Superplasticity of coarse grained aluminum alloys Chen, Zhenguo

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below. Document Version Publisher's PDF, also known as Version of record

Publication date: 2010 Link to publication in University of Groningen/UMCG research database

Citation for published version (APA): Chen, Z. (2010). Superplasticity of coarse grained aluminum alloys Groningen: s.n.

Copyright Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons). Take-down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.

Download date: 21-01-2017

RIJKSUNIVERSITEIT GRONINGEN

SUPERPLASTICITY OF COARSE GRAINED ALUMINUM ALLOYS

Proefschrift

ter verkrijging van het doctoraat in de Wiskunde en Natuurwetenschappen aan de Rijksuniversiteit Groningen op gezag van de Rector Magnificus, dr. F. Zwarts, in het openbaar te verdedigen op vrijdag 10 september 2010 om 14.45 uur door

Zhenguo Chen geboren op 12 June 1979 te Xintai, China

Promotor:

Prof. dr. J. Th. M. De Hosson

Beoordelingscommissie:

Prof. ir. L. Katgerman Prof. dr. P. Rudolf Prof. dr. H. A. De Raedt

SUPERPLASTICITY OF COARSE GRAINED ALUMINUM ALLOYS Zhenguo Chen

To my family

ISBN 978-90-77172-60-5 Zernike Institue PhD thesis series 2010-12 ISSN 1570-1530 Print: Groningen University Press Cover: Dislocation microstructure in the local necking region, a TEM micrograph of a precipitate pinned high angle grain boundary.

The work described in this thesis was carried out as part of the innovation program of the Materials innovation institute (M2i) (formerly, the Netherlands Institute for Metals Research) on Superplasticity in Superplasticity in fine- and coarse- grained materials under the project number MC4.05219.

4

Contents 1 Introduction ...........................................1 1.1 1.2

Aluminum alloys............................................................................................................. 1 Scope of the thesis........................................................................................................... 4

2 Experimental Procedures.......................9 2.1 2.2 2.3 2.4 2.5 2.6

Material and Production Process..................................................................................... 9 Mechanical Property Test.............................................................................................. 10 Scanning Electron Microscopy ..................................................................................... 11 Electron Backscattered Diffraction ............................................................................... 14 Orientations and Misorientations .................................................................................. 19 Transmission Electron Microscopy............................................................................... 23

3 Mechanical Properties .........................35 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 3.10

Experimental Procedure ................................................................................................ 36 Results........................................................................................................................... 39 The As-Received Microstructure .................................................................................. 39 Mechanical Properties................................................................................................... 40 Anisotropy and Necking Instabilities............................................................................ 47 Deformation Microstructures ........................................................................................ 52 Maximum Elongation, Anisotropy and Failure............................................................. 57 Flow Instabilities and Dislocation Microstructure ........................................................ 59 Mechanical Properties and Deformation Behavior ....................................................... 62 Conclusions................................................................................................................... 66

4 Deformation Processes ........................70 4.1 4.2 4.3 4.4 4.5 4.6

Experimental Procedure ................................................................................................ 71 Results........................................................................................................................... 72 Grain Size and Texture Evolution with Local Strain..................................................... 77 Deformation, recovery, and recrystallization. ............................................................... 80 Discussion ..................................................................................................................... 92 Conclusions................................................................................................................... 99

5 Deformation Mechanisms .................102 5.1 5.2 5.3 5.4 5.5

Strain Rate Change Test Results ................................................................................. 103 Strain Rate Sensitivity, m ............................................................................................ 104 Activation Energy ....................................................................................................... 108 Deformation Mechanism Maps................................................................................... 110 Conclusion .................................................................................................................. 112

6 Summary and Outlook.......................116 Summary.................................................................................................................................... 116 Outlook ...................................................................................................................................... 119

Acknowledgements...............................121

1 Introduction 1.1 Aluminum alloys Automotive industry demands an increasing use of aluminum alloys to reduce weight so as to improve performance and fuel consumption. However, in comparison to steel aluminum alloy sheet materials have lower formability in cold stamping processes. The 5000 series aluminum alloy (AA 5xxxx) offers an alternative approach which can be deformed to quite a high percentage at elevated temperature by the so-called superplastic forming (SPF) process[1, 2]. Superplasticity is the ability of a material to undergo very large uniform neckless tensile deformation normally over 500% elongation prior to failure at a temperature well below its melting point (Tm). Because the deformation mechanisms fall into the grain boundary sliding (GBS) regime[3], fine grain size of 10 µm, high operation temperature of 0.9Tm and slow strain rate are required as for the typical AA5083 material. Aluminum based materials that rely solely on grain boundary sliding for their superplastic properties have relatively high purity; in order to acquire their fine grain size, they require significant thermomechanical processing and thus, are quite costly. Furthermore, the low deformation strain rate, ε& , results in long forming times, and consequently, its application in the automotive industry is fairly limited. By promoting grain boundary sliding, a dramatic decrease of the forming time may be obtained, that is associated with ε& approximately equal to 10-1 s-1 and a maximum elongation to failure above 550% [4, 5]. This requires the reduction of the grain size in the sub-micrometer or nanometer scale, by applying severe plastic deformation (i.e. Equal-Channel Angular

1

Chapter 1 Pressing, or ECAP) [5, 6]. These processes, however, are currently not capable of producing low cost material in the industrial scale. The automotive industry requires alloys with a uniform maximum elongation between 200 and 300%. Since rolled products are preferred for volume component production, “engineering superplasticity” is concerned mostly with materials that do not follow the definition of superplasticity in the strictest sense and for that reason their deformation is often characterized as “enhanced ductility” or “quasi-superplasticity”[7, 8]. Since this thesis is concerned with this type of materials, the term “superplasticity”, or “coarse-grained superplasticity”, even though it is not exactly applicable in all circumstances, will be maintained. For volume component production, the most widely used alloy is the high purity fine grained AA5083 (with grain size equal, or less than 10 µm). The forming time is related to ε& of approximately 10-3 s-1 and the operating T is lower than 500°C. At these conditions, the material deforms not only by grain boundary sliding but also by a second process, which is based on viscous glide or solute drag creep (SDC) of dislocations. This mechanism is grain size insensitive and demonstrates strain rate sensitivity value of m ≈ 0.3. Indeed, a recently developed constitutive model that incorporates both mechanisms seems to predict accurately the results obtained by gas-pressure bulge tests[9] A further decrease in the cost of the primary material was found to be possible by using coarsegrained AA5083, i.e. it does not require extensive thermomechanical treatment for microstructure modification. These alloys with an initial grain size of 70 µm, exhibited a maximum elongation to failure in excess of 300% at 440°C and a strain rate of 10-2 s-1. The operation of solute drag creep alone, in coarse grained AA5083 was found to

2

Introduction promote dynamic recovery leading to a significant grain refinement of the microstructure (i.e. the average grain size decreased to 43 µm), and thus, an enhancement of plasticity[10]. Even though viscous glide and climb of dislocations occurred simultaneously, the former was considered rate controlling. Consequently, coarse-grained materials may well fulfill the industrial requirements and within this scope, the use of the low purity coarse-grained AA5182 would constitute the next step for further cost reduction. The presence of additional solutes, however, may arrest the viscous glide of dislocations rendering their climb (which depends on the thermal vacancy concentration) as the rate controlling mechanism of deformation. The presence of more precipitates at the GB interfaces may hinder the intragranular dislocation motion and result in significant stress concentrations. Even though, the presence of more precipitates at the GB may prevent large grain growth, most of the deformation may be restricted intragranularly. The subdivision of the grains into sub-grains and their continuous refinement may reach the point where discontinuous dynamic recrystallization (DDRX), cannot be prevented. Regions of the microstructure with a large distribution of softer grains (i.e. having orientations where more slip systems can be activated to carry out the deformation) will most likely deform preferentially resulting in non-uniform plasticity. The superplastic deformation of coarse-grained aluminum alloys at high temperatures is a complex phenomenon and therefore, cannot be considered solely on the basis of solute drag creep. Grain size dependent and grain size independent mechanisms take place simultaneously and modify the microstructure to such extent that it exhibits little resemblance with the original [11]. The grain “dynamic reconstruction” which is

3

Chapter 1 demonstrated by extensive refinement is often explained in terms of a process referred to as “continuous recrystallization” or “continuous dynamic recrystallization” (CDRX)[12, 13]. Theoretical modeling, as well as in-situ direct observations have shown that this is most likely a “recovery dominated transformation of the microstructure that occurs homogeneously” [14, 15]. Contrary to the ordinary dynamic or discontinuous dynamic recrystallization, this process does not produce large changes in the texture and is, presumably, a balance between the progressive increase of the sub-grain boundary (SGB) misorientation by dislocation accumulation, which transforms them into low-angle grain boundaries (LAGBs) and then to high-angle grain boundaries (HAGBs), and the migration of the adjoining high-angle grain boundaries which annihilate the highly deformed sub-grain structure. As a result, the microstructure is refined up to the point where both mechanisms (i.e. the progressive increase of the sub-grain misorientation and their absorption by high-angle grain boundary migration) reach steady-state equilibrium. According to this mechanism, recovery of the dislocation substructure can occur either by their condensation into new low-angle grain boundaries or by their absorption by migrating pre-existing high-angle grain boundaries. Even though this mechanism in not well understood, it is of central importance in the superplastically deformed alloys because it does not only soften and restore the ductility of the material, it also produces a completely new grain structure with modified grain size, and shape.

1.2 Scope of the thesis In Chapter 2, the material composition and experimental methods used throughout this thesis are introduced. The main techniques employed are Scanning Electron

4

Introduction Microscopy (SEM) attached with Orientation Imaging Microscopy (OIM also known as EBSD for Electron Backscattered Diffraction) facility and Transmission Electron Microscopy (TEM), and their aspects relevant to the observation and characterization of the microstructures like the grain sizes and grain boundaries are reviewed. Chapter 3 presents the mechanical performance studied by uniaxial tensile tests on dog-bone aluminum sheet specimens. The experiments include systematic measurements of the superplastic behavior under tension over a wide range of deformation temperature and strain rate as well as under different actuation modes of the extension bar, i.e. at constant cross-head speed or true strain rate deformation mode. In conjunction with the mechanical properties, the anisotropy and the dislocation microstructures were investigated. At the secondary necking instabilities, the average dislocation velocity increases, most dislocations break away form their solute atmospheres and thermally activated deformation occurs with high activation energies. Grain boundary sliding is prohibited due to the presence of coherent precipitates that pin effectively the grain boundary motion. In Chapter 4, the results of EBSD measurements are presented and discussed. The experiments were performed on specimens of coarse-grained AA5182 after uniaxial tension along and perpendicular to the rolling direction. The novelty of the approach relies on the fact that the EBSD map was partitioned into deformed, recovered and recrystallized microstructure. The evolution of the texture, grain size and volume fraction, sub-grain boundary, low-angle grain boundary (LAGB) and the high-angle grain boundary (HAGB) density were investigated with increasing local strain. A stable grain size and a balanced Cube texture and Goss texture with similar proportion of each

5

Chapter 1 produced large elongations. Continuous dynamic recrystallization led to homogeneous grain refinement. During deformation a dislocation controlled process produced many sub-grain boundaries (SGBs) and low-angle grain boundaries pinned by precipitates. During recovery the low-angle grain boundaries are converted into high-angle grain boundaries. Recrystallization resulted in a large volume fraction of small grains separated by high-angle grain boundaries. Continuation of the continuous dynamic recrystallization refines further the microstructure until the precipitates can no longer prevent grain boundary long range motion. Then discontinuous dynamic recrystallization leads to necking and failure. Chapter 5 deals with the deformation mechanisms of mechanical behavior of coarse-grained Aluminum alloy. Strain rate change mechanical tests were conducted on four specimens from two AA5182 alloy materials. The specimen with grain size of 21 µm demonstrated stress exponents and activation energies characteristic of grain boundary sliding, solute drag creep and dislocation glide creep (DGC) at low, intermediate and high strain rates, respectively. Solute drag creep and dislocation glide creep are dominant mechanisms governing the deformation and contributing to the high strain rate superplasticity. Based on this specimen, deformation mechanism map is constructed presenting the dominant mechanism over regions with different temperatures and stress level. It is also demonstrated that dislocation glide creep can be the only dominant responsible mechanism of superplasticity for some specimens under certain conditions. At the end of this thesis a summary and outlook is presented in Chapter 6.

6

Introduction References [1] Hojas M, Kuhlein W, Siegert K, Werle T. Superplastic Aluminum Sheet, Metallurgical Requirements, Production and Properties. Metall 1991;45:130. [2] Vetrano JS, Lavender CA, Hamilton CH, Smith MT, Bruemmer SM. Superplastic Behaviour in a Commercial 5083-Aluminum-Alloy. Scripta Metallurgica Et Materialia 1994;30:565. [3] Sherby OD, Wadsworth J. Superplasticity - Recent Advances and Future-Directions. Progress in Materials Science 1989;33:169. [4] Figueiredo RB, Kawasaki M, Xu C, Langdon TG. Achieving superplastic behaviour in fcc and hcp metals processed by equal-channel angular pressing. Materials Science and Engineering a-Structural Materials Properties Microstructure and Processing 2008;493:104. [5] Valiev RZ, Islamgaliev RK, Alexandrov IV. Bulk nanostructured materials from severe plastic deformation. Progress in Materials Science 2000;45:103. [6] Herling DR, Smith MT. Improvements in superplastic performance of commercial AA5083 aluminium processed by equal channel angular extrusion. Superplasticity in Advanced Materials, Icsam2000 2001;357-3:465. [7] Taleff EM, Henshall GA, Nieh TG, Lesuer DR, Wadsworth J. Warm-temperature tensile ductility in Al-Mg alloys. Metallurgical and Materials Transactions a-Physical Metallurgy and Materials Science 1998;29:1081. [8] Woo SS, Kim YR, Shin DH, Kim WJ. Effects of Mg concentration on the quasi-superplasticity of coarse-grained Al-Mg alloys. Scripta Materialia 1997;37:1351. [9] Taleff EM, Hector LG, Bradley JR, Verma R, Krajewski PE. The effect of stress state on hightemperature deformation of fine-grained aluminum-magnesium alloy AA5083 sheet. Acta Materialia 2009;57:2812. [10] Soer WA, Chezan AR, De Hosson JTM. Deformation and reconstruction mechanisms in coarsegrained superplastic Al-Mg alloys. Acta Materialia 2006;54:3827. [11] McQueen HJ, Evangelista E, Kassner ME. The Classification and Determination of Restoration Mechanisms in the Hot-Working of Al-Alloys. Zeitschrift Fur Metallkunde 1991;82:336. [12] Perez-Prado MT, McNelley TR, Swisher DL, Gonzalez-Doncel G, Ruano OA. Texture analysis of the transition from slip to grain boundary sliding in a continuously recrystallized superplastic aluminum alloy. Materials Science and Engineering a-Structural Materials Properties Microstructure and Processing 2003;342:216. [13] Doherty RD, Hughes DA, Humphreys FJ, Jonas JJ, Jensen DJ, Kassner ME, King WE, McNelley TR, McQueen HJ, Rollett AD. Current issues in recrystallization: a review. Materials Science and Engineering a-Structural Materials Properties Microstructure and Processing 1997;238:219. [14] Dougherty LM, Robertson IM, Vetrano JS. Direct observation of the behaviour of grain boundaries during continuous dynamic recrystallization in an Al-4Mg-0.3Sc alloy. Acta Materialia 2003;51:4367. [15] Gourdet S, Montheillet F. A model of continuous dynamic recrystallization. Acta Materialia 2003;51:2685.

