. The enhancements of shock loading and radiation damage tolerance are largely due to the interfaces of the CuNb system acting as a sink for dislocations [28], [30].

CuNb NLC panels are produced using one of two methods: physical vapor

deposition (PVD) [26] or ARB. PVD fabrication methods are limited to laboratory scale

production; however, the rolling process of ARB allows it to produce bulk panels, as seen in Figure 1.4 [29], and is commercially scalable. For this reasons the ARB material is the focus of this study. The industrial scalability of the ARB method to produce bulk quantities of this material have led to a need to join the material while maintaining its layered structure and, hence, its enhanced properties. The ability of a joining process to meet these requirements would advance the commercial viability of this material. Joining efforts of CuNb lamellar composites using traditional fusion welding techniques result in a loss of a layered structure due to the melting and resolidification of the layers. The results of which are shown in more detail in Chapter III. The solid-state nature of FSW may be able to avoid this problem. While FSW has shown it can successfully join panels of different metal alloys, the bi-metal panels joined to date have consisted of each panel being an individual alloy [6][7][8]. The successful joining of CuNb NLC may do more than just further the commercial viability of the material as the layered structure of the material may serve as an internal marker. Since the marker is the actual FSW material, it may be able to provide insight to quantifying the material flow path during the FSW process.

Figure 1.4 ARB processed material [29]

The panels used in this study have an individual layer thickness of 200 nm which corresponds to a strain input of 8.5 by the ARB process. The rolling history of the ARB process has been reported to influence the final texture of the material [31] [32]. As texture in the layers depends on the strain path, texture may be able to be used to deduce the FSW strain history by analyzing the post-weld FSW nugget texture.

1.1.2 SPD processing

Severe plastic deformation (SPD) processing is defined as a forming process which imparts strains in excess of 2 into a metal in order to produce ultra-fine grains [33].

SPD processes include: ARB [27] [34], equal angle extrusion [35], high pressure torsion

(HPT) [36] [37], and FSW [38]. Of these SPD processes, HPT and FSW share a similarity in the torsional processing path the metal experiences. Hence results obtained from HPT may be applicable to FSW regarding the nano-layer stability during processing.

Studies have been done on CuNb NLC using HPT to evaluate stability of the layer structure. Figure 1.5 illustrates the HPT process where a disc is placed between two anvils and one of the anvils is rotated, imparting a shear strain into the material.

Figure 1.5 Illustration of HPT for (a) unconstrained and (b) (c) constrained conditions [37]

CuNb NLC subjected to HPT showed the layered structure of the material to be

stable under shearing strains [39] [36]. The amount of strain accommodated with layer

retention was found to be dependent on the initial layer starting thickness. ARB CuNb

NLC panels with initial 18 nm thick layers, were able to accommodate strains () in the 10

range of 10.8 < ε < 68.5 before the layered structure destabilized into equiaxed grains.

When the initial layer thickness was increased to 200nm, the strain required to destabilize the layer structure and form equiaxed grains increased to > 3400 [39] . Although FSW differs from HPT in the fact that the FSW tool has a pin, rotates at a higher RPM, has a forward travel component, and occurs at a much higher temperature, it is similar in the fact that one of the HPT anvils and the FSW tool both rotate while the contact surface opposite them is stationary. This similarity of the strain input by the two processes may result, with proper parameter selection for FSW, in joining of the ARB CuNb NLC while retaining a layered structure.

1.2 Objective of study

The objective of this study is twofold. First, is to demonstrate the ability to join

ARB CuNb NLC panels while retaining the nano-layered microstructure. The second is to enhance the understanding of the flow/strain path the material experiences during the

FSW process.

It is proposed that the nano-layered structure of the CuNb NLC may be able to provide insight into the FSW process by acting as a marker for the material movement.

The discontinuities of the dissimilar markers used in previous studies were noted to be difficult to extrapolate back to in-situ material flow and have not provided a quantitative estimate of the strain experienced [9] [10] [11]. The texture of the material fabricated via

ARB is dependent on its processing history [32].The reorientation and refinement of the layers post joining may provide insight into the material flow during the FSW process.

Deformation mechanisms from texture studies of FSW aluminum have been previously performed [40]. The addition of layers to such a study would allow for a flow direction to 11

be coupled to a local texture for the orientation of the grains after FSW, and may be able

to provide further insight into the deformation history experienced. Thus while

demonstrating that FSW process is able to join the ARB CuNb panels of interest, it may

also provide quantitative information on the strain which the material experiences during

the process.

This study uses nominally 0.6 mm thick ARB CuNb NLC with an individual layer

thickness of 200 nm as the parent material (PM) to join using FSW. Mechanical properties are determined using tensile testing and nanoindentation, coupled with scanning electron microscopy (SEM) imaging and electron backscatter diffraction

(EBSD) detectors to correlate structural-property relations. Further details of these methods will be discussed in the sections which present that data obtained. Imaging of

the post weld layered structure is expected to provide further insight into the material

flow paths during FSW. The resulting grain orientation and/or texture are also expected

to provide additional information regarding the deformation history, or strain,

experienced during FSW.

1.3 References

[1] W. M. Thomas, E. D. Nicholas, J. C. Needham, M. G. Murch, P. Temple-Smith, and C. J. Dawes, “Method of operating on a workpiece,” U.S. Patent 5460317, Oct-1995.

[2] P. Cavaliere, R. Nobile, F. W. Panella, and A. Squillace, “Mechanical and microstructural behaviour of 2024–7075 aluminium alloy sheets joined by friction stir welding,” Int. J. Mach. Tools Manuf., vol. 46, no. 6, pp. 588–594, May 2006.

[3] I. Shigematsu, Y.-J. Kwon, K. Suzuki, T. Imai, and N. Saito, “Joining of 5083 and 6061 aluminum alloys by friction stir welding,” J. Mater. Sci. Lett., vol. 22, no. 5, pp. 353–356, 2003.

[4] J. A. Schneider, D. Williston, T. L. Murphy, C. Varner, J. Hawkins, and B. Walker, “Solid state joining of nickel based alloy, Haynes 230,” J. Mater. Process. Technol., vol. 225, pp. 492–499, Nov. 2015.

[5] C. Schneider, T. Weinberger, J. Inoue, T. Koseki, and N. Enzinger, “Characterisation of interface of steel/magnesium FSW,” Sci. Technol. Weld. Join., vol. 16, no. 1, pp. 100–107, Jan. 2011.

[6] Y. Li, E. A. Trillo, and L. E. Murr, “Friction-stir welding of aluminum alloy 2024 to silver,” J. Mater. Sci. Lett., vol. 19, no. 12, pp. 1047–1051, 2000.

[7] L. E. Murr, Y. Li, R. D. Flores, E. A. Trillo, and J. C. McClure, “Intercalation vortices and related microstructural features in the friction-stir welding of dissimilar metals,” Mater. Res. Innov., vol. 2, no. 3, pp. 150–163, 1998.

[8] J. Schneider and R. Radzilowski, “Welding of Very Dissimilar Materials (Fe- Al),” JOM, vol. 66, no. 10, pp. 2123–2129, Oct. 2014.

[9] K. Colligan, “Material Flow Behavior during Friction Stir Welding of Aluminum,” Weld. J., vol. 78, no. 7, pp. 229–237, 1999.

[10] T. U. Seidel and A. P. Reynolds, “Visualization of the material flow in AA2195 friction-stir welds using a marker insert technique.pdf,” Metall. Mater. Trans. A, vol. 32, no. 11, pp. 2879–2884, Nov. 2001.

[11] J. Schneider, R. Beshears, and A. C. Nunes, “Interfacial sticking and slipping in the friction stir welding process,” Mater. Sci. Eng. A, vol. 435–436, pp. 297–304, Nov. 2006.

[12] Y. Morisada, T. Imaizumi, and H. Fujii, “Determination of strain rate in Friction Stir Welding by three-dimensional visualization of material flow using X-ray radiography,” Scr. Mater., vol. 106, pp. 57–60, Sep. 2015.

[13] L. Ruilin, H. Diqiu, L. Luocheng, Y. Shaoyong, and Y. Kunyu, “A study of the temperature field during ultrasonic-assisted friction-stir welding,” Int. J. Adv. Manuf. Technol., vol. 73, no. 1–4, pp. 321–327, Jul. 2014.

[14] A. R. Darvazi and M. Iranmanesh, “Thermal modeling of friction stir welding of stainless steel 304L,” Int. J. Adv. Manuf. Technol., vol. 75, no. 9–12, pp. 1299– 1307, Dec. 2014.

[15] A. I. Toumpis, A. M. Galloway, L. Arbaoui, and N. Poletz, “Thermomechanical deformation behaviour of DH36 steel during friction stir welding by experimental validation and modelling,” Sci. Technol. Weld. Join., vol. 19, no. 8, pp. 653–663, Nov. 2014.

[16] A. Arora, Z. Zhang, A. De, and T. DebRoy, “Strains and strain rates during friction stir welding,” Scr. Mater., vol. 61, no. 9, pp. 863–866, Nov. 2009.

[17] Z. W. Chen and S. Cui, “Strain and strain rate during friction stir welding/processing of Al-7Si-0.3Mg alloy,” IOP Conf. Ser. Mater. Sci. Eng., vol. 4, no. 1, p. 012026, Aug. 2009.

[18] G. Buffa, J. Hua, R. Shivpuri, and L. Fratini, “A continuum based fem model for friction stir welding—model development,” Mater. Sci. Eng. A, vol. 419, no. 1–2, pp. 389–396, Mar. 2006.

[19] S. Z. Aljoaba, I. S. Jawahir, O. W. Dillon, M. H. Ali, and M. K. Khraisheh, “Modeling of friction stir processing using 3D CFD analysis,” Int. J. Mater. Form., vol. 2, no. S1, pp. 315–318, Aug. 2009.

[20] A. F. Hasan, C. J. Bennett, and P. H. Shipway, “A numerical comparison of the flow behaviour in Friction Stir Welding (FSW) using unworn and worn tool geometries,” Mater. Des., vol. 87, pp. 1037–1046, Dec. 2015.

[21] A. C. Nunes, “Metal Flow in Friction Stir Welding,” MS&T, 2006.

[22] B. M. Darras and M. K. Khraisheh, “Analytical Modeling of Strain Rate Distribution During Friction Stir Processing,” J. Mater. Eng. Perform., vol. 17, no. 2, pp. 168–177, Apr. 2008.

[23] W. Johnson, R. Sowerby, and J. Haddow, Plane-Strain Slip-Line Fields: The Theory and Bibliography. New York, NY, USA: American Elsevier publishing company Inc, 1970.

[24] J. A. Schneider and A. C. Nunes, “Quantifying the Material Processing Conditions for an Optimized FSW,” 7th Int’l Conf Trends Weld., pp. 253, 256, 2005.

[25] H. Okamoto, “Cu-Nb (Copper-Niobium),” J. Phase Equilibria Diffus., vol. 33, no. 4, pp. 344–344, Aug. 2012.

[26] A. Misra, J. P. Hirth, and R. G. Hoagland, “Length-scale-dependent deformation mechanisms in incoherent metallic multilayered composites,” Acta Mater., vol. 53, no. 18, pp. 4817–4824, Oct. 2005.

[27] I. J. Beyerlein, N. A. Mara, J. S. Carpenter, T. Nizolek, W. M. Mook, T. A. Wynn, R. J. McCabe, J. R. Mayeur, K. Kang, S. Zheng, J. Wang, and T. M. Pollock, “Interface-driven microstructure development and ultra high strength of bulk nanostructured Cu-Nb multilayers fabricated by severe plastic deformation,” J. Mater. Res., vol. 28, no. 13, pp. 1799–1812, 2013.

[28] N. A. Mara and I. J. Beyerlein, “Review: effect of bimetal interface structure on the mechanical behavior of Cu–Nb fcc–bcc nanolayered composites,” J. Mater. Sci., vol. 49, no. 19, pp. 6497–6516, Oct. 2014.

[29] I. J. Beyerlein, A. Caro, M. J. Demkowicz, N. A. Mara, A. Misra, and B. P. Uberuaga, “Radiation damage tolerant nanomaterials,” Mater. Today, vol. 16, no. 11, pp. 443–449, Nov. 2013.

[30] W. Han, M. J. Demkowicz, N. A. Mara, E. Fu, S. Sinha, A. D. Rollett, Y. Wang, J. S. Carpenter, I. J. Beyerlein, and A. Misra, “Design of Radiation Tolerant Materials Via Interface Engineering,” Adv. Mater., vol. 25, no. 48, pp. 6975– 6979, 2013.