7

Chapter 1

8

2 Experimental Procedures 2.1 Material and Production Process The material studied concerns a commercial Aluminum Alloy AA5182 in the form of rolled and annealed sheets with 1.5 mm or 2.0 mm of thickness. It is primarily a solid solution of Mg in Al, with alloying elements being Al -5.0% Mg -0.3% Mn -0.1% Cu – 0.2% Si -0.2% Fe (w.t.%). Mg is added to increase the strength through solid solution strengthening and improve their strain hardening ability. Mg forms a coherent β’ phase with Al, which is present as a fine dispersion at temperatures below 200°C. This phase is relatively stable and has a slow unbinding kinetics. However, the thermodynamically stable phase is Al3Mg2, which precipitates only upon significant annealing at temperatures immediately below 200°C. The addition of Mn is to moderately increase the strength somewhat through solution strengthening as well and to improve strain hardening while not appreciably reducing ductility or corrosion resistance. The alloying element of Cu provides substantial increase in strength and facilitates precipitation hardening, however, it can also reduce ductility and corrosion resistance. The addition of Si to aluminum reduces melting temperature and improves fluidity. Fe is the most common impurity found in aluminum but sometimes it is intentionally added to provide a slight increase in strength. The added alloy elements can form intermetallic second phases for example the Mg2Si silicides and (Mn,Fe)Al6, which is believed that some of these are located along the grain boundaries to serve as grain growth stabilizers.

9

Chapter 2

2.2 Mechanical Property Test Two commercial Aluminum alloy AA5182 materials were used with composition of Al-5.0% Mg-0.3% Mn-0.1% Cu (wt.%) and minor impurities of Si, Fe, and Cr. The Material B indicates the alloy sheet with thickness of 2 mm and average grain size of 21µm, and the Materials C has thickness of 1.5 mm and average grain size of 37µm. Dog-bone specimens were cut by Electrical Discharge Machining (EDM). The gauge directions have different orientations of Along-, Diagonal- and Perpendicular-to-the rolling direction (RD) which will be noted for Materials B and C as Ba, Bd, Bp and Ca, Cd, Cp. The gauge width is 4 mm and the length is 16 mm with a curvature section at each end connected to the grip having a radius of 2 mm. In addition, the specimens were tested using shoulder-loading rigid grips, whereby mass flow from the specimen grip to the gauge can be neglected. Moreover, the rigid grip design makes it possible to calculate true strain of the gauge from the grip displacement, the accuracy of which increases with the specimen gauge length. Strain Rate Change (SRC) tests were conducted at elevated temperatures for each material at temperature 400°C to 500°C and strain rate from 5x10-4s-1 to 2x10-1s-1. The temperatures were controlled within 3°C with thermal couples attached on the specimen and the extension grips in a three-zone furnace. Specimens were tested at true strain rate in a hydraulic-driven, computer-controlled, electromechanical testing machine to impose prescribed strain rates. Upon a single specimen the SRC tests use a series of strain rates imposed for increasing strain rate changes (5x10-4s-1 to 2x10-1s-1) or decreasing strain rate changes (1x10-1s-1 to 5x10-4s-1), with each rate held at a minimum of 2 pct engineering strain, to produce data for flow stress as a function of temperature, strain, and strain rate.

10

Experimental Procedures “Steady-state” flow stress measurements were made after each rate change by evaluating the stress transient following the rate change and measuring the stress at which the transient had fully decayed. Thereafter, the specimen was tested using increasing or decreasing strain rate change procedure. The mechanical test details will also be presented in chapter 3, 4, and 5.

2.3 Scanning Electron Microscopy The Scanning Electron Microscopy (SEM) is a type of electron microscope that study microstructural and morphological features of the sample surface by scanning it with a high-energy electron beam in a raster scanning pattern. The electrons interact with the sample matter producing a variety of signals that contain information about the sample surface topography, composition and other properties. The first SEM image was obtained by Max Knoll, who in 1935 obtained an image of silicon steel showing electron channeling contrast [1]. The SEM was further developed by Professor Sir Charles Oatley and his postgraduate student Gary Stewart and was first marketed in 1965 by the Cambridge Instrument Company as the "Stereoscan". The signals produced by an SEM include secondary and back-scattered electrons (SE & BSE), characteristic X-rays, cathodoluminescence (light), specimen current and transmitted electrons. SE detector is the most common one in SEM. As shown by the schematic Figure 2.1, in an FEI /Philips XL-30 FEG-SEM (Field Emission Gun), electrons are generated by the field emission gun using a high electrostatic field. They are accelerated with energies between 1 keV and 30 keV down through the column towards the specimen. While the magnetic lenses (condenser and objective lenses) focus the

11

Chapter 2 electron beam to a spot with a diameter of approximately 10 nm, the scanning coils sweep the focused electron beam over the specimen surface. If the microscope is operated in the backscattered mode, the result is a lateral resolution on the order of micrometers. The number of back-scattered electrons produced is proportional to the atomic number of the element bombarded. The result is that material with a high(er) atomic number produces a brighter image. To capture this information a detector is required which can either be metal, which is the least effective, but is versatile and used in environmental scanning electron microscopy (ESEM); semi-conductor, which is most common or a scintilator/light pipe/photomultiplier, which are the most efficient.

Figure 2.1: Schematic figure of the Scanning Electron Microscopy The primary electrons current is approximately 10-8 to 10-7A. The large penetration depth of the high energy electrons will cause the electrons to be trapped in the material. When studying conducting materials, the electrons will be transported away from the point of incidence. If the specimen is a non-conducting material, the excess 12

Experimental Procedures electrons will cause charging of the surface. The electrostatic charge on the surface deflects the incoming electrons, giving rise to distortion of the image. To reduce surface charging effects, a conducting layer of metal, with typical thicknesses 5-10nm, can be sputtered onto the surface. This layer will transport the excess electrons, reducing the negative charging effects. An adverse effect of the sputtered layer is that it may diminish the resolving power of the microscope, since topographical information is no longer gained from the surface of the material, but from the sputtered layer. Charging of the surface is not the only factor determining the resolution of a scanning electron microscope. The width of the electron beam is also an important factor for the lateral resolution. A narrow electron beam results in a high resolution. The spot size however, is a function of the accelerating voltage 2

2

1 ⎡i ⎤ ⎡ ∆E ⎤ 2 ⎡ 1 ⎤ 6 d = 2 ⎢ + λ2 ⎥ + ⎢ Cc ⎥ α + ⎢ Cs ⎥ α α ⎣B ⎦ ⎣ E0 ⎦ ⎣2 ⎦ 2 p

The broadening of the spot size is the sum of broadening effects due to several processes. The first contributor is the beam itself, B is the brightness of the source, i is the beam current and α its divergence angle. The second part is the contribution due to diffraction of the electrons of wavelength λ by the size of the final aperture. The last two parts are the broadening caused by chromatic and spherical aberrations. Where E0 is the electron energy and ∆E is the energy spread, Cs represent the spherical aberration and Cc is the chromatic aberration coefficient. To achieve the smallest spot size, all contributions should be as small as possible. Decreasing the accelerating voltage will not only cause the wavelength of the electrons to increase, but also the chromatic aberration increases as

13

Chapter 2

well, resulting in increasing of the spot size and, as a consequence, a decrease in resolving power of the microscope. A field emission gun has a very high brightness B, reducing the contribution in broadening due to the beam itself. The energy spread ∆E in the electron energies is also small. Together with the fact that the coefficients Cs and Cc can be reduced by optimizing the lenses for low-energy electrons, provides the FEG low voltage scanning electron microscope with very high resolving power. In this thesis scanning electron microscopy is carried out with a FEI Philips XL30-FEG or XL30s-FEG SEM equipped with a field emission gun (FEG).

2.4 Electron Backscattered Diffraction Orientation Imaging Microscopy (OIMTM) is based on automatic indexing of electron backscatter diffraction patterns (EBSD patterns or EBSPs) which can be produced in a properly equipped SEM. OIM provides a complete description of the crystallographic orientations in polycrystalline materials.

Figure 2.2: OIMTM Hardware Configuration

14

Experimental Procedures

OIM hardware A camera is mounted on the SEM and images a phosphor screen inside the specimen chamber (Figure 2.2). The electron beam is focused on a particular point of interest in the sample. The interaction of the beam and the microstructure results in an EBSD image forming on the phosphor screen, which is captured by the camera and then further processed and digitalized in a computer. The image is automatically indexed and the following data are calculated and recorded: the orientation of the crystal, a quality factor defining the sharpness of the diffraction pattern (IQ), a TSL-patented “confidence index” (CI) indicating the degree of confidence that the orientation calculation is correct, the phase of the material, and the location (in x,y coordinates) where the data was obtained on the specimen.

Figure 2.3: Formation of Kikuchi lines for EBSD Pattern Formation of EBSP The EBSD technique relies on positioning the specimen within the SEM chamber typically the surface is tilted 70° respect to the horizontal to get a good compromise

15

Chapter 2

between a high electron scattering yield and a safe configuration in the chamber. This enhances the fraction of backscattered electrons able to experience diffraction by lattice planes in the sampled volume and to escape from the specimen surface. This can be understood considering the following: when the primary electrons enter a crystalline solid, the electrons disperse beneath the surface and subsequently they are diffusely and inelastically scattered in all directions.

Figure 2.4: Automated EBSD, a) Kikuchi patterns are transformed to b) Hough space where individual high intensity peaks are detected thereafter c) an orientation is indexed.

These diffracted electrons through the Bragg angle are occurring in all directions, from each family of planes the result are two cones with one from either side of the

16

Experimental Procedures

imaginary source (Figure 2.3). The Bragg angle, for typical values of the electron wavelength and lattice interplanar spacing, is found to be about 0.5°. Consequently, the apex angle of a diffraction cone is close to 180°, i.e. the cones are almost flat. When the phosphor screen intercepts the diffractions cones, a pair of parallel conic sections results, which appears as parallel lines, i.e., the so-called Kikuchi lines. Analysis of EBSPs In commercial automated analysis of digital images, due to the imperfections in either the image data or the edge detector which is generally used for sharp brightness detection, a problem often arises of detecting the simple shapes, such as straight lines, circles or ellipses. There may be missing points or pixels on the desired shapes as well as spatial deviations between the ideal line/circle/ellipse and the noisy edge points obtained by the edge detector. To circumvent this problem one treats the data by applying the Hough transform which was developed to extract straight lines from digital images. In its current form the Hough transform analyzes the image grouping edge points into object candidates by performing an explicit voting procedure over a set of parameterized image objects[2]. The classical Hough transform was concerned with the extraction of straight lines from digital images, but later on it has been extended to identification of arbitrary shapes, for example circles and ellipses. The equation governing the Hough transform is

ρ = x cos θ + y sin θ where (x , y) describe a set of pixel coordinate forming a line in the digital image and the Hough parameter (ρ , θ ) provides a wave like function in Hough space. As the intensity of each (x , y) pixel is added, the problem of finding a Kikuchi band is now 17

Chapter 2

reduced in finding a peak of relatively high intensity in the Hough space (Fig. 2.4b). Once the bands have been detected, the reflecting planes associated with the detected bands must be identified. Two band characteristics can be used for indexing; (1) the width of a band, which is a direct function of the d-spacing through Bragg’s law, this option is a powerful tool for improved accuracy dealing with structures of low symmetry and for phase identification. (2) The angles between the (located) bands which are known and compared to a theoretical directory of interplanar angles; this is the standard method used in the OIM analysis software. The indexing routine derives an orientation solution from just three Kikuchi bands, successively analyzes all possible combinations of band triplets. The selection of the most likely indexing solution is to use the voting scheme. That is, each time the angles in a triplet of bands are compared to the look-up table of angles that are constructed from the crystallography of the sample allowing the Miler indices (h k l) associated with the bands to be identified, the solution receives a vote. The most probable solution is the one that receives the majority of votes. To assess the reliability of the indexing, several parameters such as the image quality (IQ), the confidence index (CI) and the fit between the recalculated and the detected bands may be discerned. The IQ reviews the relative quality of the pattern using the intensities of the found Hough peaks. The CI is given by CI =

(V1 − V2 ) Vtotal

where V1 and V2 are the number of votes for the first and second solutions and Vtotal represents the total possible number of votes from the detected bands. The CI will yield a value between 0 and 1. The comparison between the two highest numbers of votes

18

Experimental Procedures

gives this quantity a doubtful character and may be misleading. Especially in the case where V1 equals V2, this results a CI of 0. The pattern however may be properly indexed. In general, CI values higher then 0.1 will represent a proper indexed pattern. The fit parameter defines the average angular deviation between the recalculated and the detected bands. It is often simply a measure of how well the system is calibrated and the parameters defining the crystal structure are defined.

2.5 Orientations and Misorientations The Figure 2.5 shows a schematic of a crystal with cubic symmetry oriented in a plate of material. The term orientation describes the orientation of the principal axes of this crystal (eiC) relative to the principal axes of the sample (eiS). The Euler angles are the three rotations required to bring the principal axes of the crystal into coincidence with the principal axes of the sample.

Figure 2.5: Coordinate systems of specimen and cubic symmetric crystal In OIM the Bunge's description is used for the Euler angles. This is a so-called passive description - the rotations needed to bring the sample coordinate frame into coincidence with the crystal coordinate frame. (The converse would be the active description which would describe the rotation necessary to bring the crystal coordinate frame into coincidence with the sample frame.) In the case of Bunge’s form of the Euler

19

Chapter 2

angles (ϕ1, Φ, ϕ2) this is a rotation (ϕ1) about the e3S axis (blue-coloured minuscule z) followed by a rotation (Φ) about the e1s axis (blue-coloured minuscule x) followed by a third rotation (ϕ2) about the e3s axis again. The angles ϕ1 and ϕ2 range from 0 to 2π and Φ ranges from 0 to π. These limits form a bounded space referred to as Euler space. Other descriptions are available for representing orientations. These have their own sets of advantages and disadvantages. The Euler angle approach is used primarily due to convention. Much of the mathematics developed for the analysis of orientation and the distribution of orientations in polycrystals (texture) has been developed for the Euler angle representation. Another common description is the cknog representation, where {hkl} is the crystal plane perpendicular to the sample normal and is the crystal direction aligned with the “1” axis of the sample. This has an advantage over the Euler angle description in that it relates directly back to planes and directions in the crystal. Similar to the {hkl} representation, this description is easily related back to crystallographic directions in the material.

Figure 2.6: Definition of Euler Angles (ϕ1, Φ, ϕ2) 20

Experimental Procedures

Figure 2.7: Binned Orientation Distribution Function in Euler space

Figure 2.8: Schematic ODF shows the typical texture components in FCC Al. Table 2.1: Typical texture components in Recrystallized (RX) and Rolled (RL) Al Euler angles RX Texture

Orientation

Euler angles

Orientation

(φ1, Φ, φ2)

{h k l}

RL Texture

Cube

(0, 0, 0)

(0 0 1) [1 0 0]

Goss

Goss

(0, 45, 0)

(0 1 1) [1 0 0]

Copper

P

(54.7, 45, 0)

(0 1 1) [1 -1 1]

Brass

Q

(46.5, 18.4, 0)

(0 1 3) [3 -3 1]

S1

(56.8, 77.4, 26.6)

(2 4 1) [1 -1 2]

R

(15, 57.7, 18.4)

(1 3 2) [4 -2 1]

S2

(64.9, 74.5, 33.7)

(2 3 1) [1 -2 4]

S3

(52.9, 74.5, 33.7)

(2 3 1) [3 -4 6]

D

(270, 27.2, 45)

(4 4 11) [11 11 8]

(φ1, Φ, φ2)

{h k l}

(0, 45, 90)

(1 0 1) [0 -1 0]

(39.2, 65.9, 26.6)

(1 2 1) [1 -1 1]

(54.7, 90, 45)

(1 1 0) [1 -1 2]

21

Chapter 2

Orientations are inherently three dimensional, as shown by the Orientation Distribution Function (ODF) in Euler space (Figure 2.7). Thus to plot a set of orientations requires a method of representing three dimensions in two. This is done using two dimensional sections through the three dimensional space. To facilitate the description of how this is done we will focus on representing crystallographic orientation by Euler angles after the formalism of Bunge. Euler angles describe the rotation require to bring the sample coordinate system into coincidence with the crystal coordinate frame. The space containing all possible orientations defined as Euler angles is Euler space. The space can be thought of as a bounded rectangular volume. The three axes of the rectangular volume are the three Euler angles. A schematic is shown below along with sections through the space. In this example the constant angle sections are those along ϕ1. If 36 sections were used, then the first section would be located at ϕ1 = 5 degrees and would contain all orientations within 0 0 for the other reflections). As the transmitted wave with amplitude φ0(z) propagates through the crystal, its amplitude is depleted by diffraction and the amplitude φg(z) of the diffracted beam increases, i.e. a dynamical interaction between φ0(z) and φg(z) exists. If we assume that sg is parallel to the electron beam, the interaction can be described by the following pair of coupled differential equations, known as the Howie-Whelan equations:

dφg dz

=

πi π i −2π is z φg + φ0 e ξ0 ξg g

and

d φ0 π i π i 2π is z = φ0 + φg e g dz ξ 0 ξg where sg = |sg| and ξ0 and ξg are the extinction distances for the direct and diffracted waves, respectively. The extinction distance is a characteristic length scale determined by the atomic number of the material, the lattice parameters and the wavelength of the electrons, it typically lies between 10 and 100 nm. The above equation shows that the change in φ0(z) as a function of depth z is the sum of forward scattering and scattering from the diffracted beam, taking into account a phase change of π/2 caused by the scattering. Solving the previous 2 equations for φg gives

25

Chapter 2

φg =

⎛ πt ⎞ sin ⎜ 1 + sg2ξ g2 ⎟ ⎜ξ ⎟ 1 + sg2ξ g2 ⎝ g ⎠ i

where t is the thickness of the crystal. Accordingly, the intensity of the diffracted beam becomes

I g = φg

2

sin 2 π tseff = ξ g 2 seff 2

where seff is an effective value of sg defined by seff = (sg2 + ξg-2)1/2. In fact, by replacing seff by sg , the intensity according to the kinematical approximation is found. Absorption can be included in the dynamical theory by adding appropriate terms to both ξ0 and ξg. Mathematically speaking, absorption is included simply by allowing the arguments of the sines and cosines to become complex. Dislocations give rise to contrast because they locally distort the lattice and thereby change the diffraction conditions. If the distortion is given by a displacement field R,i.e.

dφg dz

=

πi πi φg + φ0 exp ⎡⎣ −2π i ( sg z + g ⋅ R ) ⎤⎦ ξ0 ξg

In order for a dislocation to contribute to contrast formation, the dot product g·R must be nonzero. A screw dislocation in an isotropic elastic medium has a displacement field parallel to its Burgers vector, and therefore produces no contrast when g·b = 0. General dislocations have a displacement field with more components; their image contrast also depends on g·be and g·(b×u), where be is the edge component of the Burgers vector and u is a unit vector along the dislocation line. In practice however, only very faint contrast occurs when g·b = 0 but g·be ≠ 0 and g·(b×u) ≠ 0. Therefore, the “invisibility criterion”

26

Experimental Procedures g·b = 0 is used commonly to determine Burgers vectors of dislocations in elastically

isotropic solids. The determination of a Burgers vector involves finding two reflections g1 and g2 for which the dislocation is invisible, so that b is parallel to g1×g2. The situation where only one beam φg is strongly diffracted is referred to as a twobeam condition. This type of diffraction is widely used in conventional TEM of crystalline materials because the contrast is well defined and the Burgers vectors of the dislocations can be determined as described above. By using the objective aperture in the microscope, either of the two beams can be selected to form an image and accordingly, two imaging modes may be distinguished: bright field (BF) when the direct beam φ0 is used, and dark field (DF) when the diffracted beam φg is used. Generally, the diffracted beams do not coincide with the optical axis of the microscope and consequently the DF image will not be of maximum quality due to spherical aberration. To overcome this problem, the incident beam is normally tilted in such a way that the desired diffracted beam passes along the optical axis. While two-beam conditions produce high contrast of dislocations, the resolution at which these defects are resolved is not optimal since the lattice planes around the dislocations are distorted over a relatively large area. To obtain the maximum resolution, the crystal should be tilted slightly further by such an amount that the exact Bragg condition is only fulfilled within a small region near the dislocation. In this way, a highresolution image is obtained in which the dislocation shows up as a bright line. This technique is referred to as weak-beam imaging. The deviation from the exact Bragg condition for perfect crystal is given by sg as mentioned above and can be determined accurately by the relative position of the so-called Kikuchi pattern with respect to the

27

Chapter 2

diffraction pattern. The origin of the Kikuchi pattern lies in the elastic re-scattering of inelastic scattered electrons as further explained in the following section. The width of a dislocation image is approximately 0.3ξg, i.e. several tens of nanometers in conventional bright- and dark-field imaging. This width can be detrimental to the observations of dislocations that are very closely spaced. Since the effective extinction distance decreases for increasing deviation away from the Bragg condition, the width of a dislocation image can be reduced to values in the order of 1 to 5 nm in weak-beam imaging. At lower magnifications, the average crystal orientation may vary considerably within the observed area, so that only a small area of the specimen can be set up in twobeam condition. This is especially relevant in specimens that have been deformed prior to preparation. In these cases, it is often convenient to orient the specimen close to a zone (c) axis, so that many reflections are weakly excited, independently of small changes in orientation. By using the direct beam for imaging, the dislocation structure can be imaged with good contrast over a relatively large area. However, the contrast from individual dislocations is generally smaller than in two-beam condition and not well defined since it results from many different reflections. Another TEM technique refers to high-resolution and is called HRTEM. The essence of HRTEM imaging lies in the transformation of the phase differences in the exit wave into amplitude modulation, and therefore contrast. In the Fraunhofer approximation to image formation, the intensity in the back focal plane of the objective lens is simply the Fourier transform of the wave function exiting the specimen. Inverse transformation in the back focal plane leads to the image in the image plane. However, if the phase-object approximation

28

Experimental Procedures

holds, the image of the specimen by a perfect lens shows no amplitude modulation. In practice, however, a combination with the extra phase shifts induced by defocus and lens aberrations generates suitable contrast. The influence of the extra phase shifts can be taken into account by multiplying the wave function at the back focal plane with functions describing each specific effect. As such effects can be conceived cylindrical symmetric, they can be represented as a function of the distance of the reciprocal lattice point from the optic axis, U = (u 2 + v 2 )1 2 , where u and v are the angular variables used for the distribution of amplitude or intensity in reciprocal space. The phase factor used to describe the shifts introduced by defocus and spherical aberration is[4-8]:

χ (U ) = πλ∆fU 2 + 12 π Cs λ 3U 4 with ∆f the defocus value of the objective lens and Cs the spherical aberration. The function that multiplies the exit wave is then:

B(U ) = exp(i χ (U )) If the specimen behaves as a weak-phase object, only the imaginary part of B(U) contributes to the contrast in the image, T(U):

T(U ) = sin( χ (U )) = sin(πλ∆fU 2 + 1 π Cs λ 3U 4 ) 2

which is known as the contrast transfer function (CTF) of the objective lens, so called because it determines the weight of each scattered beam transferred to the image intensity spectrum. In this way, the phase information from the specimen is converted into intensity information by the phase shift of the objective lens. The aperture inserted at the back focal 29

Chapter 2

plane of the objective lens only allows transmitted beams up to a certain angle. Its effect can then be represented by multiplication with an aperture function A(U) which is unity for U < U0 and zero outside this radius. In practice, combinations of fluctuations in high voltage and lens current, objective lens chromatic aberration, mechanical vibrations along z and usage of a partly converged beam cause loss of spatial and temporal beam coherency, which have been assumed so far. Such effects result in a smearing of the image, and compromise the ultimate resolution, since for high frequencies, the transfer function is now damped and approaching zero. Partial temporal coherence can be represented as a spread of focus, E∆, and partial spatial coherence as incident beam convergence, Eα. These effects are taken into account by multiplying sin(χ) with specific damping envelopes. A definition of resolution is not easily given for an electron microscope. In the case of completely incoherent imaging, as in optical microscopy, it is possible to define point resolution as the ability to resolve individual point objects. This resolution can be expressed (using the criterion of Rayleigh) as a quantity independent of the nature of the object. In electron microscopy, however, the illumination is more coherent and the resolving ability depends on the diffracting behaviour of the object. For example, for thick specimens, there is not necessarily a one-to-one correspondence between the projected structure of the object and its image. Therefore, the concept of microscope resolution is only meaningful for thin objects, and furthermore, one has to distinguish between: •

Point resolution, which is related to the finest detail that can be directly interpreted in terms of the specimen structure. Since the CTF depends very sensitively on defocus, and in general shows an oscillatory behaviour as a function of U, the contribution of the different scattered beams to the amplitude modulation varies. However, for particular

30

Experimental Procedures

underfocus settings the microscope approaches a perfect phase contrast microscope for a range of U before the first crossover, where the CTF remains at values close to –1. It can then be considered that, to a first approximation, all the beams before the first crossover contribute to the contrast with the same weight, and cause image details that are directly interpretable in terms of the projected potential. Optimisation of this behaviour through the balance of the effects of spherical aberration vs defocus leads to the generally accepted optimum defocus −1.2(Cs λ )1/ 2 . Designating an optimum resolution involves a certain degree of arbitrariness. However, the point where the CTF at optimum defocus reaches the value –0.7 for U = 1.49Cs−1 4 λ − 3 4 is usually taken to

give the optimum (point) resolution ( 0.66Cs−1 4λ −3 4 ). This means that the considered passband extends over the spatial frequency region within which transfer is greater than 70%. Beams with U larger than the first crossover are still linearly imaged, but with reverse contrast. Images formed by beams transferred with opposite phases cannot be intuitively interpreted. •

Information limit resolution, corresponding to the highest spatial frequency still appreciably transmitted to the intensity spectrum. This resolution, which is higher than the point resolution, is related to the finest detail that can actually be seen in the image, which however is only interpretable using image simulation. For a thin specimen, such limit is determined by the cut-off of the transfer function due to spread of focus and beam convergence (usually at 1/e2). The use of FEG sources minimizes the loss of spatial coherence. This helps to increase the information limit resolution in the case of lower voltage ( ≤ 200 kV) instruments, because in these cases the temporal coherency does not usually play a critical role. On the other hand, with higher voltage instruments,

31

Chapter 2

due to the increased brightness of the source, the damping effects are always dominated by the spread of focus and FEG sources do not contribute to an increased information limit resolution. The resolution performance can generally be improved by increasing the accelerating voltage of the microscope to reduce λ and/or decreasing the coefficient of spherical aberration of the objective lens. Unfortunately, the use of very high accelerating voltages is ultimately limited by the damage caused to the specimen and the reduction of Cs usually involves specific pole piece designs that pose restrictions to specimen tilt capabilities. Due to the finite resolution, a correct interpretation of the structure requires a careful alignment of the specimen along a low-index zone axis. In practice however, a residual crystal tilt, of a fraction of 1 mrad, usually remains. Beam tilt has a more detrimental influence on the image because the beam enters the objective lens at an angle, causing phase changes in the electron wavefront. This effect can be corrected by the voltage-centre alignment, whereby the acceleration voltage is varied to find the centre of magnification that is then placed on-axis. However, this procedure does not account for misalignments in the imaging system after the objective lens. For an accurate correction, the coma-free alignment is necessary. The process involves alternatively applying equal and opposite beam tilts to the incident beam, and subsequent comparison of the two resulting images (for two orthogonal directions). If there is residual beam tilt, one image is more distorted than the other and the beam-tilt controls should be adjusted until both tilted images display similar distortions. Nevertheless, in practice such procedure is only feasible with a computer-controlled microscope, equipped with a CCD camera, and has not yet

32

Experimental Procedures

been used in the current work; the voltage-centre alignment was used instead, inevitably resulting in some residual beam tilt. On the other hand, two-fold astigmatism can normally be easily removed by the operator, because of its clear effect on the image of amorphous edges. The microscope parameters can be kept constant from experiment to experiment by a careful mode of operation.