[31] J. S. Carpenter, S. C. Vogel, J. E. LeDonne, D. L. Hammon, I. J. Beyerlein, and N. A. Mara, “Bulk texture evolution of Cu–Nb nanolamellar composites during accumulative roll bonding,” Acta Mater., vol. 60, no. 4, pp. 1576–1586, Feb. 2012.

[32] J. S. Carpenter, R. J. McCabe, S. J. Zheng, T. A. Wynn, N. A. Mara, and I. J. Beyerlein, “Processing Parameter Influence on Texture and Microstructural Evolution in Cu-Nb Multilayer Composites Fabricated via Accumulative Roll Bonding,” Metall. Mater. Trans. A, vol. 45, no. 4, pp. 2192–2208, Apr. 2014.

[33] A. Azushima, R. Kopp, A. Korhonen, D. Y. Yang, F. Micari, G. D. Lahoti, P. Groche, J. Yanagimoto, N. Tsuji, A. Rosochowski, and A. Yanagida, “Severe plastic deformation (SPD) processes for metals,” CIRP Ann. - Manuf. Technol., vol. 57, no. 2, pp. 716–735, Jan. 2008.

[34] S.-B. Lee, J. E. LeDonne, S. C. V. Lim, I. J. Beyerlein, and A. D. Rollett, “The heterophase interface character distribution of physical vapor-deposited and accumulative roll-bonded Cu–Nb multilayer composites,” Acta Mater., vol. 60, no. 4, pp. 1747–1761, Feb. 2012.

[35] V. M. Segal, “Severe plastic deformation: simple shear versus pure shear,” Mater. Sci. Eng. A, vol. 338, no. 1, pp. 331–344, 2002.

[36] E. H. Ekiz, T. G. Lach, R. S. Averback, N. A. Mara, I. J. Beyerlein, M. Pouryazdan, H. Hahn, and P. Bellon, “Microstructural evolution of nanolayered Cu–Nb composites subjected to high-pressure torsion,” Acta Mater., vol. 72, pp. 178–191, Jun. 2014.

[37] A. P. Zhilyaev, T. R. McNelley, and T. G. Langdon, “Evolution of microstructure and microtexture in fcc metals during high-pressure torsion,” J. Mater. Sci., vol. 42, no. 5, pp. 1517–1528, Mar. 2007.

[38] R. D. Doherty, D. A. Hughes, F. J. Humphreys, J. J. Jonas, D. J. Jensen, M. E. Kassner, W. E. King, T. R. McNelley, H. J. McQueen, and A. D. Rollett, “Current issues in recrystallization: a review,” Mater. Sci. Eng. A, vol. 238, no. 2, pp. 219– 274, Nov. 1997.

[39] H. E. Ekiz, “Microstructural evolution and phase stability of Cu/Nb nanolaminates subjected to severe plastic deformation by high pressure torsion,” Dissertation, University of Illinois at Urbana-Champaign, 2015.

[40] S. Chen and X. Jiang, “Texture evolution and deformation mechanism in friction stir welding of 2219Al,” Mater. Sci. Eng. A, vol. 612, pp. 267–277, Aug. 2014.

CHAPTER II

LAYER STABILITY AND MATERIAL PROPERTIES OF FRICTION STIR WELDED

COPPER NIOBIUM NANOLAMELLAR COMPOSITE PLATES1

2.1 Abstract

Initial efforts to friction-stir weld (FSW) Cu–Nb nanolamellar composite plates

fabricated via accumulative roll bonding are reported in this study. Parent material layers

within the composite were nominally 300 nm and exhibited a hardness of 2.5 GPa. After

FSW two types of microstructures were present: a refined layered structure, and an

equiaxed nanocrystalline microstructure with grain diameters on the order of 7 nm. The

type of microstructure was dependent on location within the FSW nugget and related to

varying amounts of strain. Material hardness increased with refinement, with the

equiaxed microstructure reaching a maximum hardness of 6.0 GPa.

______

1 Published as: Layer Stability and Material Properties of Friction-Stir Welded

Cu–Nb Nanolamellar Composite Plates, Materials Research Letters, vol. 2, no. 4, pp 227-

232, 2014.

2.2 Introduction

Nanolamellar CuNb bimetallic composites have been extensively studied due to their significantly higher strength and radiation damage tolerance as compared with their 17

bulk counterparts [1][2][3][4][5][6][7]. These improvements are reported to increase as

the layer thickness (h) decreases to the nanometer (nm) scale with a maximum strength

occurring around a thickness of 2 nm.[2] These properties are reported to emanate from a

combination of heterophase interface character (α) and density (ρ), with the strength largely dominated by ρ [5][6]. CuNb nanolamellar bimetallic composites have been studied for materials produced via two fabrication routes: physical vapor deposition

(PVD) [4] and accumulative roll bonding (ARB) [8][9][10]. While the PVD method can currently only produce a small volume of laboratory scale material, the ARB process allows the production of much larger volume, engineering-scale quantities of material

[5][9]. Previous studies indicate that ρ can be controlled and ω stabilized during ARB processing via manipulation of processing parameters [10]. Thus, ARB opens the door for the fabrication of nanocomposite materials with properties tailored for specific applications via industry scalable manufacturing techniques. To further realize this goal, the stability of α and ρ under commercially available joining processes must be understood. FSW, a solid state joining process, is explored as a possible joining process for maintaining high ρ and hence the strength of the material within the weld region.

Simulations show that the stability of α, observed in previous studies [7][11][12] is

derived from the ability of specific interfacial structures to accommodate the plastic

deformation of the ARB process. Since the FSW process will introduce additional severe

plastic deformation in a manner that differs considerably from that used to synthesize the

initial composite, the stability of the preexisting α is uncertain.

In FSW, a weld tool, consisting of a pin and shoulder, is rotated and plunged into

the panel thickness, and traversed the length of the workpiece, and thereby joining the

material [13][14]. The two sides of the weld joint are referred to as the advancing side

(AS), the side where tool rotation aligns with the travel direction, and the retreating side

(RS), the side where tool rotation is opposite the travel direction. While literature reports the use of FSW to join dissimilar alloys in butt and lap welds, these studies address the joining of different homogenous bulk material on either side of the weld [15][16][17].

The current study investigates FSW as a possible joining method for a multi-layered, nano-scaled bimetal composite where both phases are resistant to chemical intermixing due to their positive heats of mixing in the solid phase.

2.3 Experimental Procedures

FSW were made in a nominal 0.6 mm thick sheet of ARB CuNb. The ARB material began with a single 2 mm thick sheet of polycrystalline Nb (99.97% pure, ATI-

Wah Chang) and two 1 mm thick sheets of polycrystalline Cu (99.99% pure, Southern

Copper and Supply). The Nb layer was sandwiched between the two Cu layers and a 60% reduction in overall thickness was performed in a single step utilizing a two-high rolling mill (Waterbury-Farrel, Brampton, Canada). Layer thickness was reduced and number of layers increased through a repetitive sequence of cutting, stacking, and roll bonding. Prior to each roll bonding step, a surface treatment consisting of a 5 minute ultrasonic acetone bath was performed followed by wire brushing. Further details of the ARB process can be found in literature [9][10]. This process was repeated until a panel consisting of 50 vol.% Cu/ 50 vol.% Nb of alternating layers with average h ≈ 300 nm was fabricated. A single piece, smooth tapered FSW tool was made from a lanthanated tungsten W-La2O3

alloy with a flat 2 mm diameter shoulder, initial pin length of 0.5 mm, upper diameter of

1 mm, and a taper of 27. All FSWs were made in a bead on plate configuration using 19

displacement control at a travel rate of 25.4 mm/m and a tool rotation of 1500 rpm.

Microstructure images were obtained using an FEI (Hillsboro, OR) Inspect F

Scanning Electron Microscope (SEM) and a FEI TITAN 300KeV transmission electron microscope (TEM). Hardness data was obtained on a Nano-XP 30 Indenter (Agilent,

Santa Clara, CA) with a nominal 150 nm radius Berkovich tip. Load vs. displacement data was collected in continuous stiffness monitoring mode for indentation depths between 200 and 1100 nm and this data was used for hardness calculations [18]. The

FSW panels were cross-sectioned normal to the welding direction using a low speed diamond saw. SEM/indentation samples were mounted in a low heat epoxy resin and metallurgically ground and polished. TEM foils were extracted and thinned utilizing a

FEI Helios Focused Ion Beam (FIB).

2.4 Results and Discussion

Figure 2.1 shows a representative SEM-based secondary electron image (SEI) of a transverse section of the FSW. Regions of interest include the parent material (PM), the thermal mechanical affected zone (TMAZ), and the weld nugget (WN). All FSW, regardless of process parameters, exhibited a weld defect on the AS crown surface. The images in Figure 2.2 document the change in layer direction and thickness within the

TMAZ region. Details of the representative regions in Figure 2.1 are shown in the SEI in

Figure 2.2a-c. Superimposed on these images are the nano hardness data with values in

GPa. To resolve the finer layers, TEM bright field imaging (BFI), on the RS, was used to obtain Figure 2.2d. It shows layers with an h of 6 nm, and a variation in material flow resulting from the asymmetrical flow field of the FSW [13].

Figure 2.1 SEI of a transverse section of FSW with regions-of-interest marked.

Note this sample was taken after 5 cm of travel with significant tool wear having occurred.

Figure 2.2 Details of FSW transverse seen in Figure 2.1.

(a) FSW nugget, (b) RS TMAZ, and (c) AS TMAZ. Location of nano indentations and hardness values in GPa are shown in (a)-(c) with (d) representing a BFI TEM of the RS WN in a layered region. The red line in (a) indicates the location where the TEM foil was extracted for Figure 2.1.

A FIB was used to extract an additional foil from the region labeled by vertical red line in Figure 2.2a. Figure 2.3 shows the higher magnification BFI TEM and accompanying ring selected area (electron) diffraction pattern (SADP), which indicates a nanocrystalline structure. Changes in diffraction contrast in the BFI TEM indicate an equiaxed grain structure approximately 5 to 10 nm in diameter. The ring SADP in Figure

2.3 is consistent with a nanocrystalline structure, which exhibits distinct Cu and Nb phases. The mottled contrast suggests the individual phases are randomly distributed. The equiaxed grains on the order of the refined h suggest a pinch off mechanism may be responsible for the grain refinement [19]. In this mechanism, the torsional motion of the

FSW process causes the sub-boundaries extending across the thickness of the elongated grains to pinch off forming equiaxed grains.

Figure 2.3 BFI TEM of WN

WN exhibits a nanocrystalline structure with equiaxed grains on the order of 5 to 10 nm in diameter. The inset accompanying SADP shows a ring structure, consistent with a polycrystalline grains, and the presence of both Cu and Nb phases.

The layered structure of the PM was maintained in parts of the FSW region as

shown in Figure 2.2, especially on the RS and show changes in orientation resulting from

the asymmetric material flow. Figure 2.2d shows a refined layered structure in the WN

region, with the majority of the layers ranging from 6 to 12 nm. Equation 2.1 can used to

calculate the true strain (εt) input in the layered regions, and, assuming layer thicknesses in Figure 2.2d (which was close to the equiaxed nanograin region) of ≥6 nm to be stable, a minimum true strain required to form the equiaxed nanograin structure can also be calculated. For Equation 2.1; t1 is the initial h of the PM and t2 is the final h within the

FSW nugget.

εt=ln(t1/t2) (2.1)

Thus for the observed reduction of h from 300 to 6 nm, the FSW process input a minimum strain of 3.9 in the equiaxed nanograin WN region. While this strain value is similar to strain ranges reported in other FSW studies of aluminum [20][21][22], other studies indicate higher strains are possible which vary as a function of location [13].

A 2.5 GPa with a standard deviation of 0.13 GPa average hardness value was obtained from the 18 nanoindentations within the PM. Within the FSW region, the hardness values varied with respect to the refinement in microstructure. A steep increase in hardness was observed in the TMAZ region, due to the gradient of h. The average hardness of 5.9 GPa was measured in the equiaxed grain region of the WN.

The nanoindentation results performed on multiple FSW showed increases in hardness as ρ increases, consistent with observations from ARB CuNb sheet material

[5][6][10]. Figure 2.4 plots CuNb ARB hardness vs. h1/2 across the length scale from the

PM to the FSW WN.[5] It also illustrates the difference in grain geometry between the

WN and an as rolled ARB panel [5].

Figure 2.4 Hardness v 1/h1/2 plot for ARB CuNb [5]. across length scales from the PM to the FSW WN.

The insets illustrate the difference between the grains of an as rolled ARB panel (elongated grains), and the WN (equiaxed grains) at the finest length scale. Grain dimensions for the ARB panel are estimated from literature [10].