33

Chapter 2 References 1.

Knoll, M., Aufladepotentiel und Sekundäremission elektronenbestrahlter Körper. Zeitschrift für technische Physik 1935, 16, 467-475.

2.

Duda, R. O.; Hart, P. E., Use of Hough Transformation to Detect Lines and Curves in Pictures. Communications of the Acm 1972, 15, (1), 11-15.

3.

Kocks, U. F.; Tome, C. N.; Wenk, H. R., Texture and Anisotropy: Preferred orientations in polycrystals and their effect on materials properties. Cambridge University Press 1998.

4.

Williams, D. B.; Carter, C. B., Transmission Electron Microscopy: A Textbook for Materials Science. Springer 2009.

5.

Wouter A. Soer and Jeff Th. M. De Hosson, In-situ transmission electron microscopy, ed. Florian Banhart, World Scientific, 2008, pp.115-160. and W.A. Soer, PhD thesis, Un. of Groningen, pp.:1132, 6-1-2006; D.T.A. Matthews, PhD thesis, Un. of Groningen 13-6-2008, P. A. Carvalho, PhD thesis, Un.of Groningen. pp:1-162, 26-1-2001

6.

J.Th.M. De Hosson, A. van Veen, "Nanoprecipitates and nanocavities in functional materials", Encycl. of Nanoscience and nanotechnology, ASP, USA, Vol. 7, 297-349 (2004)

7.

J.Th.M. De Hosson, "Transmission Electron Microscopy of metals and alloys", in Handbook of Microscopy, eds. S. Amelinckx, D. van Dyck, J. van Landuyt, G. van Tendeloo, Volume 3, p. 5 111, VCH, N.Y. (1997)

8.

J. M. Cowley, in High-Resolution Transmission Electron Microscopy (eds. P. Buseck, J. Cowley, L. Eyring), Oxford University Press, New York (1988) p. 3.

34

3 Mechanical Properties The term superplasticity, in the strictest sense, is defined as the elongation in excess of 500% without necking and failure due to cavitation coalescence. The dominate deformation mechanism is based on diffusion controlled grain boundary sliding (GBS) [1], i.e. the sliding of adjacent grains with respect to each other. The materials normally have grain size smaller than 10 µm and are customarily deformed at temperatures, T, close to 0.9 Tm and at strain rates, ε& , typically around 10-4 s-1. Aluminum based materials that rely solely on grain boundary sliding may achieve a maximum elongation to failure above 550% [2] associated with ε& approximately equal to 10-1 s-1, which requires the reduction of the grain size in the sub-micrometer or nanometer scale, by applying severe plastic deformation (i.e. Equal-Channel Angular Pressing, or ECAP) [3]. These processes, however, are currently not capable of producing low cost material on the industrial scale. For volume component production, the most widely used aluminum alloy is the high purity fine grained AA5083 (with grain size equal, or less than 10 µm), which is deformed not only by grain boundary sliding but also by a second process of viscous glide or solute drag creep (SDC) of dislocations. The automotive industry requires alloys with a uniform elongation between 200 and 300%. Since rolled products are preferred for volume component production, “engineering superplasticity” is concerned mostly with materials that do not follow the definition of superplasticity in the strictest sense and for that reason their deformation is often characterized as “enhanced ductility” or “quasisuperplasticity” [4,5]. Since the present study is concerned with this type of materials, the

35

Chapter 3

term “superplasticity” or “coarse-grained superplasticity”, even though it is not exactly applicable, will still be maintained. A further decrease in the cost of the primary material was found to be possible with the use of coarse-grained AA5083 with an initial grain size of 70 µm, exhibited a maximum elongation to failure in excess of 300% at 440°C and at 10-2 s-1. The operation of solute drag creep alone, in coarse grained AA5083 was found to promote dynamic recovery leading to a significant grain refinement of the microstructure (i.e. the average grain size decreased to 43 µm), and thus, an enhancement of plasticity [6]. Consequently, coarse-grained materials may well fulfill the industrial requirements and within this scope, the use of the low purity coarse-grained AA5182 would constitute the next step for further cost reduction. In this chapter we present a full account of the mechanical properties of the coarse-grained AA5182 aluminum alloy deformed under uniaxial extension. The anisotropy of the materials was determined by extending at strains, ε, of up to 250% and measuring regions of the gauge which exhibited the most uniform area reduction. By employing transmission electron microscopy (TEM), an investigation in the dislocation microstructures revealed the deformation behaviour of this material and finally, we discuss our results outlining the potential of this alloy for future automotive applications.

3.1 Experimental Procedure The materials used in the present study were commercial aluminum AA5182 alloys with composition Al – 5.0 wt.% Mg – 0.3 wt.% Mn – 0.1 wt.% Cu, minor impurities of Si, Fe, and Cr and an as-received grain size of approximately 21 µm (20.7

36

Mechanical Properties

µm) and 37 µm (37.1 µm). In at.%, their exact composition was Al – 5.80 Mg – 0.12 Mn – 0.05 Cu – 0.08 Fe – 0.77 Si – 0.01 Cr. Based on their as-received grain size they will be denoted, henceforth, as 21G and 37G. Specimens for tensile tests were produced by Electrical Discharge Machining (EDM) from the 2 and 1.5 mm thick rolled and shortly annealed metallic sheets, for the 21G and the 37G material, respectively. Their gauge direction was parallel, at 45° and at 90° to the rolling direction (RD). To determine the conditions for maximum elongation prior to failure, tensile tests were performed, first, at constant cross-head speed (CCS) at temperatures between 350 and 500°C and at initial ε& between 10-1 and 10-3 s-1. Subsequently, tests measuring the maximum elongation were carried out at constant true ε& (TSR) of 10-1 and 10-2 s-1 and at T, between 400 and 450°C for the material 21G and between 425 and 500°C for the material 37G. Test up to 50, 100, 150, 200, 250 and wherever possible up to 300% ε, were performed at 425°C and at true strain rate of 10-2 s-1 for both materials, since at these conditions the average value of the maximum elongation to failure achieved for the material 21G was the highest. All tensile tests were carried out in air using a computer controlled 810 Material Testing System (MTS) equipped with a three-zone split furnace. Each specimen was placed between two stainless steel grips and fastened with two plates, the furnace was closed and the target temperature was set in the computer control of the facility, so as to achieve the fastest heating rate possible. The target temperature was reached in approximately 15 minutes and the test commenced after a very short stabilization period of about 90 seconds. The short heating cycle was chosen so as to approximate as much as possible, the conditions of blow forming in the industry and to restrict static recrystallization. During the test, the temperature variation in the three zones was less than ± 3°C. Upon failure, the furnace

37

Chapter 3

was immediately opened and the specimens were quenched by water spray prior to removal. The temperature decrease during quenching was larger than 50°C s-1. Because this rapid quenching produced very high contraction, upon reaching the predetermined value of ε, the computer control was set so as to immediately unload the specimens. The ε distributions over the gauge length were determined from optical photographs (grayscale pixel uncompressed bitmaps) from the face and the profile of each side of the failed specimens, using the GOM – ARAMIS system. For the specimens deformed up to specific ε, the ε distribution was calculated using one picture for the face and one for the profile of the gauge. The images were combined, aligned, converted into black and white and processed by a MATLAB (Ver. R2007b) program, which enabled the calculation of the area reduction per pixel along the extended gauge. Subsequently, the specimens were sectioned and mechanically polished according to the Struers method with diamond suspensions of decreasing grade sizes and finished with a colloidal alumina suspension, OPU, having a particle size of 30nm. For the investigations of microstructure, electron backscattered diffraction (EBSD) was carried out in an XL-30 Scanning Electron Microscope equipped with a Field Emission Gun (FEG-SEM) and a fluorescent screen operating at 20 kV. The data from the orientation imaging microscopy (OIM) were collected and analyzed using the TSL application developed by AMETEC Inc.--EDAX. The step size of 1µm was selected so as to compromise between scanning appreciably large areas in certain time and being able to collect more than 40 data points for the majority of the grains. Electron dispersive X-ray analysis (EDS) was carried out on an XL-30 Environmental FEG-SEM. Subsequently, characteristic areas of the gauge were punched, ground to a thickness of 60 µm, mounted on a copper ring and thinned to

38

Mechanical Properties

perforation using the GATAN precision ion polishing (PIPS) facility. The dislocation microstructures were observed and recorded in a Jeol 2010F transmission electron microscope (TEM) equipped with a Field Emission Gun (FEGTEM), operating at 200 kV.

3.2

Results

3.3 The As-Received Microstructure

Figure 3.1: Automatic Inverse Pole Figures, obtained by EBSD from the as-received AA5182; (a) material 21G and (b) material 37G Figure 3.1 presents the inverse pole figures (IPF) of the as-received 21G (a) and 37G (b) material, demonstrating a weak recrystallized texture, comprising equiaxed

39

Chapter 3

grains. Some small non-indexed areas, no more than a few pixels, can be discerned at the top and right of Figure 3.1(a) and middle and bottom of Figure 3.1(b) corresponding, most likely, to precipitates. These were located very often close and/or exactly at the GB interfaces and were identified as being rich in Mn, Cu and Fe and slightly depleted in Al and Mg. Their size most often was between 0.5 – 1.5 µm, but could in rare occasions reach 10 µm [7]. The average dislocation density as it was determined by TEM was relatively low (i.e. 5 × 1010 ± 1010 m-2).

3.4 Mechanical Properties Representative true stress – engineering strain curves are presented elsewhere [8]. Their shape was completely reproducible up to ε where local necking developed prior to the specimen final fracture. Tests performed at 10-2 s-1 and at T equal to 425°C for both materials (see Section 3.3) showed that “visible” local necking was developed well above 250% ε. Prior to this ε, the variations of the true stress were recorded and were minimal (mostly much lower than 2 MPa) not only along the stress – strain curve of a particular specimen but also between specimens with the same orientation extended at the same testing conditions. We can assume, therefore, that within a certain degree of accuracy and up to the point where notable local necking has developed, the data are representative of a “steady state” deformation. This ε, however, at the onset of local necking, could not be predicted for each specimen despite repeated tests (more than ten) at exactly the same conditions. As a result the maximum elongation to failure showed significantly large error margins. Figure 3.2 shows the variation of the maximum elongation to failure with T for each ε& for the alloy 21G (Figure 3.2a, 3.2c and 3.2e) and 37G (Figure 3.2b, 3.2d and 3.2f). The top two plots represent experiments with specimens cut along the RD, the 40

Mechanical Properties

Figure 3.2: Maximum elongation to failure against temperature at different strain rates, for the material 21G (a), (c) and (e) and 37G (b), (d) and (f). The areas correspond to experiments at cross-head speed, while the scatter and line plots to those at true strain rate.