Figure 2.4 shows a 20% increase in hardness in the equiaxed 7 nm grain size region of the WN compared to that of the ARB panel with h = 7 nm. This increase in hardness likely arises due to different deformation mechanisms in the equiaxed grains of the WN.[23][24][25][26][27][28] Increased hardness suggests that friction stir processing of CuNb plates could be used to form a 3-dimensional nanocrystalline structure, to further increase the maximum obtainable strength. The significance in a 20% increase in hardness over an already extremely hard material (considering its bulk counterparts) could be of particular importance in high wear applications, and other areas where an even further increase in surface, or through-thickness, hardness is desired.

Although resolvable with TEM, the small size presented by the grains precluded

an electron backscatter diffraction study of ω within the weld nugget. Studies of ARB

sheets had well-defined layer directions, which allowed the use of a normal direction

inverse pole figure to act as a measure of the interface plane normal distribution. With no

layer directions present in the weld nugget, bulk texture measurements via neutrons or x-

rays will not provide an understanding of α. With grain diameters on the order of 7 nm,

this also limits the applicability of local texture measurements such as precession electron

diffraction within a TEM, a technique used in the past for investigations of CuNb

nanoscale multilayers [29][30].

With regards to the FSW process, there was a defect on the AS of the FSW

region. This type of defect is typically seen in FSW when insufficient down-force is

applied [31]. Although the shoulder was in contact with the material during welding, a

substantial influence on the material flow from the shoulder was not observed as

evidenced by the minimal material movement on the RS seen in Figure 2.1. Weld

material on the AS, beyond the defect and not shown in Figure 2.1, was also in intimate

contact with the shoulder, but material-movement/shoulder-interaction was virtually

nonexistent. It is possible that adding features to the shoulder would help increase the

lateral material movement and help eliminate the volumetric defect [14].

Tool wear during welding was significant with pin length decreasing by 50% after

5 cm of FSW length. Different material selection for the tool should improve tool life and

preserve features added to improve material flow. Although improvements in tool life and preventing defect formations are still areas of concern, FSW seems to be a viable

option for joining nanolamellar bi metallic composites while maintaining their enhanced strength.

2.5 Conclusion

FSW of the ARB panels lead to both refined layers, and a nanocrystalline structure. Where layers were present, directional changes were observed due to the non- symmetric flow field of the FSW [32][33]. The initial h of 300 nm was reduced to grains with 3D nanoscale diameters of 7 nm following the FSW process. Within the WN hardness values of the equiaxed grain region exceeded those of ARB CuNb panels with h equal to the equiaxed grain diameter. Hardness values were noted to increase from 2.5

GPa in the starting PM and increased to a maximum of 6.0 GPa as the length scale decreased.

2.6 Acknowledgments

This work is supported by the Los Alamos National Laboratory Directed

Research and Development (LDRD) project 20130764ECR. This work was performed, in part, at the Center for Integrated Nanotechnologies, an Office of Science User Facility operated for the U.S. Department of Energy (DOE) Office of Science. Los Alamos

National Laboratory, an affirmative action equal opportunity employer, is operated by

Los Alamos National Security, LLC, for the National Nuclear Security Administration of the U.S. Department of Energy under contract DE-AC52-06NA25396. Electron microscopy was performed at the Los Alamos Electron Microscopy Laboratory.

2.7 References

[1] N. A. Mara, D. Bhattacharyya, P. Dickerson, R. G. Hoagland, and A. Misra, “Deformability of ultrahigh strength 5nmCu∕Nb nanolayered composites,” Appl. Phys. Lett., vol. 92, no. 23, p. 231901, Jun. 2008.

[2] A. Misra, J. P. Hirth, and R. G. Hoagland, “Length-scale-dependent deformation mechanisms in incoherent metallic multilayered composites,” Acta Mater., vol. 53, no. 18, pp. 4817–4824, Oct. 2005.

[3] A. Misra, M. J. Demkowicz, X. Zhang, and R. G. Hoagland, “The radiation damage tolerance of ultra-high strength nanolayered composites,” Jom, vol. 59, no. 9, pp. 62–65, 2007.

[4] W. Han, M. J. Demkowicz, N. A. Mara, E. Fu, S. Sinha, A. D. Rollett, Y. Wang, J. S. Carpenter, I. J. Beyerlein, and A. Misra, “Design of Radiation Tolerant Materials Via Interface Engineering,” Adv. Mater., vol. 25, no. 48, pp. 6975– 6979, 2013.

[5] I. J. Beyerlein, N. A. Mara, J. S. Carpenter, T. Nizolek, W. M. Mook, T. A. Wynn, R. J. McCabe, J. R. Mayeur, K. Kang, S. Zheng, J. Wang, and T. M. Pollock, “Interface-driven microstructure development and ultra high strength of bulk nanostructured Cu-Nb multilayers fabricated by severe plastic deformation,” J. Mater. Res., vol. 28, no. 13, pp. 1799–1812, 2013.

[6] J. S. Carpenter, S. J. Zheng, R. F. Zhang, S. C. Vogel, I. J. Beyerlein, and N. A. Mara, “Thermal stability of Cu–Nb nanolamellar composites fabricated via accumulative roll bonding,” Philos. Mag., vol. 93, no. 7, pp. 718–735, 2013.

[7] S. Zheng, I. J. Beyerlein, J. S. Carpenter, K. Kang, J. Wang, W. Han, and N. A. Mara, “High-strength and thermally stable bulk nanolayered composites due to twin-induced interfaces,” Nat. Commun., vol. 4, p. 1696, Apr. 2013.

[8] S.-B. Lee, J. E. LeDonne, S. C. V. Lim, I. J. Beyerlein, and A. D. Rollett, “The heterophase interface character distribution of physical vapor-deposited and accumulative roll-bonded Cu–Nb multilayer composites,” Acta Mater., vol. 60, no. 4, pp. 1747–1761, Feb. 2012.

[9] J. S. Carpenter, S. C. Vogel, J. E. LeDonne, D. L. Hammon, I. J. Beyerlein, and N. A. Mara, “Bulk texture evolution of Cu–Nb nanolamellar composites during accumulative roll bonding,” Acta Mater., vol. 60, no. 4, pp. 1576–1586, Feb. 2012.

[10] J. S. Carpenter, R. J. McCabe, S. C. Vogel, A. D. Rollett, N. A. Mara, and I. J. Beyerlein, “Process Parameter Influence on Texture Evolution in Cu-Nb Multilayer Composites Fabricated via Accumulative Roll Bonding,” Metall. Mater. Trans. A, vol. xx, p. xxx, 2014.

[11] C. A. Bronkhorst, J. R. Mayeur, I. J. Beyerlein, H. M. Mourad, B. L. Hansen, N. A. Mara, J. S. Carpenter, R. J. McCabe, and S. D. Sintay, “Meso-Scale Modeling the Orientation and Interface Stability of Cu/Nb-Layered Composites by Rolling,” JOM, vol. 65, no. 3, pp. 431–442, Mar. 2013.

[12] B. L. Hansen, J. S. Carpenter, S. D. Sintay, C. A. Bronkhorst, R. J. McCabe, J. R. Mayeur, H. M. Mourad, I. J. Beyerlein, N. A. Mara, S. R. Chen, and G. T. Gray III, “Modeling the texture evolution of Cu/Nb layered composites during rolling,” Int. J. Plast., vol. 49, pp. 71–84, Oct. 2013.

[13] J. A. Schneider and A. C. N. Jr, “Characterization of plastic flow and resulting microtextures in a friction stir weld,” Metall. Mater. Trans. B, vol. 35, no. 4, pp. 777–783, Aug. 2004.

[14] R. S. Mishra and Z. Y. Ma, “Friction stir welding and processing,” Mater. Sci. Eng. R Rep., vol. 50, no. 1–2, pp. 1–78, Aug. 2005.

[15] C. Schneider, T. Weinberger, J. Inoue, T. Koseki, and N. Enzinger, “Characterisation of interface of steel/magnesium FSW,” Sci. Technol. Weld. Join., vol. 16, no. 1, pp. 100–107, Jan. 2011.

[16] I. Galvão, R. M. Leal, A. Loureiro, and D. M. Rodrigues, “Material flow in heterogeneous friction stir welding of aluminium and copper thin sheets,” Sci. Technol. Weld. Join., vol. 15, no. 8, pp. 654–660, Nov. 2010.

[17] Y. F. Sun, H. Fujii, N. Takaki, and Y. Okitsu, “Microstructure and mechanical properties of dissimilar Al alloy/steel joints prepared by a flat spot friction stir welding technique,” Mater. Des., vol. 47, pp. 350–357, May 2013.

[18] W. C. Oliver and G. M. Pharr, “Measurement of hardness and elastic modulus by instrumented indentation: Advances in understanding and refinements to methodology,” J. Mater. Res., vol. 19, no. 01, pp. 3–20, 2004.

[19] R. D. Doherty, D. A. Hughes, F. J. Humphreys, J. J. Jonas, D. J. Jensen, M. E. Kassner, W. E. King, T. R. McNelley, H. J. McQueen, and A. D. Rollett, “Current issues in recrystallization: a review,” Mater. Sci. Eng. A, vol. 238, no. 2, pp. 219– 274, Nov. 1997.

[20] A. Arora, Z. Zhang, A. De, and T. DebRoy, “Strains and strain rates during friction stir welding,” Scr. Mater., vol. 61, no. 9, pp. 863–866, Nov. 2009.

[21] Z. W. Chen and S. Cui, “Strain and strain rate during friction stir welding/processing of Al-7Si-0.3Mg alloy,” IOP Conf. Ser. Mater. Sci. Eng., vol. 4, no. 1, p. 012026, Aug. 2009.

[22] G. Buffa, J. Hua, R. Shivpuri, and L. Fratini, “A continuum based fem model for friction stir welding—model development,” Mater. Sci. Eng. A, vol. 419, no. 1–2, pp. 389–396, Mar. 2006.

[23] W. Yang and H. Wang, “Mechanics modeling for deformation of nano-grained metals,” J. Mech. Phys. Solids, vol. 52, no. 4, pp. 875–889, Apr. 2004.

[24] H. Van Swygenhoven, A. Caro, and D. Farkas, “A molecular dynamics study of polycrystalline fcc metals at the nanoscale: grain boundary structure and its influence on plastic deformation,” Mater. Sci. Eng. A, vol. 309–310, pp. 440–444, Jul. 2001.

[25] H.-H. Fu, D. J. Benson, and M. A. Meyers, “Analytical and computational description of effect of grain size on yield stress of metals,” Acta Mater., vol. 49, no. 13, pp. 2567–2582, 2001.

[26] V. Yamakov, D. Wolf, S. R. Phillpot, A. K. Mukherjee, and H. Gleiter, “Dislocation processes in the deformation of nanocrystalline aluminium by molecular-dynamics simulation,” Nat. Mater., vol. 1, no. 1, pp. 45–49, Sep. 2002.

[27] M. A. Meyers, A. Mishra, and D. J. Benson, “Mechanical properties of nanocrystalline materials,” Prog. Mater. Sci., vol. 51, no. 4, pp. 427–556, May 2006.

[28] R. Raj and M. F. Ashby, “On grain boundary sliding and diffusional creep,” Metall. Trans., vol. 2, no. 4, pp. 1113–1127, Apr. 1971.

[29] J. S. Carpenter, X. Liu, A. Darbal, N. T. Nuhfer, R. J. McCabe, S. C. Vogel, J. E. LeDonne, A. D. Rollett, K. Barmak, I. J. Beyerlein, and N. A. Mara, “A comparison of texture results obtained using precession electron diffraction and neutron diffraction methods at diminishing length scales in ordered bimetallic nanolamellar composites,” Scr. Mater., vol. 67, no. 4, pp. 336–339, Aug. 2012.

[30] X. Liu, N. T. Nuhfer, A. D. Rollett, S. Sinha, S.-B. Lee, J. S. Carpenter, J. E. LeDonne, A. Darbal, and K. Barmak, “Interfacial orientation and misorientation relationships in nanolamellar Cu/Nb composites using transmission-electron- microscope-based orientation and phase mapping,” Acta Mater., vol. 64, pp. 333– 344, Feb. 2014.

[31] Y. G. Kim, H. Fujii, T. Tsumura, T. Komazaki, and K. Nakata, “Three defect types in friction stir welding of aluminum die casting alloy,” Mater. Sci. Eng. A, vol. 415, no. 1–2, pp. 250–254, Jan. 2006.