41

Chapter 3

middle for tests with specimens at 45° and the last two for tests with specimens perpendicular to the RD. The plots are arranged so as to compare the performance of the two alloys at the same conditions. The colored and patterned areas represent experiments at cross-head speed, blue, red and green for tests at initial strain rate, ε&0 of 10-1, 10-2 and 10-3 s-1, respectively. Their spread corresponds to the error on the maximum elongation to failure at each temperature. The scatter and line plots represent values from the tests at true strain rate and are colored dark blue and dark red for the ε& of 10-1 and 10-2 s-1, respectively. It is apparent that at cross-head speed the maximum elongation to failure of the 21G lies consistently above 300% at 425°C and at 10-2 s-1, with values as high as 410% whereas the similar high values at 450°C and at 10-1 s-1 are accompanied by a very large error margin as specimens have demonstrated values as low as 220%. Only for tests with specimens oriented perpendicular to the RD the variation at these conditions was found to be relatively small (i.e. between 340 and 380%). The optimum deformation conditions for the material 21G were reproduced accurately for the experiments carried out at true strain rate, with the specimens along the RD performing slightly better than those cut at 45° and considerably better than those cut perpendicularly to the RD. At cross-head speed and at all the conditions tested, the alloy 37G showed inferior performance at cross-head speed compared with that of the 21G at the optimum conditions. Its only advantage over the 21G was observed at and above 475°C, where its maximum elongation prior to failure was attained systematically at 10-1 s-1 with values close and/or slightly above 300%. At true strain rate conditions along and/or at 45° to the RD the 37G showed the best performance at 500°C and at 10-2 s-1, reaching peak values similar to those of the 21G at

42

Mechanical Properties

425°C and at 10-2 s-1 (i.e. 350%). It seems likely that its behavior may be better at these orientations at T > 500°C. The use of this material at such conditions, however, may almost certainly require the use of lubricants between the material and the mould. This would increase the forming costs and is, therefore, undesirable by the industry. At 90° to the RD, finally, at 500°C and at 10-2 s-1, its performance decreased slightly below 300%, i.e. it exhibited a similar behaviour than that of the specimens of 21G having the same orientation at 425°C and at 10-2 s-1. The steady state true flow stress, σ, was obtained from the stress – strain curves as the average between 5 and 10% ε. These values were compensated by dividing with the Young’s elastic (dynamic) modulus for pure Al [9]: E = 77630 − 12.98 T − 0.03084 T 2

(1)

where E is expressed in MPa. The slope on the plots of the logarithm of the strain rate vs. the logarithm of σ/E (not shown here, for brevity) is equivalent to the stress exponent, n, in the phenomenological equation for creep [1]: p

n

⎛ Q ⎞ ⎛b⎞ ⎛σ ⎞ ε& = A⎜ ⎟ ⎜ ⎟ exp⎜ − c ⎟ ⎝d ⎠ ⎝E⎠ ⎝ RT ⎠

(2)

where A, is a constant that depends on the material the stacking fault energy and the deformation mechanism, b is the magnitude of the Burgers vector, d is the grain size, p is the grain-size exponent, Qc, or Q is the activation energy for creep (and/or viscous glide or climb of dislocations) and R is the universal gas constant. The slope from such plots showed that n, does not vary significantly for the strain rates investigated, despite their large differences. The strain rate sensitivity, m = 1/n is plotted as a function of T for experiments carried at cross-head speed (areas) and true strain rate (line and scatter plots) in Figure 3.3. At cross-head speed and at T lower than

43

Chapter 3

Figure 3.3: Plots of the variation of the strain rate sensitivity, m with temperature for the alloys 21G (a) and 37G (b). The areas correspond to experiments at cross-head speed, whereas the scatter and line plots to those at true strain rate. 400°C, m ranges between 0.20 to 0.26 for the alloy 21G and between 0.20 and 0.23 for the 37G. Throughout the entire T regime the error in the variation of m for the former is significantly larger than that of the latter. m may exceed 0.30 at T > 425°C for the 21G but for the coarser 37G that may occur only between 450 and 475°C. At true strain rate, however, m is significantly lower for both alloys. For the coarser 37G, it ranges between 0.20 and 0.26 at 425°C, down to 0.19 to 0.25 at 500°C with a marginally decreasing trend, whereas for the 21G, m demonstrates a significantly larger scatter (e.g. between 0.14 and 0.28, at 400°C, for the specimens perpendicular to the RD) and on average it increases between 400 and 425°C. These low values between 0.20 and 0.25, are most likely indicative that for both alloys, a combination of viscous glide and climb of dislocation occurs during deformation, with the latter mechanism exerting a larger influence, especially for the experiments at true strain rate. The large scattering, however, demonstrated by the alloy 21G at 400°C, resulting in an m value lower than 0.15, most

44

Mechanical Properties

likely indicates that at this T, the deformation at ε& of 10-2 and especially 10-1 s-1 is very close and/or crosses over to the power-law breakdown regime [4,10-11].

Figure 3.4: Plots of the variation of the activation energy, Q with modulus compensated flow stress, σ/E for the alloys 21G (a) and 37G (b). The areas correspond to experiments at cross-head speed, whereas the scatter and line plots to those at true strain rate. The vertical separation between data at different temperatures for the same alloy (i.e. at constant grain size and constant σ/E) in the plots of the logarithm of the strain rate vs. the logarithm of σ/E, is proportional to the activation energy of the deformation [1214], according to: Q = −R

∂ ln ε& ∂1 T σ

(3) E

Hence, the result of the linear regression of the ε& logarithm with 1/T multiplied by (–R) is equal to Q. These values plotted against σ/E are shown in Figure 3.4. For the alloy 21G, at cross-head speed, irrespective of specimen orientation and at low values of σ/E (i.e. at high T and/or low initial ε& ) Q ranges between 97 and 125 kJ mol-1 having an average value of close to 110 kJ mol-1, significantly lower than that for Mg diffusion in Al, (widely considered equal to 136 kJ mol-1), which is associated with solute drag creep, as

45

Chapter 3

well as viscous glide of dislocations [10,15]. Surprisingly, however, the average value of 110 kJ mol-1 coincides with that for grain boundary sliding [14,16–17]. At the intermediate values of σ/E, Q acquired a range from 116 to 150 kJ mol-1 and an average value of 133 kJ mol-1 which is very close to that of solute drag creep and, finally, at high σ/E, Q settles at 130 kJ mol-1 with a substantially small error margin. Since Q for the selfdiffusion of Al has been considered equal to 143 kJ mol-1 but it can vary from 120.4 up to 144.4 kJ mol-1 depending on the method of calculation [6,18], it is not surprising that for the material 37G, at low σ/E (i.e. at high T and/or low initial ε& ) Q ranges between 140 and 156 kJ mol-1, with average values of 142, 153 and 147 kJ mol-1 for the specimens along, at 45° and perpendicular to the RD demonstrating that at least at these conditions the self diffusion of Al plays a dominant role in the deformation (i.e. dislocation climb seems to be the rate controlling mechanism of deformation). Due to the high temperatures, however, the presence of some of the alloy species present in the matrix (i.e. presumably Cr, Fe and Cu) may provide obstacles in the dislocation motion, at least in high temperatures. With decreasing T, however, (and the corresponding σ increase) Q consistently decreases below 140 kJ mol-1 indicating that at low T dislocation motion may be affected by the Mg diffusion. At true strain rate (i.e. the data depicted by the line and scatter plots) and for the alloy 21G, at intermediate values of σ/E, Q has values close and/or slightly lower than that of the Mg diffusion but as the σ/E increase (at low T and/or high σ) Q become very high reaching values close to 180 kJ mol-1. The variation of Q with σ/E showed similar trends irrespective to the orientation of the specimens and the high values may be indicative of obstacles to the dislocations, either by coherent precipitates or by segregation of alloy species, especially close to the grain boundary

46

Mechanical Properties

interface, which inhibit their intragranular motion. Similar high values of Q are observed for the alloy 37G. These, however, were consistently calculated, irrespective of the specimen orientation, only at low to intermediated values of σ/E and were close to 170– 180 kJ mol-1. As the value of σ/E increased, Q showed a decrease to values between that of Al self-diffusion, for the specimens oriented perpendicularly to the RD and/or that of the Mg diffusion in the fcc Al, for those at 45° to the RD. For the specimens along the RD, however, Q increased again to an average value ≈ 170 kJ mol-1, with increasing σ/E, demonstrating, perhaps, a notable anisotropic response for this alloy at low T and/or at high ε& . Table 1 Lankford Coefficients and the parameters r and ∆r for the planar anisotropy of the alloys 21G and 37G. Alloy / Testing Conditions

r0

r45

r90

r

∆r

-2

-1

0.935

0.961

1.0069

0.966

0.009

-2

-1

21G / 425°C, 10 s , failure

1.106

1.112

1.071

1.100

-0.024

21G / 425°C, 10-1 s-1, failure

1.031

1.085

1.096

1.074

-0.022

1.017

0.963

1.036

0.995

0.064

21G / 425°C, 10 s , 50%

-2

-1

-2

-1

37G / 425°C, 10 s , 50% 37G / 475°C, 10 s , failure

1.163

1.043

1.185

1.109

0.131

37G / 500°C, 10-2 s-1, failure

1.130

1.165

1.066

1.132

-0.067

37G / 475°C, 10-1 s-1, failure

1.059

1.161

1.137

1.130

-0.063

1.025

1.124

1.079

1.088

-0.072

-1

-1

37G / 500°C, 10 s , failure

3.5 Anisotropy and Necking Instabilities To calculate the planar anisotropy of Al alloys fabricated in the form of sheets, first specimens along, at 45° and perpendicularly to the RD were extended up to a ε prior to the development of diffuse necking, i.e. where σ shows still an increasing trend with ε, within the uniform plastic deformation range. Study of the stress – strain curves, showed

47

Chapter 3

that the onset of diffuse necking was systematically above 50% ε for both materials deformed at 10-2 s-1 and thus the straining of selected specimens up to 50% would ensure that the specimen was well within the uniform plastic deformation regime. Even though this value is above 20% (according to ASTM E517), it was considered appropriate for specimens exhibiting superplastic properties, since the planar anisotropy has been also customarily estimated at half the fracture strain and/or after the specimen has failed [20]. Optical photography of the specimens using the GOM – ARAMIS system was instrumental in determining the width and the thickness ε, i.e. εw and εt, respectively. The Lankford Coefficients as a measure of the plastic anisotropy of a rolled metal sheet, rθ, the r and the ∆r values for both alloys, are presented in Table 1, as they were determined for specimens deformed up to 50% ε and from uniformly deformed regions of the gauge in specimens extended up to failure. They were calculated according to the formulae: rθ =

εw εt

r=

r0 + 2r45 + r90 4

∆r =

r0 − 2r45 + r90 2

(4a, b, c)

It becomes immediately apparent that the material 21G is essentially isotropic, whereas the 37G is moderately anisotropic, due to its slightly high ∆r value. Experiments up to 50% for the alloy 37G at 475 and at 500°C were not carried out since they may produce only marginally better values. Extensions at 425°C and at 10-2 s-1 up to 200, 250 and 350% ε showed that the material 21G deformed quite uniformly up at 250% (e.g. Figure 3.5), but profuse local necking developed at 300% ε. The material 37G, however, showed significant local ε variations even at 200% and, furthermore, not only significant necking but numerous secondary necking instabilities at 250%. Due to the significantly better performance of

48

Mechanical Properties

the alloy 21G, which can potentially fulfill the industrial criteria, this chapter focuses, henceforth, primarily on this alloy.

Figure 3.5: Images of an as-received specimen from the alloy 21G (a), specimens extended up to 250% strain at 425°C and at 10-2 s-1, along (b), at 45° (c) and perpendicularly to the RD (d). (e) Distribution of the local engineering strain as it was determined from images obtained with the GOM – ARAMIS system. Deforming up to failure at the optimum conditions, however, (i.e. at 425°C and at 10-2 s-1) very often resulted in the development of secondary necking instabilities, an unpredicted but frequent behavior, especially when elongations in excess of 400% were achieved [21]. Figure 3.6 shows a specimen of the alloy 21G extended up to almost 500% showing two secondary necking instabilities one to the left and one to the right of the primary failure point. A statistical study of all specimens extended up to failure that 49

Chapter 3

involved photographing each specimen using ARAMIS and measuring the distance between successive instabilities, including secondary instabilities as well as primary failure points, from one side of the specimen’s gauge to the other produced Figure 3.7.

Figure 3.6: Specimen from the alloy 21G deformed at 425°C and at 10-2 s-1 up to 500% strain showing secondary necking instabilities; the area HT was subjected only to static recrystallization, SN denotes the secondary necking, UDT and UDG are the adjacent uniformly deformed regions and T is the area tip. A third, but weaker necking instability can be seen on the left (at approximately 1.5 cm from the left grip).

Figure 3.7: The variation of the distance between successive instabilities with T for each ε& , for the alloy 21G (a) and 37G (b). For the alloy 21G, the distance between successive instabilities maximizes at the optimum deformation conditions and at cross-head speed with ε&0 of 10-2 s-1 it showed a slightly larger spacing compared with that at true strain rate. At 10-1 s-1, largest spacings were exhibited by the specimens deformed at true strain rate. For the material 37G the maximum in the distance between successive instabilities was observed at 475°C. The

50

Mechanical Properties

apparent decrease at 500°C occurred solely because at this T, they were far more numerous, i.e. occasionally three and more secondary instabilities were detected in a single specimen, and as a result their apparent spacing decreased. The experiments at cross-head speed systematically showed larger distances between successive necks. For both alloys, finally, they were most likely to appear in specimens along the RD; their presence in specimens at 45° was less frequent and in those perpendicularly to the RD was particularly rare. Table 2 Data obtained from the EBSD analysis. The comments refer to the transition from UDG to secondary necking instabilities SN and to UDT (i.e. UDG→SN→UDT). Parameters / Spec. Orientation Grain Size (µm) / All Orientations

Grain Partition UDG

SN

UDT Comments

Average Deformation Recovery Recrystallization Cube Texture Deformation Volume Fraction (%) Recovery / All Orientations Recrystallization Goss Texture Deformation Volume Fraction (%) Recovery / All Orientations Recrystallization Grain Deformation Volume Fraction (%) Recovery /Along the RD Recrystallization

26.6 35.3 24.0 15.5 26.0 16.2 14.9 17.9 14.5 8.6 6.2 11.8 82.1

24.2 32.5 20.3 14.5 30.1 18.7 20.4 18.9 18.0 11.7 7.4 11.4 81.3

25.6 33.9 22.2 15.7 27.9 18.7 22.2 16.2 17.8 10.0 6.9 13.8 79.4

Grain Volume Fraction (%) /At 45° to the RD Grain Volume Fraction (%) / Perpendicular to RD Specimen Orientation Along the RD

6.3 14.9 78.8 6.5 14.8 78.8 UDG

11.9 19.4 68.7 5.1 11.3 83.6 SN

7.5 15.3 77.2 10.2 15.8 74.0 UDT

56 32 12 58 32 10 69 23 8

61 25 14 55 28 17 60 34 6

56 31 12 57 30 13 66 21 13

Deformation Recovery Recrystallization Deformation Recovery Recrystallization GBs Vol. Fr. (%) HAGBs SGBs LAGBs At 45° to the RD HAGBs SGBs LAGBs Perpendicular to RD HAGBs SGBs LAGBs

Grain Refinement: Change UDG→SN→UDT Decrease by 8%, Increase by 4% Decrease by 15%, Increase by 9% Decrease by 6%, Increase by 8% Local Maximum on SN Increase → SN and constant → UDT Increase → SN and increase → UDT Local Maximum on SN Increase → SN and decrease → UDT Increase → SN and decrease → UDT Small increase → SN very small decrease → UDT Slight decrease → SN and increase → UDT Marginal decrease UDG → SN → UDT Large increase → SN and large decrease → UDT Large increase → SN and large decrease → UDT Large decrease → SN and large increase → UDT Marginal decrease → SN, large increase → UDT Decrease → SN and increase → UDT Increase → SN and large decrease → UDT Behaviour on SN Local Maximum Recrystallization Dominant Local Minimum Small Local Maximum Small Local Minimum Small Local Minimum Sharp Local Maximum Recovery Dominant Sharp Local Minimum Sharp local MaximumDeformation Dominant Constant increaseExtensive Recovery in UDT

51

Chapter 3

The cumulative results of the EBSD analysis on the secondary necking instabilities (SN) and on those of the adjacent uniformly deformed sections (i.e. UDG and UDT, for closer to the grip and tip, respectively), with the microstructure partitioned in deformed, recovered and recrystallized grains, according to the criteria outlined in [8], are presented in Table 2. In summary, Table 2 shows that grain refinement occurred at the secondary necking instabilities. The Cube and Goss texture of the deformed grains maximized. For the specimens along the RD, where the appearance of secondary necking instabilities were most frequently observed, the variations on the volume fractions of the three partitions were minimal with recrystallization being extensive, suggesting that the internal structure and orientation of the grains is mostly responsible for the necking development (i.e. the necking most likely indicates the conclusion of significant internal grain deformation that was followed by recovery and then significant recrystallization). For those at 45° to the RD, deformation is followed by extensive recovery, which results in a comparatively increase of the low-angle grain boundaries at the expense of sub-grain boundaries (SGBs), whereas for the specimens oriented perpendicularly to the RD the apparent increase in the volume fraction of the sub-grain boundaries at the secondary instabilities (deformation) serves as a starting point for more extensive recovery closer to the tip, where a higher volume fraction of low-angle grain boundaries was observed.

3.6 Deformation Microstructures EDS of the matrix of the deformed specimens showed that it contained 5.38±0.33 at.% Mg, 0.12 ± 0.01 at.% Mn, 0.08 ± 0.01 at.% Cu, 0.04 ± 0.01 at.% Fe and 0.01±0.01

52

Mechanical Properties

at% Cr and the balance in Al. Figure 3.8, 3.9 and 3.10, show the characteristic microstructures of the areas denoted UDG, UDT and SN, in Figure 3.6. The local engineering ε in these areas was, 150, 250 and 500%, respectively. Figure 3.8a and 3.8b, obtained in strong beam condition close to the [001] pole. Figure 3.8a shows a bimodal deformation often encountered within a single grain, with two sub-grain boundaries on the left and a dislocation density approximately equal to 4×1011 ± 1011 m-2 and a very high dislocation density on the right, i.e. 4×1015±1015 m-2. Figure 3.8b shows three subgrain boundaries close to a triple junction in an area relatively free of dislocations. Figures 3.9a, 3.9b and 3.9c, obtained also at [001] beam direction, show microstructures with more uniform dislocation distribution and densities between 1014 and 1015 m-2. The microstructure of each grain showed between 6 and 10 coherent precipitates (i.e. no size and strain field change, was observed when they were tilted between ±g), with diameter ≈ 0.5 and 1.0 µm (e.g. Figure 3.9a) and very often were observed close and/or exactly at the grain boundaries. Some of them acted as sub-grain boundary initiation sites (e.g. Figure 3.9a and 3.9c), which further subdivided the grain structure into sub-grains (Figure 3.9b). In Figure 3.10, the number of sub-grain boundaries decreased markedly. Only a few were observed relatively away from precipitates (e.g. Figure 3.10a). Most grains demonstrated exceptionally uniform dislocation distribution with density ≈ 1015 m-2. Precipitates in the grain interior did not produce dislocation entanglements (i.e. they were successfully cut by dislocations, e.g. Figure 3.10b), whereas they successfully pinned the motion of the low-angle grain boundaries / high-angle grain boundaries (Figure 3.10c). In summary, slip and climb occur simultaneously with bimodal deformation in the UDG and more uniform distribution in UDT. In those regions, most of the sub-grain

53

Chapter 3

boundaries develop in the vicinity and/or emanate from submicrometer-sized coherent precipitates. In the secondary necking instabilities, the deformation seems to deviate from a “core and mantle” microstructure. The intergranular dislocation motion occurs with relative ease despite the presence of precipitates. Sub-grain boundaries are few and isolated from those precipitates inside the matrix, whereas the precipitates at the GB interfaces act as efficient pinning agents of the low-angle grain boundary and/or highangle grain boundary migration.

Figure 3.8: Representative microstructures in section UDG: (a) A bimodal deformation showing sub-grain boundaries (left) and high dislocations density(right); (b)Three subgrain boundaries around a triple junction in a microstructure relatively free of dislocations.

54

Mechanical Properties

Figure 3.9: Representative microstructures in section UDT: (a) A precipitate initiating a sub-grain boundary (top) and two more sub-grain boundaries, emanating from triple junctions with a sub-grain (shown at the bottom part of the image); (b) a sub-grain boundary by sub-grain boundaries (white arrowheads) with two precipitates at the topright corner (black arrowheads) (c) a long sub-grain boundary formed at a junction next to a precipitate.

55

Chapter 3

Figure 3.10: Representative microstructures in the secondary necking section: (a) A long sub-grain boundary far from a precipitate; (b) A high density of uniformly distributed dislocations crosses a precipitate and (c) a precipitate pinning a high angle grain boundary.

56

Mechanical Properties

3.7 Maximum Elongation, Anisotropy and Failure Previous studies on the mechanical properties of AA5182, at the same ε& (crosshead speed) showed a maximum elongation to failure that was consistently below 150% , even though the post-deformation grain size measured at the grip of the specimens was between 19 and 24 µm [22], i.e. similar with the as-received alloy 21G. Most tests, however, were carried out below 400°C. A test at 450°C and at 10-1 s-1 showed a maximum elongation to failure of only 95% and this was explained by the fact that the parameter ε&0 /D was larger than the value 1013 m2, which indicates the onset of power-law breakdown (PLB) [23]. D was the lattice self-diffusivity of Al (estimated by using D0 = 1.7×10-4 m2 s-1 and Q = 142 kJ mol-1). Even though comparison in the previous section was carried out using this value for the Al self-diffusivity, an average value from experimental and theoretical parameters recently reported [18,19], produced D0 = 8.675×10-5 m2 s-1 and Q = 128.6 kJ mol-1. Using the latter, the parameter ε&0 /D becomes equal to 2.25×1012 m2, i.e. lower than that indicating power-law breakdown. This, as well as the low values reported for the maximum elongation to failure for the Al-Mg-Mn ternary alloys (even though the latter can be attributed to exceptionally large grain size variations [4]) cannot be easily reconciled with the large values in the present study. In the first study [22], the samples were allowed to stabilize in the testing T for over 1 hour. A heat treatment of the material 21G at 450°C for 1 hour showed a significant grain growth with the average grain size reaching 50 µm, indicating that the present alloys with the weakly recrystallized texture is essentially in a metastable state (i.e. significant internal energy is still stored resulting in a large driving force for grain boundary motion).

57

Chapter 3

Thus, the good properties demonstrated by the material 21G are most likely due to the fabrication methods. Examination of the large elongations of the specimens along to the RD using EBSD showed that continuous dynamic recrystallization (CDRX) produced an almost stable average grain size with increasing local ε (with values between 22 and 30 µm in most of the gauge of the post-mortem specimen). The specimens oriented perpendicularly to the RD, however, demonstrated an almost continuous grain refinement from 28 down to 17 µm, which most likely triggered the onset of discontinuous dynamic recrystallization and resulted in premature final fracture [8]. This is the main reason for the specimens that were oriented perpendicularly to the RD exhibiting lower values of maximum elongation prior to failure. In this picture, the larger values demonstrated when the specimens were extended at cross-head speed can be explained in terms of a combination of a continuously decreasing true ε& and heating, which resulted in an increased recrystallization rate, reduced grain refinement (i.e. the rate of progress of the continuous dynamic recrystallization was lessened), rendering the average grain size more stable. This is, most likely, the main reason for the similarity in the values of the maximum elongation to failure at cross-head speed irrespective of specimen’s orientation. Failure occurred after the onset of discontinuous dynamic recrystallization which resulted in rapid local necking and final fracture. In summary, the metastable microstructure of the alloy 21G, most likely responsible for its isotropic response, its fairly uniform elongation up to 250% and its maximum elongation to failure between 300 and 400%, indicates that this alloy is capable of fulfilling the criteria for automotive applications (i.e. uniform elongation between 200

58

Mechanical Properties

and 300% at high ε& ). On the other hand, the large grain size of the alloy 37G showed good maximum elongation at higher T, but its anisotropic nature and the numerous necking instabilities render it quite unfavorable for industrial use. The large variations in the maximum elongation to failure, however, suggest that proper pressure profiles have to be investigated during blow forming, if the 21G AA5182 is to be utilized successfully for automotive products.

3.8 Flow Instabilities and Dislocation Microstructure Secondary necking instabilities produced by unstable material flow (i.e. flow localization) have been observed when subjecting sub-micrometer sized aluminum alloys in high ε& elongations [24], but they were considerably more frequent for coarse grain alloys [4,21]. Shorter gauge length has been documented that it is instrumental in attaining more uniform elongations [25], whereas cavity interlinkage (and not cavity nucleation and growth) limits the uniformity in elongation leading to “quasi-superplastic” behavior [26]. In the present case, the gauge length was only 16 mm, and no significant cavitation interlinkage (not even nucleation and growth) was observed in the regions where secondary necking instabilities instabilities developed. It is more likely that lack of cavitation interlinkage are associated with the decrease of the number of grains in the specimen’s cross-section and/or the increase in the probability of cross-sections containing an uneven volume fraction of soft and hard grains. Thus, cross-sections with more soft grains, i.e. grains in orientations where many slip systems can operate easily and accomplish the imposed deformation, will deform (thin down) preferentially. Indeed, the texture analysis showed that the Cube and Goss component of the deformed grains

59

Chapter 3

maximized in these regions. Grains of material with fcc lattice which possess Cube and Goss orientation have both eight (out of the twelve) active slip systems. Given the fact that the grains with Goss orientation are rotated by 45° around an axis parallel to the RD with respect to that of grains with Cube orientation, their coherency, may probably favour the rotation of deformed grains into alternating Cube and Goss orientations, so as to allow for the operation of multiple slip systems across the grain boundary interfaces. Thus microstructures with a more evenly balanced Cube and Goss texture, are most likely softer and lead to large local elongations. This behavior would be detected in the stress – strain curves only by using an extensometer (an optical device with similarities to the GOM – ARAMIS system), which is capable of dividing the gauge into sections, measure all local elongations across all sections comprising the specimen’s gauge, and record one stress – strain curve for each section. As detected by this method, the development of a secondary necking, a sudden increase in the applied stress would be recorded for that section (larger local elongation would result in a thinner cross section). In secondary necking local area, continuous dynamic recrystallization has obviously progressed further leading to grain refinement (i.e. a decrease in the average grain size by 5 – 15%). This is accompanied by a slight coarsening in the adjoining uniformly deformed sections. However, the uniformly deformed region that lies closer to the tip is not expected to exhibit grain coarsening up to the same value as that closer to grip, since the former is more likely to have experienced more deformation (i.e. the continuous dynamic recrystallization progressed further). This refinement was observed mostly for the grains exhibiting deformation and especially recovery but their coarsening closer to the tip was about 4 and 9%, respectively. Only the large volume fraction of the

60

Mechanical Properties

grains that showed recrystallization coarsened up to a size that was slightly higher than that prior to the appearance of the secondary necking instabilities (i.e. coarsened by 8% after the secondary necking instabilities, as opposed to a 6% refinement). Increased recrystallization, portrayed as the conclusion of internal grain deformation and recovery, characterizes the secondary necking instabilities of the specimens along the RD, whereas extensive recovery, associated with the sharp increase of the volume fraction of the lowangle grain boundaries [8], characterizes the secondary necking instabilities of the specimens oriented at 45° to the RD. Consistent with the continuous grain refinement of the specimens oriented perpendicularly to the RD, the instable region demonstrates extensive deformation and serves as a transition area to a more gradual recovery closer to the tip, as the abrupt increase of the sub-grain boundaries in the secondary necking instabilities was succeeded by their conversion to low-angle grain boundaries in the UDT. This behavior at the optimum deformation conditions (i.e. at 425°C and 10-2 s-1 for the alloy 21G and at and above 475°C and at 10-2 s-1 for the 37G) may potentially decrease the applicability of the coarse grained AA5182. Unstable flow is highly undesirable, since it may lead to components of uneven thickness, in regions where forming progressed further than an equivalent extension of 250%. The study of the TEM microstructures showed sub-grain boundaries in all three regions of the gauge. They were far more numerous in the UDT, most appeared to originate from coherent precipitates, and subdivided the microstructure into sub-grains. They were less numerous in the UDG region; the dislocation density in their vicinity was 4 × 1011 ± 1011 m-2 and they appeared to be relatively isolated from regions of extensive dislocation glide (with density as high as 4 × 1015 ± 1015 m-2) and, finally, only a couple were observed in the secondary necking

61

Chapter 3

instabilities. Their presence in the uniformly deformed regions suggests that, at these conditions, the material follows the high stress “five-power-low” creep, where recovery dominates and dislocation climb, as manifested by the sub-grain boundary formation, is rate controlling [10,27-29]. Thus in the secondary necking instabilities instabilities, where the cross-section has decreased (and the effective applied stress has increased) the material crosses over to the power-law-break-down regime, where thermally activated glide presumably dominates [11,30], with the coherent precipitates pinning effectively the drastic grain modification of the microstructure [31,32].