[32] L. E. Murr, Y. Li, R. D. Flores, E. A. Trillo, and J. C. McClure, “Intercalation vortices and related microstructural features in the friction-stir welding of dissimilar metals,” Mater. Res. Innov., vol. 2, no. 3, pp. 150–163, 1998.

[33] Y. Li, L. E. Murr, and J. C. McClure, “Solid-state flow visualization in the friction-stir welding of 2024 Al to 6061 Al,” Scr. Mater., vol. 40, no. 9, pp. 1041– 1046, Apr. 1999.

CHAPTER III

MAINTAINING THE NANOLAMELLAR STRUCTURE OF ACCUMULATIVE

ROLL BONDED PLATES DURING FRICTION STIR WELDING2

3.1 Abstract

Copper Niobium (CuNb) nanolamellar composite panels were friction stir welded

(FSW) to evaluate the stability of the lamellar structure. The lamellar structure of the parent material (PM) was maintained during a single pass of the FSW but layer structure was lost during a second pass on the reverse side. The ultimate tensile strength for the single pass FSW was 606 MPa and 394 MPa for the double pass FSW. Digital image correlation (DIC) was used to spatially resolve the localized deformation during tensile testing of the inhomogeneous FSW microstructure. The results indicate that through careful selection of FSW parameters, the structure and strength can be maintained during joining of nanolamellar composites.

______

2 A version of this chapter has been submitted for publication to Metallurgical and Materials Transactions A as: Cobb, J.C., Schneider, J.S., Carpenter, J.S., Lovato, M.L., Liu, C., Vachhani, S., Mara, N.A., “Maintaining the nanolamellar structure of accumulative roll bonded plates during friction stir welding”

3.2 Introduction

Copper Niobium (CuNb) nanolamellar composites, consisting of alternating layers of Cu and Nb, have been the subject of significant research into their fabrication and control of their properties. As compared to the pure coarse-grained constituent materials, these nano-lamellar materials have shown improved strength at room temperature [1][2][3], and elevated temperature [4] [5], as well as increased radiation damage resistance [6][7][8][9]. Mechanical properties of CuNb lamellar composites are reported to increase over that of the bulk materials as the individual layer thickness (h) decreases into the sub-micron range and peak at an thickness, h, of approximately 2 nm

[1]. Thus the mechanical properties can be tailored by controlling the individual layer thickness [1][4].

In this study, CuNb nano-lamellar composites were fabricated via accumulative roll bonding (ARB) of alternating sheets of Cu and Nb. As compared to other methods of fabrication, such as physical vapor deposition PVD [1], ARB produces panels large enough for commercial scale production of components [8] [10]. To realize the commercial production of a component using ARB fabricated CuNb nano-lamellar composites, joinability represents a significant challenge. The enhanced properties of

CuNb nano-lamellar composites emanate from the bulk nanolamellar morphology and local interface character. As such, it is desirable to maintain a nanolamellar structure in the post weld microstructure. Figure 3.1 shows the resulting structure of an electron beam welded (EBW) joint of a CuNb lamellar composite with a 20µm initial average layer thickness [11]. As can be observed in Figure 3.1a, there is a loss of the lamellar

structure in the weld region. A close up of this region of the joint in Figure 3.1b, shows a dendritic Nb structure in a Cu solution which formed during the solidification.

Figure 3.1 a) Optical microscopy image of an electron beam welded CuNb nanolamellar composite overview. (b) Higher magnification of the boxed region in (a) shows the edge of the weld melt zone where a dendritic Nb structure has formed within a Cu matrix.

The FSW process, illustrated in Figure 3.2a, joins material in the solid state by plunging a rotating tool into the butting surfaces of the material to be joined and traversing the rotating tool along the weld length. A backing anvil supports the weld

panel to counteract the downward forces imposed by the tool. Tool rotation results in

severe plastic deformation which “stirs” the faying surfaces of the material together [12]

[13]. This results in a non-symmetric weld zone in which two sides are referred to as either the advancing side (AS), where tool rotation and travel direction align, or the

retreating side (RS), where tool rotation and travel direction are opposed. The resulting

transverse FSW macrostructure is illustrated in Figure 3.2b. The FSW nugget is separated from the parent material (PM) by two regions with varying microstructures.

The first is a heat affected zone (HAZ) in which the material experiences an increase in

temperature without modification of the macrostructure. The next is a thermal-

mechanical affected zone (TMAZ) where the heated material undergoes some degree of

mechanical deformation. In monolithic bulk materials, the FSW nugget is characterized

by refined, equiaxed grains.

Figure 3.2 (a) FSW diagram showing the process parameters, workpiece material, tool, advancing and retreating sides. (b) A schematic of the transverse view of a FSW which contains the weld nugget, thermo-mechanically affected zone (TMAZ), heat affected zone (HAZ), and parent material (PM).

A kinematic model based on slip-line theory has been developed to calculate

shear strains () based on the tool rotation (ω), tool travel (V), and stir zone radius approximated by the pin radius (R) as given in Equation 3.1 [14].

� � � = (3.1) �

This approach assumes that the material adheres to the tool during the entire rotation for the calculations of in-plane shear strain experienced by the material [14].

Equation 1 derives from the similarity of the solid state flow field around the FSW tool, using wire marker studies define the flow paths [15], to that of slip lines illustrated in

Figure 3.3 [16]. As the material enters on the AS, it traverses a longer arc around the tool than material entering on the RS. This can be used to explain the differences in the metallurgical evolution of the FSW macrostructure due to changes in shear strain during the non-symmetric FSW process.

Figure 3.3 Schematic of plan view of FSW zone in which material enters in slip lines around the FSW tool [after 16].

The material travels from right to left while the tool rotates in a clockwise direction.

This estimation of shear strain the in the FSW process fails to capture the out of

plane strain paths of the through thickness material flow [14]. Previous work [17] showed

that the strain input by the FSW process changes the PM microstructure by further

refining h and misorienting the layer directions [17]. All welds had areas which 35

experienced high strain input during the FSW process, especially on the AS that resulted in the initial 200 nm thick layers being thinned to approximately 10 nm prior to forming equiaxed nano-grain regions with an average diameter of 7 nm. This equiaxed nanostructured region exhibited an increased hardness of 20% over an unwelded CuNb material with layer thickness of h = 7 nm. However, no FSW were made that were volumetrically consolidated [17]. In this study, FSW panels were made without volumetric defects unlike in [17] which allowed the mechanical properties and microstructure to be evaluated.

3.3 Experimental

FSW was performed on ARB CuNb nano-lamellar panels with a nominal panel thickness of 0.5 mm with an individual layer thickness, h, of nominal 200 nm. Details of the ARB panel fabrication are documented elsewhere [17]. A single piece FSW tool was machined from a tungsten- 1% lanthanide (W-1%La2O3) alloy with a flat 6.4 mm shoulder diameter, 0.8 mm pin diameter, and 0.5 mm pin length. FSWs were made with the tool tilted 2⁰ on a machine operated in force control with a load of 1500 N. The tool was rotated at 2000 rpm at a travel rate of 6 mm/min for the 40 mm long FSWs. Due to shortening of the pin, due to wear, during initial FSWs to establish process parameters, the final length of the pin used for the panels was approximately 0.2 mm. Thus in the first set of single pass FSWs, there was a lack of penetration (LOP) on the root side due to insufficient pin length. The second set of FSW panels were made using a double pass on the top and bottom side to eliminate the LOP as illustrated in Figure 3.4. The double pass FSW was configured to place the AS from the first pass FSW on the opposite side on the second pass. Although the pin of the tool was slightly shorter than half the 36

workpiece, the material directly below the pin is also stirred which results in a complete through-thickness FSW after two passes.

Figure 3.4 Diagram showing the transverse view layout of the double pass weld.

The FSW panels were cross-sectioned transverse to the welding direction using a low speed diamond saw. Metallographic samples were mounted in a low heat epoxy resin and metallurgically ground, with a final vibratory polish in a colloidal silica solution prior to imaging and hardness measurements. Microstructural images were obtained using a FEI (Hillsboro, OR) Inspect F Scanning Electron Microscope (SEM) operated at

10-20 keV. Hardness data was collected using an Agilent Nano Indenter XP-30 system equipped with a Berkovich tip calibrated using a SiO2 standard [18]. Average hardness was calculated from the load vs. displacement data collected in continuous stiffness mode for indentation depths between 200 and 1100 nm [18].

Electro discharge machining (EDM) was used to cut the tensile dogbones to the dimensions shown in Figure 3.5. The tensile tests were conducted with the load applied transverse to the FSW direction. Prior to tensile testing of the single pass FSW, the root side LOP region was removed by grinding to a final surface finish of 800 grit. Although 37

the crown side of the single pass FSW had a slight depression from the tool marks, this side was not polished due to the thinness of the specimen. This resulted in a slightly reduced cross-sectional area of the single pass FSW tensile specimen. This reduced area was used for the stress calculations. Due to the small volume, only one tensile specimen was prepared from the single pass FSW. The double pass FSW test specimens were polished smooth on both the root and crown surfaces to remove the tooling marks using

SiC 1200 grit paper. Two tensile specimens were prepared from the double pass FSW.

Figure 3.5 Tensile dogbone dimensions with panel transverse direction (TD) and rolling direction (RD) labeled.

The FSW was made along the RD of the ARB panels. All dimensions are in mm.

Tensile tests were performed on an Instron model 1125, with a 1 kN load cell, at a

constant crosshead velocity of 0.2 mm/min. Digital image correlation (DIC) was used as

a 10 mm non-contact extensometer to record the overall specimen elongation in addition

to spatially resolved strains. Prior to tensile testing, a speckle pattern was applied to the

crown side of the FSW specimens for collection of DIC data. A CCD camera, with a

resolution of 1628 × 1236 pixels, recorded the speckle images at a frame rate of 1fps.

The speckle images were processed using Correlated Solutions Vic2D 2013 software to

obtain the strain field over the sample surface.

3.4 Results and Discussion

The mechanical properties calculated from the tensile test data of single and

double pass FSWs are summarized in Table 1 along with the PM properties from [3].

The single pass FSW had an ultimate tensile strength (UTS) of 606 MPa or a 14% reduction in strength from the PM. The double pass FSWs had an average tensile strength of 394 MPa which is a 43% reduction from PM strength. Reduction in strength within a welded region is common, and the difference in strength between the single and double pass is attributed to changes in microstructure resulting from the second pass

FSW. The double pass FSWs were noted to fail on the 2nd pass AS of the FSW nugget in

the TMAZ. The single pass FSW had heavy localized deformation in the AS TMAZ, but

failed in the weld nugget on the AS.

Table 3.1 Tensile properties of PM, single pass FSW, and double pass FSW.

UTS (MPa) YS (MPa) 0.2% offset Elongation to Failure (%)

CuNb ARB PM [27] 687 518 2.62 Single pass FSW 606 536 1.40 Double pass FSW Specimen #1 404 377 1.53 Specimen #2 383 383 1.87

As the microstructure varies across the FSW zone, strain incompatibilities can result in reduced bulk ductility. Although the bulk strain measurements listed in Table 1 indicate a reduction in ductility for the FSWs, the DIC results show highly localized strains. For the single and double pass FSWs, DIC indicated an average bulk elongation to failure of about 1.5%. Figure 3.6 shows that on the AS TMAZ region, localized strains on the order of 9% and 20% were accommodated prior to failure for the single pass and double pass, respectively.

Figure 3.6 DIC xx strain field maps showing large local strain on the AS for single (a) and double pass (b) FSWs.

Corresponding scale-bar shows the strain with units of mm/mm.

Further evidence of the ductility of the FSW region can be observed in the SEM secondary electron image of the fracture surfaces of the double pass FSW in Figure 3.7.

Necking, expected from regions of large localized strain from DIC, can be seen on both edges of the FSW. Dimples can be observed which also indicate a ductile type failure.

Figure 3.7 SEM secondary image of the center of the double pass FSW fracture surface showing ductile fracture features.

Metallographic specimens prepared from the PM, single and double pass FSW

were imaged to document the microstructure. Figure 3.8 shows an SEM backscattered

image of the PM ARB CuNb nano-lamellar with the light regions corresponding to the

Nb layers and the dark regions corresponding to the Cu layers.

Figure 3.8 SEM backscatter image of PM microstructure showing the nano-lamellar structure.

The lighter layers are Nb and the darker layers are Cu.

The high aspect ratio of the FSWs, makes it impractical to show the entire weld nugget region while viewing the microstructure in any detail. Because of this images shown in Figures 3.9-3.12 are from subsections of the FSW regions with their approximate locations in the weld region shown. Figure 3.9 is a transverse view of the center of the single pass FSW nugget prior to removal of the root side LOP. The varying strain input during the FSW process caused a non-uniform change in layer orientation, and a non-uniform reduction of h in the weld nugget.

Figure 3.9 (a) Location of SEM image in FSW region. (b) SEM backscatter image of the single pass FSW nugget before removal of root side lack of penetration (LOP) region for the tensile test specimens.

Figure 3.10 shows a higher magnification image of the region of the highest shear strain, near the AS edge of the FSW nugget. A significant reduction in h and misorientation are noted in this region; however, the layers of the material can still be resolved throughout the single pass FSW.

Figure 3.10 (a) Location of SEM image in FSW region for the single pass FSW. (b) SEM backscatter image of the AS of the FSW nugget showing the most extreme changes in direction and layer refinement.

Although the double pass FSW was found to retain a layered structure in the

majority of the FSW nugget, Figure 3.11 shows the layered structure is no longer resolvable near the AS of the last pass. The loss of the visible layered structure is attributed to the additional strain input by the second pass of the FSW tool. Chapter II showed the loss of visible layers coupled with measured hardnesses of 6.0 GPa or higher as the CuNb nano-layers fragment into equiaxed grain structures composed of Cu and Nb grains of approximately 7 nm diameter [17].

Figure 3.11 (a) Location of SEM image in FSW region for the double pass FSW. (b) SEM image of transverse section of the double pass FSW showing loss of visible layered structure, in the SEM, at the AS interface between the nugget and the TMAZ region.

Figure 3.12 shows the hardness data in the double pass FSW corresponding to the different microstructural regions, with maximum hardness occurring on the AS of the second pass. These harder regions had hardness in excess of 6 GPa and correlated with regions of the weld where the layered structure could not be resolved. This hardness is similar to that found in a previous study where transmission electron microscopy was used to resolve the fragmentation of the nano layers into equiaxed nanograins of about 7 nm diameter [17]. Three indents spanning the thickness of the TMAZ region, shown in the far right of Figure 3.12, averaged a hardness value of 1.8 GPa with a standard deviation of 0.16 GPa, 30% softer than the 2.5 GPa of the PM. TMAZ region is where the tensile failure occurred, and the lower hardness in this region correlates with the large reduction in ultimate tensile stress (UTS) from the PM. The reduction in hardness could 45

be attributed to the elevated temperature during both of the FSW passes, as annealing studies have shown softening in CuNb nanolamellar composites above 773 K (500° C) due to coarsening of the layers [5].

Figure 3.12 (a) Location of SEM image in FSW region for the double pass. (b) FSW Hardness map across the double pass FSW at the AS nugget/TMAZ interface. Hardness values are shown in GPa.

Layers of tortuous morphology and vortices, similar to those seen in Figure 3.10

are seen throughout the FSW regions with alteration in microstructure relative to the PM

becoming progressively more severe closer to the AS for both single and double pass

FSWs. This is attributed to the larger strains experienced on the AS. The flow paths of

the CuNb FSW display extreme bending of the layers and formation of vortices.

Although these features are not typically reported in similar material FSWs, they have

been noted when welding dissimilar materials. Early FSW studies of dissimilar materials

often reported swirls and vortices in the microstructure and were speculated to be due to a 46

chaotic, disorderly, turbulent flow [19-21]. In later studies the turbulent flow theory was discounted as various marker studies showed evidence of an orderly flow of the material

[22-24]. The strain input needed to cause the loss of a visible layered structure is unknown; however, an approximation can be calculated from Equation 3.1 [14]. Using the FSW parameters of this experiment for a complete tool rotation, assuming no slipping of the tool and material, the nominal shear strain is estimated to be 838. This is the predicted in plane strain and corresponds to the material entering on the most extreme point on the AS. Not all material rotating around the tool experiences the same amount of strain as was illustrated in Figure 3.3. If the material enters the FSW zone towards the

AS it may experience 75% of the tool rotation, which reduces the predicted in-plane shear strain to approximately 630. In comparison as the material enters the FSW zone from the RS, it may only experience 25% of the tool rotation for an estimated in-plane shear strain of 210.

The microstructural evolution observed with FSW of CuNb shares some characteristics with those reported in high pressure torsion (HPT) of ARB nano-lamellar

CuNb with h = 200 nm [25]. This severe torsional deformation process also exposes the material to large shear strains. With progressing levels of strain in the HPT process, the layers first become wavy, then misoriented (including folds, swirls, and vortices), followed by evolution into an equiaxed nanograin structure [25]. The layered structure was maintained up to shear strains in excess of 3400 before fragmenting into an equiaxed nanograin structure. It is noted that the strain to reach an equiaxed nanograin structure is higher for HPT than the strain estimated, using Equation 1, for the double pass FSW, which would be a maximum of 1676, or double the maximum predicted strain of 838 for

the single pass FSW. Although both HPT and FSW input a rotational shearing strain, the difference in the strain rate, geometry, tilt, tool travel, and weld temperature could have contribute to this difference.

A closer look at the layers in the FSW nugget show small length-scale waviness in the layers and vortices and swirls that resemble Kelvin-Helmholtz instabilities. A

Kelvin-Hemholtz instability typically occurs in fluids when a velocity differential causes a shear layer at the interface to become unstable causing turbulence and mixing [26].

Figure 3.13 shows an example of layer waviness from the center of the double pass FSW.

Figure 3.13 Waviness swirls and vortices in the FSW region of a double pass weld in the CuNb nanolayers.

In this study, as the strain increased in the FSW nugget (moving from the RS

toward the AS) the waviness, and vortices increased. In the double pass FSW the layers

have fragmented completely on the AS. Similar multi-layer evolution has also been

reported in CuAu nanolayers where cyclic sliding of a spherical nanoindenter tip was

used to deform the material [27]. As the cycle count increased the layer progressed from 48

straight, to wavy, then forming vortices that were also reported to resemble a Kelvin-

Helmholtz instability, before forming a final microstructure of equiaxed nanograins. This same type of layer evolution of waviness, swirls, vortices have also been reported in HPT

CuNb nanolayers [25][28] although its deformation profile was attributed to folding mechanisms found in plate tectonics. It is interesting to note that all three of these processes share a common velocity component, shown illustrated in Figure 3.14, of the top layers moving faster than the layers beneath them.

Figure 3.14 Common velocity profile of FSW, HPT, and cyclic sliding.

The cyclic sliding study imparts this velocity gradient by dragging a spherical indenter over the top of the material while keeping the bottom surface stationary. The flow paths in HPT are due solely to rotation as the top anvil turns on a stationary lower anvil. Assuming sticking of the metal specimen to the anvils, this would impart a velocity gradient between the top and bottom anvils. During the FSW process, the metal experiences a similar rotational flow as in HPT, as the rotating FSW tool would impart

similar flow as the rotating anvil in HPT. It should be noted for FSW there is also addition flow component which is an out of plane, through thickness material flow

[22][23][24]. The common velocity gradient in FSW, HPT, and cyclic sliding, coupled with similar layer evolution that resemble Kelvin-Helmholtz instabilities, suggest that

Kelvin-Helmholtz instabilities may explain the observed microstructural evolution.

A Kelvin-Helmholtz type instability may play an important role in the grain refinement and decreasing the strain needed to form an equiaxed nanograin structure. As the continually increasing strain thins the layers, coupled with the turbulent flow of the instability, the tendency for grain pinch off increases [29] resulting in the formation of equiaxed nanograins found in the double pass FSW in this study and in others [17].

3.5 Conclusions

Two sets of CuNb nano-lamellar composite plates with a layer thickness of h ≈

200 nm were successfully consolidated in a FSW butt joint. The single pass FSW was made only on the top side and the double pass FSW was made on the top and bottom sides. The single pass FSW was found preferable to the double pass FSW since the lamellar structure was maintained along with an associated higher UTS. The weld nuggets for both FSW configurations showed layers that became progressively disordered, and refined in h, toward the AS. This is attributed to the progressively increasing strain experienced from the RS to AS. The addition of the FSW pass on the bottom of the double pass FSW changed the microstructure in a predictable and logical fashion, with the increase in strain due to the extra FSW pass causing further layer disorder. After a single pass FSW, the nano-layered microstructure of the PM remained resolvable, albeit tortuous and refined, while after the second pass FSW regions of the 50

layered structure fragmented into nano, equiaxed grains on the AS of the second FSW pass.

Microstructural evolution in the FSWs was consistent with morphologies seen in

HPT, and cyclic sliding studies performed on nanolamellar composites. Kelvin-

Helmholtz instabilities are consistent with the observed microstructural waviness, folds, and vortices. This may precede the formation of equiaxed grains through a grain pinch off mechanism. This study shows that the mechanical properties of the ARB plates joined by FSW are highly dependent on the processing parameters that control the shear strain input and ultimate post FSW microstructure.

3.6 Acknowledgements

This work is supported by the Los Alamos National Laboratory Directed

National Laboratory, an affirmative action equal opportunity employer, is operated by

3.7 References

[1] A. Misra, J. P. Hirth, and R. G. Hoagland, “Length-scale-dependent deformation mechanisms in incoherent metallic multilayered composites,” Acta Mater., vol. 53, no. 18, pp. 4817–4824, Oct. 2005.

[2] N. A. Mara, D. Bhattacharyya, P. Dickerson, R. G. Hoagland, and A. Misra, “Deformability of ultrahigh strength 5 nm Cu∕Nb nanolayered composites,” Appl. Phys. Lett., vol. 92, no. 23, p. 231901, 2008.

[3] I. J. Beyerlein, N. A. Mara, J. S. Carpenter, T. Nizolek, W. M. Mook, T. A. Wynn, R. J. McCabe, J. R. Mayeur, K. Kang, S. Zheng, J. Wang, and T. M. Pollock, “Interface-driven microstructure development and ultra high strength of bulk nanostructured Cu-Nb multilayers fabricated by severe plastic deformation,” J. Mater. Res., vol. 28, no. 13, pp. 1799–1812, 2013.

[4] J. S. Carpenter, S. J. Zheng, R. F. Zhang, S. C. Vogel, I. J. Beyerlein, and N. A. Mara, “Thermal stability of Cu–Nb nanolamellar composites fabricated via accumulative roll bonding,” Philos. Mag., vol. 93, no. 7, pp. 718–735, 2013.

[5] S. Zheng, J. S. Carpenter, R. J. McCabe, I. J. Beyerlein, and N. A. Mara, “Engineering Interface Structures and Thermal Stabilities via SPD Processing in Bulk Nanostructured Metals,” Sci. Rep., vol. 4, Feb. 2014.

[6] A. Misra, M. J. Demkowicz, X. Zhang, and R. G. Hoagland, “The radiation damage tolerance of ultra-high strength nanolayered composites,” Jom, vol. 59, no. 9, pp. 62–65, 2007.

[7] W. Han, M. J. Demkowicz, N. A. Mara, E. Fu, S. Sinha, A. D. Rollett, Y. Wang, J. S. Carpenter, I. J. Beyerlein, and A. Misra, “Design of Radiation Tolerant Materials Via Interface Engineering,” Adv. Mater., vol. 25, no. 48, pp. 6975– 6979, 2013.

[8] J. S. Carpenter, S. C. Vogel, J. E. LeDonne, D. L. Hammon, I. J. Beyerlein, and N. A. Mara, “Bulk texture evolution of Cu–Nb nanolamellar composites during accumulative roll bonding,” Acta Mater., vol. 60, no. 4, pp. 1576–1586, Feb. 2012.

[9] C. A. Bronkhorst, J. R. Mayeur, I. J. Beyerlein, H. M. Mourad, B. L. Hansen, N. A. Mara, J. S. Carpenter, R. J. McCabe, and S. D. Sintay, “Meso-Scale Modeling the Orientation and Interface Stability of Cu/Nb-Layered Composites by Rolling,” JOM, vol. 65, no. 3, pp. 431–442, Mar. 2013.

[10] J. S. Carpenter, R. J. McCabe, S. J. Zheng, T. A. Wynn, N. A. Mara, and I. J. Beyerlein, “Processing Parameter Influence on Texture and Microstructural Evolution in Cu-Nb Multilayer Composites Fabricated via Accumulative Roll Bonding,” Metall. Mater. Trans. A, vol. 45, no. 4, pp. 2192–2208, Apr. 2014. 52

[11] J. Carpenter, “Unpublished data from Dr. John Carpenter,” 2013.

[12] R. S. Mishra and Z. Y. Ma, “Friction stir welding and processing,” Mater. Sci. Eng. R Rep., vol. 50, no. 1–2, pp. 1–78, Aug. 2005.