3.9 Mechanical Properties and Deformation Behavior The fact that the material response lies between the power five law and powerlaw-break-down regime is reflected in the low values of m. These lie between 0.20 and 0.30 at cross-head speed and for the alloy 37G at true strain rate, but can decrease significantly below 0.20 for the material 21G at true strain rate, around the optimum deformation conditions, especially for the experiments carried out along the rolling direction (the thickest olive green line and scatter plot in Figure 3.4a). At 425°C, the specimens oriented perpendicularly to the RD exhibit the largest scatter in the value of m, presumably due to extensive continuous dynamic recrystallization which produces an increasingly finer grain size, thus leading to premature failure (i.e. a low maximum elongation). Admittedly, these values were calculated from the values of σ between 5 and 10% ε, i.e. prior to the development of diffuse necking and/or the onset of secondary necking instabilities, but there seems to be an agreement with the TEM observations of the post-mortem microstructures as well as with the interpretation of the macroscopic

62

Mechanical Properties

cross-section decrease (with the corresponding increase in the σ) during the development of secondary necking instabilities. One further evidence supporting the response type of the present materials is provided by the values of σ/µ, which at 425°C and at 10-2 s-1 for the material 21G is equal to 1.8×10-3. For the alloy 37G at the same ε& and at 475 and 500°C are equal to 1.3–1.4×10-3 and 1.1×10-3. All these values are very close to 10-3 (i.e. marginally higher), which characterizes the transition to power low behavior [33]. All tests at ε& equal to 10-1 s-1, however, even at cross-head speed, lie well within the powerlaw break down regime. These affect mostly Q at high σ/E (low T and/or high σ), which for the material 21G increase consistently up to 185 kJ mol-1, and for the material 37G reach 170–180 kJ mol-1 at low to intermediate values of σ/E (low T) and 170 kJ mol-1 at high σ, for specimens along the RD. One important issue that is necessary to clarify concerns the validity of the thermally activated glide and obstacle controlled glide [11,34-36] in the region of powerlaw-break-down. These studies, showed that the former is associated with Q between 171 and 175 kJ mol-1 and that the high σ associated with this regime indicate that the dislocations have broken away from their solute atmosphere. The latter, on the other hand, is associated with Q lower than that of self-diffusion. These mechanisms are essentially the same [37], but in the present case some considerations regarding the nature of these obstacles may prove quite useful in explaining the high dislocation densities encountered in the TEM microstructures and the deviation from the core and mantle model (i.e. the unexpected uniform dislocation distribution) at the secondary necking instabilities. The average critical velocity, υc , above which solute atmospheres cannot be formed around the dislocations is given by the relationship [6]:

63

Chapter 3

υc =

3 + 2 2 A Ds 2 b 2 k BT

(5)

where A is the elastic energy interaction equal to 3µbΩεa/π, µ the temperature dependent shear modulus, b the magnitude of the Burgers vector (bAl = 2.8635 Å), Ω the atomic volume of a particular solute in Al, εa the absolute value of the misfit strain i.e. the absolute value of the change of the lattice constant after replacing an Al with a solute atom (or the solute-solvent size difference [29]) and Ds is the diffusivity of the solute. Here, the volumes of all solutes in fcc Al and the absolute value of the linear size factor (lsf) were used for Ω and εa, respectively [38]. For the diffusivity of each solute, the average from experimental and theoretical values were used (i.e. D0-Mg = 4.521×10-5 m2 s1

, QMg = 125.7 kJ mol-1, D0-Mn = 3.051×10-2 m2 s-1, QMn = 219.8 kJ mol-1, D0-Cu =

2.11×10-4 m2 s-1, QCu = 127.2 kJ mol-1, D0-Fe = 4.178×10-1 m2 s-1, QFe = 213.2 kJ mol-1, D0-Si = 4.604×10-5 m2 s-1, QSi = 121.8 kJ mol-1, D0-Cr = 3.534×10-1 m2 s-1, QCr = 251.8 kJ mol-1 and D0-Ti = 1.12×10-1 m2 s-1, QTi = 241.4 kJ mol-1) [18,19]. Figure 3.11, shows the -2 -1 -4 υc , as a function of T. For ε& equal to 10 s an average dislocation velocity υ of 10 m

s-1 requires a mobile dislocation density of 3.5×1011 m-2, i.e. four orders of magnitude lower that the total dislocation density observed. Since Mg concentrations more than 2 at.% result in super-saturation of the dislocation cores, it can be concluded that the solute drag of mobile dislocations by Mg solutes dominates in AA5182 [23]. Even though Mn, Cu, Fe, Si and Cr have been mostly detected in the precipitates, a possible minute content in the matrix of both alloys is expected, but it is well below their solubility limit in Al. These atoms are approximately 46, 38, 37, 16% and 57% smaller than the Al atoms, and

64

Mechanical Properties

since Cu and Si can also reach the dislocation cores they may well contribute to the solute drag, or even pin the dislocation motion.

Figure 3.11: Critical dislocation velocity for the formation of solute atmospheres in AA5182. The gray shaded area corresponds to the T regime used in the present study. The velocities above the curve of a particular solute correspond to values where that solute is unable to reach the dislocation cores, whereas below solute drag and/or dislocation pinning occurs by that solute. In summary, solute drag by Mg atoms dominates at ε& = 10-2. Cu and Si atoms present in the matrix inhibit the dislocation motion. Dislocations that encounter occasional Fe, Mn, Cr or Ti atoms may be arrested and rendered immobile requiring dislocation multiplication mechanisms for continued deformation. Dislocation glide and climb occur concurrently with the latter being the rate controlling mechanism. The coherent precipitates inside the grain microstructure can be cut successfully by dislocations especially when the applied stress locally increases. The local ε difference by 250% between secondary necking instabilities and UDT of Figure 3.6, corresponds to an increase of approximately 2.5 times in the effective stress and, consequently, by a maximum of 2.55 ≈ 100 times in the average dislocation velocity. Hence, if the mobile

65

Chapter 3

dislocations move with a υ ≈ 10-4 m s-1 in the uniformly deformed regions, they will move with a maximum υ ≈ 10-2 m s-1 in the secondary necking instabilities and thus, most will escape their solute atmospheres. In such cases, their motion will be controlled strictly by thermal activation processes that exhibit high Q values. The contribution of the tests at 10-1 s-1, which lie well into the power-law breakdown regime is, most likely, the reason for the increase of Q, from that of the Al self-diffusion and Mg diffusion in the fcc Al, considered in this study (i.e. 128.6 and 125.7 kJ mol-1, respectively) to those calculated. Finally, the average value of 110 kJ mol-1 for the material 21G at low σ/E shows that at these conditions grain boundary sliding becomes important. But due to the presence of coherent precipitates grain boundary sliding is inhibited and thus, failure occurs rapidly with a very low maximum elongation.

3.10 •

Conclusions

The coarse-grained AA5182 aluminum alloys 21G and 37G, exhibited optimum deformation conditions at ε& equal to 10-2 s-1 and at T equal to 425°C and above 475°C, respectively, with maximum elongations to failure between 300 and 400% along, and at 45° to the RD and approximately equal to 300% perpendicularly to the RD. These large values are mostly due to their metastable, weakly recrystallized as-received microstructure.

•

The alloy 21G is essentially isotropic exhibiting consistently uniform deformation at all orientations up to an elongation of 250%, at 425°C and at 10-2 s-1, whereas the 37G is slightly anisotropic and demonstrated significant deviations from uniformity

66

Mechanical Properties

at 200%, as well as significant necking and numerous secondary necking instabilities at 250%. •

Secondary necking instabilities were most likely to appear at the optimum deformation conditions, especially when elongations in excess of 400% were achieved, but their development could not be very well predicted. For both alloys they were most likely to appear in specimens along the RD; their presence in specimens at 45° was less frequent and in those perpendicularly to the RD was particularly rare. They are associated with regions of the gauge that contain a large volume fraction of soft grains and produce microstructures that exhibit maxima in the Cube and Goss component of the deformed grains and slight grain refinement compared with the adjoining thicker and more uniformly deformed regions.

•

The deformation of both alloys at 10-2 s-1 lies at the border between the high stress power five law regime, where dislocation climb is the rate controlling mechanism and the power-law breakdown phenomenon.

•

The microstructure of the secondary necking instabilities is indicative of power-law breakdown deformation, where the dislocations have acquired high average velocity and thus have managed to break away form their solute atmospheres. Hence thermal activation processes with high Q control the deformation.

•

The high activation energies calculated in this study are most likely due to the influence of the tests at 10-1 s-1 which lay well in the power-law breakdown regime.

•

At high T and low ε& , Q is close to that of grain boundary sliding but the coherent precipitates at the grain boundary interfaces pin effectively their motion and as a result the maximum elongation of both alloys is very low.

67

Chapter 3 References [1]

Sherby OD, Wadsworth J. Prog Mater Sci 1989;33:169.

[2]

Figueiredo RB, Kawasaki M, Xu C, Langdon TG. Mater Sci Eng A 2008;493:104

[3]

Valiev RZ, Islamgaliev RK, Alexandrov IV. Prog Mater Sci 2000;45:103

[4]

Taleff EM, Henshall GA, Nieh TG, Leseur DR, Wadsworth J. Metall Mater Trans A 1998;29:1081.

[5]

Woo SS, Kim YR, Shin DH, Kim WJ. Scripta Mater 1997;37:1351.

[6]

Soer WA, Chezan AR, De Hosson JThM. Acta Mater 2006;54:3827

[7]

Chen Z, Kazantzis AV, De Hosson JThM. Mat.-wiss u Werkstofftech 2008;39:259.

[8]

Chen Z, Kazantzis AV, De Hosson JThM. Acta Mater (submitted).

[9]

Köster W. Z Metallkd 1948;39:1.

[10]

Yavari P, Langdon TG. Acta Metall 1982;30:2181.

[11]

Soliman MS. J Mater Sc 1987;22:3529.

[12]

Barrett CR, Ardell AJ, Sherby OD. Trans ASM 1964;20:200.

[13]

Bae DH, Ghosh AK. Acta Mater 2000;48:1207.

[14]

Kulas M-A, Green WP, Taleff EM, Krajewski PE, McNelley TR. Metall Mater Trans A 2005;36:1249.

[15]

Sherby OD, Taleff EM. Mater Sci Eng A 2002;322:89.

[16]

Kulas M-A, Green WP, Taleff EM, Krajewski PE, McNelley TR. Metall Mater Trans A 2006;37:645.

[17]

Green WP, Kulas M-A, Niazi A, Oishi K, Taleff EM, Krajewski PE, McNelley TR. Metall Mater Trans A 2006;37:2727.

[18]

Simonovic D, Sluiter MH. Phys Review B 2009;79:054304.

[19]

Du Y, Chang YA, Huang B, Gong W, Jin Z, Xu H, Yuan Z, Liu Y, He Y, Xie F-Y. Mater Sci Eng A 2003;363:140.

[20] Kocks UF, Tomé CN, Wenk H-R. Texture and Anisotropy, Cambridge: Cambridge University Press; 2000. p. 421. [21] Kazantzis AV, Chen Z, De Hosson JThM, Chezan AR, Zhuang L, Rathev P, De Smet P. In: Hirsch J, Skrotzki B, Gottstein G, editors. Aluminium Alloys: Their Physical and Mechanical Properties. Proceedings of the 11th International Conference in Aluminium Alloys. Aachen, Germany: Wiley – VCH Verlag GmbH; 2008, p. 1423. [22]

Taleff EM, Nevland PJ, Krajewski PE. Metall Mater Trans A 2001;32:1119.

[23]

McNelley TR, Michel DJ, Salama A. Scripta Metall 1989;23: 1657.

[24]

Xu C, Furukawa M, Horita Z, Langdon TG. Acta Mater 2003;51:6139.

[25]

Mohamed FA, Langdon TG. Acta Metall 1981;29:911.

[26]

Langdon TG. Metal Sc 1982;16:175.

[27]

Yavari P. Mohamed FA, Langdon TG. Acta Metall 1981;29:1495.

[28]

Kassner ME, Pérez-Prado M-T. Prog Mater Sci 2000;45:1.

[29]

Mohamed FA, Langdon TG. Acta Metall 1974;22:779.

[30]

Pharr GM. Scripta Metall 1981;15:713.

68

Mechanical Properties [31]

Sitdikov O, Sakai T, Avtokratova E, Kaibyshev R, Tsuzaki K, Watanabe Y. Acta Mater 2008;56:821.

[32]

Humphreys FJ, Hatherly M. Recrystallization and related annealing phenomena. Oxford: Elsevier; 2004.

[33]

Sherby OD, Burke PM. Prog Mater Sci 1968;13:325.

[34]

Soliman MS, Mohamed FA. Metall Mater Trans A 1984;15:1893.

[35]

Raj SV, Langdon TG. Acta Metall Mater 1991;39:1817.

[36]

Raj SV, Langdon TG. Acta Metall Mater 1991;39:1823.

[37]

Kocks UF, Argon AS, Ashby MF, Prog Mater Sci 1975;19:110.

[38]

King HW. J Mater Sci 1966;1:79.

69

4 Deformation Processes The application of Electron Backscattered Diffraction, (EBSD), enabled remarkable studies in the microstructure development of single phase Al – Mg alloys with increasing degrees of rolling reduction, as well as in the grain boundary, (GB) characterization and texture evolution of superplastic Al – Li and Al – Cu – Zr alloys with increasing strain, ε , in tension [1-3]. The novelty of the approach followed in the present chapter, relies on the fact that the EBSD maps were initially partitioned into deformed, recovered, (RV) and recrystallized, (RX) portions. Subsequently, the evolution of the texture, the grain size, the grain volume fraction, the sub-grain boundary (SGB), low-angle boundary (LAGB) and high-angle grain boundary (HAGB) density were investigated using prescribed, sequentially increasing incremental values of the local strain, ε L , along the post-mortem gauge of tensile specimens. This approach permitted the comparative evaluation of the progress of these three processes, and thus produced important conclusions with respect to the total mechanism responsible for the microstructure modification along the gauge of specimens from AA5182 aluminum alloy upon “superplastic” extension. Studies on the evolution of the texture as a function of strain, ε , in terms of grain size, sub-grain boundary, low-angle grain boundary and high-angle grain boundary volume fraction, length and densities will most certainly help in understanding the deformation mechanisms responsible for the modification of the microstructure. Such knowledge is the first and most important step in the understanding of the macroscopic constitutive behaviour of these materials and will help in the development of materials with enhanced formability at elevated temperatures [4]. In the same spirit, instrumental in our present work is the partitioning of the EBSD maps into deformed, recovered and

70

Deformation Processes

recrystallized regions. This was accomplished by means of a criterion which is based on the internal structure of individual grains. The comparative development of these three processes was characterized using a series of microstructural parameters calculated by the analysis of OIM data with increasing ε L . Finally, we explain these results in terms of the current theory of the high T superplastic deformation of aluminum alloys.