[13] H. Farrokhi, A. Heidarzadeh, and T. Saeid, “Frictions stir welding of copper under different welding parameters and media,” Sci. Technol. Weld. Join., vol. 18, no. 8, pp. 697–702, Nov. 2013.

[14] A. C. Nunes, “Joining of Advanced and Specialty Materials (JASM) Including Affordable Joining of Titanium and Joining Technologies for MMCs,” MS&T, 2006.

[15] J. A. Schneider and A. C. Nunes, “Quantifying the Material Processing Conditions for an Optimized FSW,” 7th Int’l Conf Trends Weld., pp. 253-256, 2005.

[16] W. Johnson, R. Sowerby, and J. Haddow, Plane-Strain Slip-Line Fields: The Theory and Bibliography. New York, NY, USA: American Elsevier publishing company Inc, 1970.

[17] J. Cobb, S. Vachhani, R. M. Dickerson, P. O. Dickerson, W. Han, N. A. Mara, J. S. Carpenter, and J. Schneider, “Layer Stability and Material Properties of Friction-Stir Welded Cu–Nb Nanolamellar Composite Plates,” Mater. Res. Lett., pp. 1–6, May 2014.

[19] L. E. Murr, Y. Li, R. D. Flores, E. A. Trillo, and J. C. McClure, “Intercalation vortices and related microstructural features in the friction-stir welding of dissimilar metals,” Mater. Res. Innov., vol. 2, no. 3, pp. 150–163, 1998.

[20] L. E. Murr, E. A. Trillo, R. D. Flores, and J. C. McClure, “Solid-State Flow Associated With the Firction_Stir Welding of Dissimilar Metals.” The Minierals Metals & Materials Society, 1999.

[21] Y. Li, E. A. Trillo, and L. E. Murr, “Friction-stir welding of aluminum alloy 2024 to silver,” J. Mater. Sci. Lett., vol. 19, no. 12, pp. 1047–1051, 2000.

[22] J. Schneider, R. Beshears, and A. C. Nunes, “Interfacial sticking and slipping in the friction stir welding process,” Mater. Sci. Eng. A, vol. 435–436, pp. 297–304, Nov. 2006.

[23] T. U. Seidel and A. P. Reynolds, “Visualization of the material flow in AA2195 friction-stir welds using a marker insert technique.pdf,” Metall. Mater. Trans. A, vol. 32, no. 11, pp. 2879–2884, Nov. 2001.

[24] K. Colligan, “Material Flow Behavior during Friction Stir Welding of Aluminum,” Weld. J., vol. 78, no. 7, pp. 229–237, 1999.

[25] H. E. Ekiz, “Microstructural evolution and phase stability of Cu/Nb nanolaminates subjected to severe plastic deformation by high pressure torsion,” University of Illinois at Urbana-Champaign, 2015.

[26] H. G. Lee and J. Kim, “Two-dimensional Kelvin–Helmholtz instabilities of multi- component fluids,” Eur. J. Mech. - BFluids, vol. 49, pp. 77–88, Jan. 2015.

[27] Z.-P. Luo, G.-P. Zhang, and R. Schwaiger, “Microstructural vortex formation during cyclic sliding of Cu/Au multilayers,” Scr. Mater., vol. 107, pp. 67–70, Oct. 2015.

[28] E. H. Ekiz, T. G. Lach, R. S. Averback, N. A. Mara, I. J. Beyerlein, M. Pouryazdan, H. Hahn, and P. Bellon, “Microstructural evolution of nanolayered Cu–Nb composites subjected to high-pressure torsion,” Acta Mater., vol. 72, pp. 178–191, Jun. 2014.

[29] R. D. Doherty, D. A. Hughes, F. J. Humphreys, J. J. Jonas, D. J. Jensen, M. E. Kassner, W. E. King, T. R. McNelley, H. J. McQueen, and A. D. Rollett, “Current issues in recrystallization: a review,” Mater. Sci. Eng. A, vol. 238, no. 2, pp. 219– 274, Nov. 1997.

CHAPTER IV

TEXTURE ANALYSIS OF FRICTION STIR WELDED COPPER NIOBIUM

NANOLAMELLAR COMPOSITES

4.1 Introduction

The second objective of this study proposed that the nano layered structure of the

CuNb NLC may be able to provide insight into the FSW process by acting as a marker for the material movement. While Chapter 3 presented detailed microscopy imaging of the layer deformation following FSW, this chapter presents an effort to characterize the corresponding texture within the FSW nugget. Because the texture of the material and the maintenance of a layered nanostructure are dependent on the hot working history, the reorientation and refinement of the layers post joining may provide insight into the material flow during the FSW process. Previous studies in the literature have shown the texture of material fabricated via ARB is dependent on its processing history [1] [2]. This section discusses efforts undertaken to characterize the texture within the layers in the

FSW nugget. Various spectrometers and detectors were used in transmission electron microscopy (TEM) and scanning electron microscopes (SEM) to obtain information about the elemental nature and grain orientation within the nano layers. Energy dispersive spectroscopy (EDS) preformed in a TEM was used to obtain elemental maps to confirm that layers were maintained post FSW in regions of interest. All electron backscatter diffraction analyses were conducted in an SEM. Conventional electron

backscatter diffraction (cEBSD) of bulk specimens was used to obtain texture information within layer thickness > 150 nm. For thinner layers that could not be resolved using cEBSD, TEM foils were analyzed using transmission Kikuchi diffraction

(TKD) within an SEM., also known as transmission electron backscatter diffraction

(tEBSD).

4.1.1 Energy dispersive X-ray spectroscopy (EDS)

EDS is a spectroscopy method that can be used in electron microscopes, such as the TEM or SEM, to obtain elemental information of a specific location. For EDS the beam of an electron microscope is used to ionize an atom by ejecting an inner core-shell electron. An electron from a weaker outer shell takes its place, and this process is repeated until the last atom falls into a core shell from the conduction band. Each of these electron transitions has the possibility to generate a characteristic X-ray. A detector in microscope is used to analyze the energy of these characteristic X-rays for elemental analysis [3]. Using the scanning mode of a STEM or SEM, an EDS system can be used to obtain maps of the elemental distribution. This provides a qualitative method to verify the extent to which the elementally distinct layers were maintained.

4.1.2 Conventional electron backscatter diffraction (cEBSD)

Collection of electron backscatter diffraction (EBSD) data can be utilized for mapping the grain orientation in crystalline materials [4] [5] [6] [7] . A typical EBSD setup is illustrated in Figure 4.1 [8] . It consists of tilting a flat polished sample at a 70 degree angle normal to the electron beam of a scanning electron microscope (SEM). The sample should be highly polished and free of surface defects or oxidation films as these

artifacts will interfere with the backscatter process. As the electron beam interacts with the sample, it experiences inelastic collisions with the atoms of the sample and is incoherently scattered. This is followed by some of the scattered electrons being reflected back and elastically channeled outward along crystalline planes. These backscattered electrons give rise to Kikuchi band patterns which can be used to index the crystal structure and orientation. These electron backscatter patterns (EBSP) are recorded by a digital camera behind a phosphorus screen as it is illuminated by the electrons striking the screen [8].

Figure 4.1 Typical EBSD setup showing the primary electron beam focused on the sample, and detector for collection of the electron backscatter patterns (EBSP).

Since EBSD patterns are dependent on the lattice parameters and of the crystalline material, it is possible to distinguish between phases [9] [10] as well as determine the individual grain size and orientation [10] [11] [12] . High resolution EBSD techniques are also being explored to measure lattice strains in a material [13] [14] .

Because of scattering as the beam interacts with the sample, the interaction volume with the sample is larger than the original beam size [4] [15]. Typically the ability to analyze an individual grain relates to the interaction volume relative to the grain size. If the interaction volume is larger than the grain, additional patterns will be collected from neighboring grains [1], [3], [4] [12]. Therefore, if the interaction volume is reduced the EBSD resolution can be increased. The interaction volume is affected by several variables including the electron beam probe diameter, the atomic number of the sample, and the electron beam accelerating voltage. Probe diameters are related to the emitter in the SEM with the smallest diameter being generated by a field emission emitter and the largest by a tungsten emitter. At a 10 kV accelerating voltage and a probe current of 5 pA a field emission emitter is probe size is 3 nm and a tungsten emitter 20 nm [16] .

The interaction volume also varies by the atomic number of the sample, with a lower atomic number having a larger interaction volume [4] [15] [7]. The typical 70 degree tilt of the sample also affects the interaction volume, as shown in Figure 4.2, by stretching the interaction volume perpendicular to the tilt axis of an EBSD sample [4].

Figure 4.2 Schematic illustrating an electron beam interaction volume for an EBSD setup for a material with a 30nm grain size [4].

As can be observed, EBSD patterns will also be generated from grains surrounding the initial grain in initial contact with the beam.

The interaction volume in a SEM can be reduced by using a lower accelerating voltage. However a reduction in accelerating voltage causes increased sensitivity to surface contamination and beam drift, both of which cause a degradation in pattern quality [4]. This coupled with current EBSD pattern detection technology limits the use of EBSD to a grain size on the order of 100 nm when using a field emission detector [7]

[15]. EBSD of highly strained materials present an additional challenge as high dislocation densities cause a degradation of patterns which also result in poor indexing

[12].

4.1.3 Transmission Kikuchi Diffraction (TKD)

In 2012, a new method for obtaining EBSD patterns was reported and shown schematically in Figure 4.3 [17]. Known as transmission Kikuchi diffraction (TKD), it records patterns transmitted from the back side of an electron transparent foil used in transmission electron microscopy (TEM) [18]. In addition to a difference in sample thickness, the main difference in setup between cEBSD and TKD is the orientation of the sample with respect to the electron beam and EBSD detector. Since the patterns are recorded from the bottom of the sample, the sample is tilted in the opposite direction of cEBSD and documented at negative α angles. Several different α tilt angles are reported in literature with most ranging between 0 to -20 degrees[17] [18] [19], but tilts as large as

-60 degrees have been reported [20].

Figure 4.3 Infrared image of a typical TKD setup in a SEM chamber [17] with the tilt angle α being a negative angle compared to a cEBSD setup.

TKD is reported to have an order of magnitude better resolution of grains than cEBSD due to the decreased interaction volume with the thin TEM foil [19] [21]. A resolution of about 10 nm has been reported in nanocrystalline nickel samples produced by electrodeposition [22]. Because the TKD patterns originate from the bottom 5 - 10 nm of the sample, analysis of samples with overlapping grains through the thickness is possible [20].

4.2 Experimental Procedure

4.2.1 EDS

Figure 4.4 shows the location of three TEM foils (Samples B, C, and D) that were removed from the AS of a transverse section of the single pass FSW for analysis.

Elemental maps were obtained in a Philips CM200 transmission electron microscope

(TEM) using the scanning transmission electron microscope (STEM) capability. The

EDS detector was an EDAX r-TEM. Using EDS elemental maps were made of Sample

C to ensure that distinct layers of Cu and Nb were present.

4.2.2 cEBSD

Sample preparation for cEBSD consisted of sectioning the material in approximately the center of the FSW length, perpendicular to the weld direction, mounting the sample in a phenolic resin, followed by metallographically grinding and polishing, with a final vibratory polish using colloidal silica. EBSD scans were made using a FEI Inspect F SEM equipped with an EDAX EBSD system. A total of two scans were made both near the FSW centerline, one the RS of the centerline and one on the AS

of the centerline. The RS scan size was a 10 µm box and was made using a 50 nm step size. The AS scan size was a 66 µm box and was made using a 100 nm step size.

4.2.3 TKD

The three TEM foils shown in Figure 4.4, were milled, lifted out, and thinned using a Hitachi FB-2000A SEM with a dual focused ion beam (FIB). Heavy curtaining of the surface occurred during the FIB thinning process. In an attempt to remove the curtaining and flatten the foil, a Fischione 1040 nanomill was used. However, no visible reduction in curtaining was achieved as the curtaining can still be observed in Figure 4.5.

Figure 4.4 SEI SEM Image of the transverse section of the single pass FSW nugget.

Locations of the TEM foil removal specimens, which were used in the TDK analysis, are shown.

Figure 4.5 SEM secondary electron image of Sample C prepared by FIB process

TKD was performed using three different SEM and EBSD systems due to pattern indexing issues. All SEMs were equipped with a field emission source. Operating conditions of the various systems are summarized in Table 1.

A JEOL JSM-7400F SEM equipped with an EDAX/TSL EBSD system was able to record EBSD patterns of visible high quality, but the software was unable to properly index them. A JEOL 6500F SEM equipped with an Oxford INCA EBSD system was able to record and index patterns, but the SEM was unstable for the long acquisition times required for EBSD scans. A Zeiss SUPRA 40 SEM equipped with an EDAX/TSL EBSD system was not able to initially index patterns. This issue was resolved by setting the focal working distance (WD) on the SEM to 8 mm and the EDAX/TSL software focal

WD to 9 mm. All TKD orientation data presented in this work was generated using the

Supra 40 SEM at a WD of 8mm, with an accelerating voltage of 30 kV, a -15 degree tilt, and using a step size of 5 to 10 nm.

Table 4.1 SEM settings for TKD

JEOL JSM7400F JEOL 6500F Zeiss Supra 40 EBSD system EDAX/ TSL Oxford EDAX/ TSL /software /INCA

kV 15 to 30 20 to 30 30 WD (mm) 3 to 8 3 to 5 5 to 15 Tilt (degrees) -5 to -15 -15 -15

4.3 Results

4.3.1 EDS

EDS mapping of sample C verified the distinct layers of Cu and Nb in the post

FSW nugget as shown in Figure 4.6. Sample C was on the AS of the centerline of the

FSW nugget. The layers in these images range from 20 nm to 340 nm. Note the large variation in layer thickness as compared with the starting PM of nominally 200 nm that was shown in Figure 3.8.

Figure 4.6 STEM - EDS maps of sample C from the AS of the FSW nugget which shows the distinct layers of Cu and Nb.

4.3.2 cEBSD scans

Pole figures from cEBSD scans for Cu and Nb are shown are shown in Figure 4.7 and Figure 4.8, respectively. Similar texture is observed between the PM and the RS of the FSW nugget. This texture weakens when crossing the centerline of the FSW to the

AS. For the Cu, an approximately 1.4 to 5 times random texture is noted for the initial

PM (Figure 4.7d) as well as the RS FSW nugget (Figure 4.7c). The Cu grains on the AS show a weakened intensity in the pole figures dropping to approximately 1 to 1.6 times random. Similar information is shown in Figure 4.8 for the Nb between the PM and the

RS weld nugget in which the texture ranges from 1.4 to 7 times random. The Nb texture shows a similar weakening on the AS reducing to 1 to 1.6 times random.

Figure 4.7 Secondary electron SEM image with red arrows showing the locations of cEBSD scans in the single pass FSW nugget.

Locations are shown by the red arrows. b) Pole figures of the Cu just past the centerline of the FSW on the AS. c) Pole figures of the Cu just before the centerline of the FSW on the RS. d) Cu grains pole figure of the PM. The legend for all pole figures is give as multiples of random.

Figure 4.8 a) Secondary electron SEM image with red arrows showing the locations of cEBSD scans in the single pass FSW nugget. b) Pole figures of the Nb just passed the centerline of the FSW on the AS. c) Pole figures of the Nb just before the centerline of the FSW on the RS. d) Pole figures of the Nb in the PM.

The legend for all pole figures is given as multiples of random.

4.3.3 TKD scans

Figure 4.9 shows the TKD scans for sample B. The phase distribution map is shown in Figure 4.9a with the Cu phase in red and the Nb phase in green. Figure 4.9b shows the Cu grains and their resolved orientations and Figure 4.9c shows the Nb grains and their resolved orientations. The colors in the orientation images are referenced to the stereographic triangle in Figure 4.9d. The results from foils C and D are shown in Figure

4.10 and Figure 4.11, respectively, and are laid out in the same manner as in Figure 4.9.

The images presented have not had a software cleanup of the grains, so areas that appear as black and white speckles in parts c) and d) are regions where the grains were unable to be indexed. 67

Figure 4.9 a) Phase distribution map for sample B with the Cu phase shown in red and the Nb phase in green. b) Orientation map of Cu grains. c) Orientation map of Nb grains. d) A stereographic triangle relating the colors to orientations.

Resolved grain /layer thicknesses range from 50-200 nm with most being in between 70- 140 nm.

Figure 4.10 a) Phase distribution map for sample C with the Cu phase shown in red and the Nb phase in green. b) Orientation map of Cu grains. c) Orientation map of Nb grains. d) A stereographic triangle relating the colors to orientations.

Resolved grain /layer thicknesses range from 70-260 nm with most being approximately 150 nm.

Figure 4.11 a) Phase distribution map for sample D with the Cu phase shown in red and the Nb phase in green. b) Orientation map of Cu grains. c) Orientation map of Nb grains. d) A stereographic triangle relating the colors to orientations.

Resolved grain /layer thicknesses range from 30-450 nm with all except the largest Cu layer ranging between 30-230mm.

The phase distribution maps in Figures 4.10 and 4.11 show distinct layers of the

Cu and Nb. Although a limited number of grains could be resolved, the orientation maps show distinct grain boundaries between the individual grains within a layer. While the individual Cu and Nb layers can still be observed in Figure 9a, which is closer to the centerline of the FSW nugget, more variation in layer thickness is observed. It also appears that the software was able to resolve more grains in this location than in Figures

4.6 and 4.7 which are further towards the AS of the FSW nugget centerline.

4.4 Discussion

4.4.1 cEBSD and TKD Scans

The pole figures from the cEBSD scans in Figure 4.7 and 4.8 show a similar texture in the Cu and Nb layers on the RS with that of the PM. Texture is observed to weaken when crossing from the RS to the AS of a transverse section of the FSW nugget.

This weakening of texture would indicate that the AS of the FSW nugget material has undergone additional strain compared to that experienced by the RS of the FSW nugget. 69

Marker studies have indicated a difference in flow paths and hence strain that material experiences as it enters from the AS vs the RS of the FSW nugget centerline. In earlier FSW tungsten tracer wire studies, a difference in the material flow patterns was noted for material entering on the AS of the FSW tool [23] vs that entering on the RS.

Excerpts from this study are presented in Figure 4.12, and show material that entered on the AS of the tool was scattered in its final deposited position in the wake in the x and y positions. In contrast material the wire entering on the RS of the tool flowed orderly around the tool and deposited in the wake in a similar x and y positions as when it entered. The study interpreted these results of scattered material flow on the AS, indicative of material that has made several rotations around the tool through the thickness of the workpiece resulting in an out of plane flow path.

Figure 4.12 Inverted x-ray radiograph images from the tungsten wire tracer study showing a) the plan view of the wire entering on the RS exiting in an orderly fashion b) a plan view of the wire entering the AS exiting in a scattered fashion in the x and y coordinates c) a transverse view of the wire from (b) showing scattered z coordinates [23].

A more recent in-situ marker which used an embedded spherical tungsten marker

to measure the strain rate also observed material that flowed several times around the

FSW tool [24]. This study described this material as being entrapped in a “flow zone”

where the strain rate was very small (0.21 s-1) due to the measured angular velocity of the material in the flow zone remaining almost constant. This contrasts with the higher measured strain rate of the material as it enters and exits the flow zone with a maximum strain rate of -13.4 s-1 .

The weakening of texture from the cEBSD analysis from the RS to the AS of the

CuNb FSW suggests that the “flow zone” strain is non-trivial when it comes to the final microstructure of the material. All material that passes through the FSW nugget experiences the varying amounts of shear strain as it enters, is swept around the tool, and exits in the wake of the FSW. Material entering on the AS of the FSW centerline experiences an additional strain/flow path as it is transported around the tool multiple times. This strain is apparently significant enough to weaken the texture observed in cEBSD scans of the AS FSW nugget.

Although the TKD orientation maps in Figures 4.9 – 4.11 showed layers of Cu and Nb, too few grains were indexed for a meaningful statistical orientation analysis.

However within the grains indexed, the patterns did not appear to indicate a preferential orientation. This would agree with the cEBSD scan from the AS in Figures 4.7 and 4.8, which show a textural weakening from the RS to the AS. It is known from TEM analysis presented in chapter II of this study, that if addition strain via FSW is input, the final microstructure contains randomly oriented equiaxed nano grains. Keeping in mind the final microstructure, the evolution from a strong PM texture on the RS to a more random orientation on the AS seems logical. The difference in the layer orientations and grain

size between a 86 nm layer thickness CuNb NLC processed solely by ARB [9] and the

layers of the FSW analyzed by TKD, where many of grains were on the same order of

lengthscale, in interesting. The ARB material, shown in Figure 4.13, has extremely

elongated grains with an obvious texture especially in the Nb were the layers are almost a

solid color indicating crystallographic alignment of the grains. This difference in texture

and grain size is attributed to the difference in strain paths undergone to achieve the reduction in layer thickness.

Figure 4.13 cEBSD results from h=86 nm ARB CuNb sample.

a) Cu layer orientation map. b) Nb layer orientation map results [9].

4.4.2 Factors influencing indexing of TKD

The TKD scans shown in Figures 4.9-4.11 represent the best results obtained to

improve the number of indexed points by varying working distance, accelerating voltage,

and tilt of the specimen. Time constraint on the availability of the equipment prevented a

full investigation into all of these parameters. Literature reports multiple possible factors

that can degrade scan quality of TKD patterns such as thickness variations, sample strain,

and sample preparation. TKD resolutions of <10 nm grains have been reported in Nickel

samples that were electrodeposited. These samples were ideal cases for TKD due to

uniform thickness, low dislocation density, and absence of strain. Monte Carlo

simulations show TKD sample thickness directly affects the physical resolution of the

electron beam [21], with the resolution decreasing as the foil becomes thicker. These

simulations show a change in thickness for a Cu foil from 50 nm to 150 nm results in a change in physical resolution from 25 nm to 75 nm [21]. The samples used in this study vary in thickness from top to bottom, with the top of the sample being the top of Figure

4.5 and the bottom being the bottom of Figure 4.5. The change in thickness is due to the physics of the FIB thinning process. The scans of samples B and C, shown in Figures 4.9 and 4.10, have regions where layers index better than others. Sample thickness variations may contribute to this.

The strained microstructure of the FSW nugget could also contribute to poor indexing, as SAE 8620 strained via hard turning showed grain maps with a large percentage of unindexed areas [25]. Although TKD patterns are reported to be less sensitive than cEBSD to strained material [19], the highly strained grains in the CuNb

FSW nugget may still have a negative effect on scan quality [25].

FIB damage is another possible contributing factor in poor pattern indexing as ion induced FIB damage was reported to reduce indexable patterns in 316 stainless steel to

48% as opposed to >90% for the same material prepared by electropolishing [26]. TKD operational parameters of accelerating voltage, working distance, and sample tilt, have also reported to affect pattern indexability [20] [21]. For this study accelerating voltage was varied from 15 to 30 kV, and working distances of 3 to 15 mm were investigated with an optimal 30 kV and working distance of 3 mm determined by pattern quality.

Limited access time to the SEM for scans resulted in tilt angles of -5, -10, and -15 degrees being investigated with the -15 degree tilt found to be the best. Tilt angles are reported to affect proper indexing of patterns due to the geometry of the EBSD detector position in current SEMs [21]. Low angles of tilt are reported in literature as resulting in

higher resolution by means of a smaller interaction volume [17] [19]. However low back- tilt angles coupled with the physical geometry of a typical EBSD detector position causes a higher intensity of Kikuchi bands at the bottom of the phosphorus detection screen which causes problems with the EBSD software indexing algorithms [20][21]. Although

Monte Carlo simulations predict a back-tilt of 20 degrees for optimal resolution in a 100 nm thick gold sample [21], experiments by Suzuki found 40 degree back-tilt angle to be the optimal balance between resolution and screen brightness with aluminum and tempered martensite samples [20].

The reason the Supra 40 SEM with EDAX EBSD system was only able to index the EBSD patterns at a focal WD of 8mm with the software WD set to 9 mm is still under investigation. The software had recently been calibrated by an EDAX service engineer and the software appeared to work well with EBSD of other materials. Possible future work of using high resolution TEM to measure lattice parameters may be beneficial to see if severe lattice distortion is present, which would require an adjustment of pattern calibration factors.

4.5 Conclusions

EDS was used to confirm that distinct Cu and Nb layers were maintained in the

FSW nugget. The cEBSD scans revealed Cu and Nb on the lower strained RS of the

FSW nugget maintained the texture of the PM. Both cEBSD and TKD results from the higher strained AS of the nugget showed a weakening of texture which could be due to the additional strain input caused by the material making multiple passes around the tool.

The additional strain caused an out-of-plane flow of material in marker studies [23].

cEBSD was able to provide index a higher number of grains adjacent to the FSW centerline on both the AS and RS of the nugget.

To determine orientation data for the layers on the more strained AS of the FSW nugget, TEM foils were analyzed using TKD. Since only a limited number of grains from each sample were able to be indexed, this prevented obtaining statistically viable texture measurements. However, the grains within the layers that were indexed on the

AS, showed a much smaller aspect ratio and random orientation that similar length scale material produced solely by ARB. Possible contributors to the poor indexing by TDK as reported in literature include: non-uniform sample thickness, too low of a tilt angle in

TDK, and possible FIB damage.

4.6 Acknowledgements

This work was supported in part by the Los Alamos National Laboratory Directed

Research and Development (LDRD) project 20130764ECR at the Center for Integrated

Nanotechnologies, an Office of Science User Facility operated for the U.S. Department of Energy (DOE) Office of Science. Los Alamos National Laboratory, an affirmative action equal opportunity employer, is operated by Los Alamos National Security, LLC, for the National Nuclear Security Administration of the U.S. Department of Energy under contract DE-AC52-06NA25396. JBC acknowledges the summer support to be in residence and obtain the specimens used in the FSW studies.

A portion of this research using TDK was conducted at the ORNL SHaRE User

Facility under grant CNMS2015-049. The assistance of Dr. Donovan Leonard on TDK characterization is gratefully acknowledged. ORNL is sponsored by the Division of

Scientific User Facilities, Office of Basic Energy Sciences, U.S. Department of Energy. 76

The equipment used at Mississippi State University includes the FE-SEM obtained under NSF-IMR Grant #DMR-0216703 and 02070615 and the Friction Stir

Welding equipment obtained under a NASA Federal Initiative on “Advancing Disruptive

Manufacturing Research”.

4.7 References

[1] J. S. Carpenter, R. J. McCabe, S. J. Zheng, T. A. Wynn, N. A. Mara, and I. J. Beyerlein, “Processing Parameter Influence on Texture and Microstructural Evolution in Cu-Nb Multilayer Composites Fabricated via Accumulative Roll Bonding,” Metall. Mater. Trans. A, vol. 45, no. 4, pp. 2192–2208, Apr. 2014.

[2] J. S. Carpenter, S. C. Vogel, J. E. LeDonne, D. L. Hammon, I. J. Beyerlein, and N. A. Mara, “Bulk texture evolution of Cu–Nb nanolamellar composites during accumulative roll bonding,” Acta Mater., vol. 60, no. 4, pp. 1576–1586, Feb. 2012.

[3] D. B. Williams and C. B. Carter, Transmission electron microscopy: a textbook for materials science, 2nd ed. New York: Springer, 2008.

[4] S. Zaefferer, “A critical review of orientation microscopy in SEM and TEM,” Cryst. Res. Technol., vol. 46, no. 6, pp. 607–628, Jun. 2011.

[5] J. A. Schneider and E. A. Kenik, “Microstructural Influences on the Development and Growth of Small Fatigue Cracks in the Near Threshold Regime,” J ASTM Int, vol. 2, pp. 1–11, 2005.

[6] A. Deal, X. Tao, and A. Eades, “EBSD geometry in the SEM: simulation and representation,” Surf. Interface Anal., vol. 37, no. 11, pp. 1017–1020, Nov. 2005.

[7] D. R. Steinmetz and S. Zaefferer, “Towards ultrahigh resolution EBSD by low accelerating voltage,” Mater. Sci. Technol., vol. 26, no. 6, pp. 640–645, Jun. 2010.

[8] T. Maitland and S. Sitzman, Electron backscatter diffraction (EBSD) technique and materials characterization examples, vol. 14. Springer Berlin, 2007.

[9] I. J. Beyerlein, N. A. Mara, J. S. Carpenter, T. Nizolek, W. M. Mook, T. A. Wynn, R. J. McCabe, J. R. Mayeur, K. Kang, S. Zheng, J. Wang, and T. M. Pollock, “Interface-driven microstructure development and ultra high strength of bulk nanostructured Cu-Nb multilayers fabricated by severe plastic deformation,” J. Mater. Res., vol. 28, no. 13, pp. 1799–1812, 2013.

[10] B. L. Hansen, J. S. Carpenter, S. D. Sintay, C. A. Bronkhorst, R. J. McCabe, J. R. Mayeur, H. M. Mourad, I. J. Beyerlein, N. A. Mara, S. R. Chen, and G. T. Gray III, “Modeling the texture evolution of Cu/Nb layered composites during rolling,” Int. J. Plast., vol. 49, pp. 71–84, Oct. 2013.

[11] R. D. Doherty, D. A. Hughes, F. J. Humphreys, J. J. Jonas, D. J. Jensen, M. E. Kassner, W. E. King, T. R. McNelley, H. J. McQueen, and A. D. Rollett, “Current issues in recrystallization: a review,” Mater. Sci. Eng. A, vol. 238, no. 2, pp. 219– 274, Nov. 1997. 78

[12] H. Borkar, S. Seifeddine, and A. E. W. Jarfors, “In-situ EBSD study of deformation behavior of Al–Si–Cu alloys during tensile testing,” Mater. Des., vol. 84, pp. 36–47, Nov. 2015.

[13] D. Fullwood, M. Vaudin, C. Daniels, T. Ruggles, and S. I. Wright, “Validation of kinematically simulated pattern HR-EBSD for measuring absolute strains and lattice tetragonality,” Mater. Charact., vol. 107, pp. 270–277, Sep. 2015.

[14] M. D. Vaudin, W. A. Osborn, L. H. Friedman, J. M. Gorham, V. Vartanian, and R. F. Cook, “Designing a standard for strain mapping: HR-EBSD analysis of SiGe thin film structures on Si,” Ultramicroscopy, vol. 148, pp. 94–104, Jan. 2015.

[15] F. J. Humphreys, Y. Huang, I. Brough, and C. Harris, “Electron backscatter diffraction of grain and subgrain structures—resolution considerations,” J. Microsc., vol. 195, no. 3, pp. 212–216, 1999.

[16] B. Hafner, “Scanning electron microscopy primer,” Charact. Facil. Univ. Minn.- Twin Cities, pp. 1–29, 2007.

[17] R. R. Keller and R. H. Geiss, “Transmission EBSD from 10 nm domains in a scanning electron microscope: TRANSMISSION EBSD IN THE SEM,” J. Microsc., vol. 245, no. 3, pp. 245–251, Mar. 2012.

[18] M. Meisnar, A. Vilalta-Clemente, A. Gholinia, M. Moody, A. J. Wilkinson, N. Huin, and S. Lozano-Perez, “Using transmission Kikuchi diffraction to study intergranular stress corrosion cracking in type 316 stainless steels,” Micron, vol. 75, pp. 1–10, Aug. 2015.

[19] P. W. Trimby, Y. Cao, Z. Chen, S. Han, K. J. Hemker, J. Lian, X. Liao, P. Rottmann, S. Samudrala, J. Sun, J. T. Wang, J. Wheeler, and J. M. Cairney, “Characterizing deformed ultrafine-grained and nanocrystalline materials using transmission Kikuchi diffraction in a scanning electron microscope,” Acta Mater., vol. 62, pp. 69–80, Jan. 2014.

[20] S. Suzuki, “Features of Transmission EBSD and its Application,” JOM, vol. 65, no. 9, pp. 1254–1263, Sep. 2013.

[21] R. van Bremen, D. Ribas Gomes, L. T. H. de Jeer, V. Ocelík, and J. T. M. De Hosson, “On the optimum resolution of transmission-electron backscattered diffraction (t-EBSD),” Ultramicroscopy, vol. 160, pp. 256–264, Jan. 2016.

[22] P. W. Trimby, “Orientation mapping of nanostructured materials using transmission Kikuchi diffraction in the scanning electron microscope,” Ultramicroscopy, vol. 120, pp. 16–24, Sep. 2012.

[23] J. A. Schneider and A. C. Nunes, “Quantifying the Material Processing Conditions for an Optimized FSW,” 7th Int’l Conf Trends Weld., pp. 253, 256, 2005.

[24] Y. Morisada, T. Imaizumi, and H. Fujii, “Determination of strain rate in Friction Stir Welding by three-dimensional visualization of material flow using X-ray radiography,” Scr. Mater., vol. 106, pp. 57–60, Sep. 2015.

[25] V. Bedekar, R. Shivpuri, A. Avishai, and R. S. Hyde, “Transmission Kikuchi Diffraction study of texture and orientation development in nanostructured hard turning layers,” CIRP Ann. - Manuf. Technol., vol. 64, no. 1, pp. 73–76, 2015.

[26] W. Zieliński, T. Płociński, and K. J. Kurzydłowski, “Transmission Kikuchi diffraction and transmission electron forescatter imaging of electropolished and FIB manufactured TEM specimens,” Mater. Charact., vol. 104, pp. 42–48, Jun. 2015.

CHAPTER V

CONCLUSIONS

This study investigated the possibility of joining CuNb NLC plates by FSW while maintaining their layered structure, and using the post weld microstructure to provide insights into the flow and strain paths experienced during FSW. Chapter II showed the possibility of maintaining a refined layered structure post FSW, but that the layered structure would refine to the point of fragment into equiaxed nanograins if too much strain was input into the material.

Chapter III demonstrated the ability to join CuNb NLC without volumetric defect

using FSW. Two sets of panels were joined for evaluation of mechanical properties. One

was joined by a single pass of the FSW tool on the crown side, and the second joined by a

pass of the FSW tool on the crown and root side. The single pass FSW maintained a

layered microstructure throughout the FSW nugget. While the layered structure of the

double pass FSW was lost in some regions due to the additional strain input during

second FSW pass. The single pass weld had an UTS of 606 MPa vs. 404 MPa for the

double pass weld. This was attributed to softening of the TMAZ region in the double pass

weld due to its longer time at elevated temperature.

Visual analysis of the post FSW layers revealed an evolution of flow patterns that

match Kelvin-Helmholtz instabilities commonly seen in fluids exposed to a shear flow.

Other NLC with a component of shear deformation similar to FSW were also reported exhibit similar Kelvin-Helmholtz-like layer evolution [1] [2].

Chapter IV presents results using various EBSD techniques to characterize the texture. The PM texture of the ARB was maintained across the RS of a transverse section of the FSW nugget. This texture weakened after crossing the FSW centerline to the AS. The weakening of texture was attributed to the additional out of plane strain experienced by the material on the AS of the FSW nugget. TKD was used to analyze layers on the AS too refined to be resolved by cEBSD. Difficulties indexing TKD patterns prevented obtaining a meaningful statistical orientation analysis.

Chapter IV discusses additional approaches which may result in higher resolution

TKD maps in future work.

5.1 References

[1] E. H. Ekiz, T. G. Lach, R. S. Averback, N. A. Mara, I. J. Beyerlein, M. Pouryazdan, H. Hahn, and P. Bellon, “Microstructural evolution of nanolayered Cu–Nb composites subjected to high-pressure torsion,” Acta Mater., vol. 72, pp. 178–191, Jun. 2014.

[2] Z.-P. Luo, G.-P. Zhang, and R. Schwaiger, “Microstructural vortex formation during cyclic sliding of Cu/Au multilayers,” Scr. Mater., vol. 107, pp. 67–70, Oct. 2015.

CHAPTER VI

FUTURE WORK

To complete the analysis of grain orientation on the AS of the FSW nugget requires further investigation. Chapter IV discussed several things that could be changed to improve the likelihood of indexing the TKD patterns. These methods focus on specimen preparation and imaging conditions. Lower beam energy could be used in the

FIB process to minimize damage and curtaining. A higher angle of back-tilt may also result in increased indexing.

The lattice distortion, due to residual strain in the FSW material could also affect the ability to index patterns using EBSD. High resolution TEM could be used to measure the lattice distortion differences on the AS and RS. It is recommended to prepare TEM foils from the RS of the FSW nugget to obtain TDK data.

With grain orientation data obtained for the AS and RS of the FSW nugget, the data could be used in a plasticity model, such as a visco plastic self-consistent (VPSC) model, to validate the strain type and amount input by the FSW process. This analysis would consider the starting texture of the PM, assumptions regarding the FSW flow path, and matching the resulting texture predicted with the known final texture.

SEM AND EBSD SAMPLE PREPERATION METHODS

Samples were prepared cutting with a liquid cooled diamond wafering blade.

They were then mounted in a low heat epoxy. This was followed by grinding using SiC paper of 220, 400, 600, 800, 1200 grit consecutively, then polished using a 3µm and

0.3µm diamond solution on a buffing pad, placed in a 4 hour vibratory polish of alumina, followed by a 15 minute vibratory polish in colloidal silica.

STRESS VS. STRAIN CURVES FOR P.M., SINGLE PASS FSW, AND DOUBLE

PASS FSW

Figure A.1 Stress vs. strain curve for PM, single pass FSW, and double pass FSW

APPENDIX C