4.1 Experimental Procedure The material used in the present study was commercial aluminum AA5182 alloy with composition Al – 5.0 wt. % Mg – 0.3 wt. % Mn – 0.1 wt. % Cu, minor impurities of Si, Fe, and Cr and an as-received grain size of approximately 21 µm (20.7 µm). Specimens for tensile tests were produced by Electrical Discharge Machining (EDM) from the 2 mm thick rolled and flash annealed metallic sheets, with the gauge direction parallel and perpendicular to the rolling direction (denoted as RD and TD, respectively). Tensile tests were performed at constant true ε& (TSR), equal to 10-1 and 10-2 s-1 and at T, between 400 and 450°C in air using a computer controlled 810 Material Testing System (MTS) equipped with a three-zone split furnace. Each specimen was placed between two stainless steel grips and fastened with two plates, the furnace was closed and the target temperature was set in the computer control of the facility, so as to achieve the fastest heating rate possible. The target temperature was reached in approximately 15 minutes and the test commenced after a very short stabilization period of about 90 seconds. The short heating cycle was chosen so as to approximate, as much as possible, the conditions of blow forming in the industry and to restrict static recrystallization. During the test the temperature variation in the three zones was less than ± 3°C. Upon failure the furnace

71

Chapter 4

was immediately opened and the specimens were quenched by water spray prior to removal. The temperature decrease during quenching was larger than 50°C sec-1. The strain distributions over the gauge length were determined from optical photographs (grayscale pixel uncompressed bitmaps) from the face and the profile of each side of the failed specimens, using the GOM – ARAMIS system. The images were combined, aligned, converted into black and white and processed by a MATLAB (Ver. R2007b) program which enabled the calculation of the area reduction per pixel along the extended gauge. Subsequently, the specimens were sectioned and mechanically polished according to the Struers method with diamond suspensions of decreasing grade sizes and finished with a colloidal alumina suspension, OPU, having a particle size of 30nm. For the investigations of microstructure, EBSD was carried out in an XL-30 Scanning Electron Microscope equipped with a Field Emission Gun (FEG-SEM) and a fluorescent screen operating at 20 kV. The Orientation Imaging (OIM) data were collected and analyzed using the TSL application developed by AMETEC Inc. – EDAX. The step size was selected to be 1 µm so as to compromise between examining appreciably large areas covering very often the entire thickness of the deformed gauge and being able to collect more than 40 data points from the majority of the grains.

4.2 Results All specimens, irrespective of their orientation with respect to the rolling direction, exhibited their maximum elongation to failure at 425°C and at 10-2 s-1. Figure 4.1 shows representative true stress – engineering strain curves of the specimens RD and TD extended at 425°C at 10-1 and 10-2 s-1 (true strain rate TSR). Past the elastic regime, all

72

Deformation Processes

curves exhibited sharp peaks prior to the establishment of steady state flow, indicative of a low initial density of mobile dislocations. They demonstrate the need for dislocation multiplication processes because, at the onset of plasticity, most dislocations are pinned by solutes (i.e. Mg) or impurity atmospheres, as in the case of low carbon steels, where strain ageing is demonstrated by a subsequent softening after sharp peaks at the onset of plastic deformation. This subsequent softening is due to the increase of the mobile dislocation density and/or because the dislocations have managed to break away from their pinning atmosphere [5]. Past the softening the flow stress, σ f gradually increases (i.e. presumably during a stage of work hardening) up to approximately 35% and 50% of nominal elongation for the specimens deformed at 10-1 and 10-2 s-1, respectively. The values of σ f were taken from each tests, as the average between 5 and 10% of engineering ε and were equal to 60.25 ± 0.26 MPa (RD) and 59.33 ± 1.43 MPa (TD), at 10-1 s-1 and 35.13 ± 0.78 MPa (RD) and 35.35 ± 1.19 (TD), at 10-2 s-1. Contrary to what is observed in Figure 4.1, for four out of several specimens, there is practically no difference in the shape of the curves between the RD and TD specimens up to 10% of maximum elongation, as well as up to the maximum σ f value at the end of the work hardening stage. The only notable difference is, perhaps, the larger uncertainty in the σ f for the TD specimens compared to that of the RD specimens. Past the work hardening stage, σ f decreases with increasing ε , indicating the development of diffuse necking. Again, no difference could be discerned in the slope of this decrease between different specimens (RD, as well as TD), as the curves could be placed practically on top of one another (i.e. they were completely reproducible up to this point). Significant differences,

73

Chapter 4

70

o

60

True Stress / MPa

-1 -1

RD TSR 425 C, 10 s o -1 -1 TD TSR 425 C, 10 s o -2 -1 RD TSR 425 C, 10 s o -2 -1 TD TSR 425 C, 10 s

50 40 30 20 10 0

0

50

100

150

200

250

300

350

400

Engineering Strain (%) Figure 4.1: True stress – Engineering strain curves for the specimens extended along the rolling direction (RD) and perpendicular to the rolling direction (TD).

however, were observed at ε where σ f started to decrease rapidly, i.e. at the onset of local necking, just prior to the specimen’s ultimate failure. This resulted in significant differences in the values of the maximum elongation to failure between specimens with the same orientation, deformed at the same conditions. Their average values were approximately equal to 238% (RD) and 248% (TD), for the specimens deformed at 10-1 s1

and 360% (RD) and 297% (TD), for the specimens deformed at 10-2 s-1. The curves in

Figure 4.1, correspond to specimens that showed elongation to failure as close as possible to the average values, i.e. 242% (RD) and 254% (TD), at 10-1 s-1 and 367% (RD) and 308% (TD), at 10-2 s-1. Detailed description of the mechanical properties, the anisotropy the pre and post deformation dislocation microstructures are reported in the previous chapter [6].

74

Deformation Processes

(a)

(d)

(b)

(e) 200

10 9 8 7 6 5 4 3 2 1 0

10 9 8 7 6 5 4 3 2 1 0

εL

400

100

εL

200

10

20

30

40

50

60

70

length along the gauge (mm)

80

90

10

20

30

40

50

60

70

80

length along the gauge / mm

(c) (f) Figure 4.2. The RD and TD specimens, (a, d) before and (b, e) after the tensile tests at 425°C and at 10-2 s-1 and the distribution of ε L along their gauge, (c) and (f). Figure 4.2 shows the specimens RD and TD before (Figure 4.2a and 4.2d) and after (Figure 4.2b and 4.2e) the tensile deformation at 425°C and at 10-2 s-1, as well as their corresponding ε L distribution across their gauges (Figure 4.2c and 4.2f). Knowing the area reduction at each point across the gauge, i.e. the ratio A0/A, where A0 and A are the area of the cross-section before and after the deformation, respectively, ε at each position across the gauge (i.e. the ε L ) was calculated by employing the volume conservation of the specimen (i.e. the fact that A0 × l 0 = A × l , where l0 and l are the lengths of the gauge before and after deformation) and using the formula ε L = ( A0 / A) − 1 .

75

Chapter 4

Evidence of localized secondary necking (secondary necking instabilities) was frequently observed in specimens after deformation. These instabilities were associated with specimens that often showed unusually large maximum elongations but their development was completely unpredictable. A secondary necking instability was observed in the gauge of the specimen RD at 44 mm (Figure 4.2(b)). In the diagram of Figure 4.2(c), it is observed at approximately 53 mm (at ε L ≈ 535%), because sections of the specimen grip closer to the gauge (on the right of the specimen hole at the left side and on the left of the specimen hole at the right side) were also included in the calculation of the ε L so as to evaluate, if possible, the flow of material from the grips. These instabilities were rationalized as evidence of quasi-superplastic flow [7] and were observed even in specimens of aluminum alloys with sub-micrometer grain size produced by equal-channel angular pressing (ECAP), after extension at 400°C and at the same ε& [8]. Selecting appropriate areas for EBSD analysis is crucial in obtaining a complete profile of the evolution of the properties of the microstructure as a function of ε L . Large sections of the gauge, for instance, especially close to the grips (i.e. between 20 and 40 mm in Figure 4.2(c)) had minimal variation in ε L (between 180 and 230%), whereas the section between 40 and 50 mm presented areas where ε L varied from 260 up to 500%. The selection of the appropriate areas and their careful preparation having good EBSD signals was quite laborious, so as to make sure that the uncertainty in the values of ε L for the areas chosen was less than ±2.5%.

76

Deformation Processes

4.3 Grain Size and Texture Evolution with Local Strain. Figure 4.3 shows the evolution of the average grain size (Figure 4.3a and 4.3d) and the volume fractions of the recrystallization texture components as a function of ε L (Figure 4.3b&c and Figure 4.3e&f), for the specimens RD and TD, respectively. These components correspond to the Cube (001)[100], Goss (011)[100], P (011)[ 1 1 1 ], Q (013)[ 3 3 1 ] and R (132)[ 421 ] texture components. The Cube texture means, for example, the grains have preferential orientations with (001) plane parallel to the rolling plane and [100] direction parallel to the rolling direction of the material. The data from the asreceived material are plotted at ε L of 50%, and that from the grip section at ε L of 0%. At the grip, the grain size was 20.4µm for the RD specimen, which was quite close to the asreceived value of 20.7µm. In the TD specimen, however, the grain size was increased to the average value of 24.4 µm. The short heating cycle, therefore, seems to have no effect on increasing the grain size in the RD specimen. The modest increase in the average grain size in the TD specimen, however, indicated that static recrystallization did indeed take place but it was moderate. In the RD specimen, deformation of up to 65% of ε L leads to grain growth to an average value of 26.5 µm. Subsequently, with increasing ε L and up to 400%, the grain size fluctuated between 29.8 and 26.5 µm. At ε L above 400%, refinement decreased the grain size down to 14.4µm (at ε L equal to 1415%), before the significant amount of dynamic recrystallization taking place close to the failure point, which increased the average grain size to 43 µm. In the TD specimen, from the 24.4 µm at the grip the average grain size increased to 27.1 µm at 55% ε L , decreased slightly to 25.8 µm at 90%

77

Chapter 4 50

o

50

-2 -1

RD, 425 C, 10 s

45 40 35 30 25 20

35 30 25 20

15

15

10

0

100

200

300

400

500

7000

10

0

200

400

εL (%)

(a) 40

Cube

Goss

30 25 20 15 10 5 0

100

200

1000

7000

300

400

500

Goss

30 25 20 15 10 5 0

0

800

Cube

35

Volume Fraction (%)

Volume Fraction (%)

35

600 ε L (%)

(d)

40

7000

0

200

400

600

800

1000

7000

εL (%)

εL (%)

(b)

(e) 10

10

P

8

Q

P

R Volume Fraction (%)

Volume Fraction (%)

-2 -1

40

Grain Size (µm)

Grain Size (µm)

o

TD, 425 C, 10 s

45

6 4 2 0

Q

R

8 6 4 2 0

0

100

200

300

ε L (%)

(c)

400

500

7000

0

200

400

600

800

1000

7000

εL (%)

(f)

Figure 4.3. Microstructure evolution as a function of ε L for specimens RD and TD: (a) and (d) the average grain size, (b, e) the volume fractions of the Cube and Goss RX texture components and (c, f) the volume fractions of the P, Q and R RX texture components.

78

Deformation Processes ε L before growing to an unusually high value of 39.5 µm at 120%. From this high value

it decreased, suddenly at first and then continuously with some minimal variation, down to 17.6 µm at 1015%. It increased to 45.2µm at 2400% and finally reached an average value of 29.8 µm at 6830%. Both values, at this very high ε L regime, are average grain size values from regions that are located at the onset of the local necking and close to the failure point, respectively, and consequently, it is safe to assume that dynamic recrystallization (DRX) did indeed take place at this area, prior to failure. This is confirmed also by the relative increase of the R (for the RD specimen), as well as the R and P (for the TD specimens) texture components, which together with the significant decrease of the Cube and Goss components, are indicative of a major texture change (i.e. the texture becoming more random) in the necking area. For both specimens the texture develops from that characterizing a weak recrystallized microstructure in the as-received state (where all RX components have a volume fraction between 5 and 8%) to that being dominated by the Cube and Goss components suggesting perhaps a combination of deformation and recrystallization processes. The dominance of these components becomes pronounced above 100% ε L , for the RD specimen and much sooner (i.e. above 55% ε L ) for the TD specimen, whereas their volume fractions increase above 10%, reaching and exceeding 30% (for the Cube component in the TD specimen). The only observed difference between the RD and the TD specimen is the clear supremacy of the Cube over the Goss texture in all sections of the gauge for the latter, apart from the regions at the local necking area where major texture changes occurred. On the other hand, the RD specimen showed, in general, stronger Goss component with volume fractions closer to 20%. Finally, almost the entire gauge of both specimens shows

79

Chapter 4

insignificant fractions of the P, Q and R components (i.e. their volume fraction decreased from a value between 6 – 7% at the grip, down to 2% for the RD specimen and less than that for the TD specimen). Other common components like those characterizing rolling textures, i.e. Goss, Copper, S, D and Brass, as well as those indicating a fibrous microstructure (i.e. having α−, β−, γ−, or τ−fiber texture) were also investigated. At the failure point of the RD

specimen, the S texture reached a volume fraction of 10% and the brass texture a volume fraction of 4%, whereas at the failure point of the TD specimen the α- fiber showed a volume fraction of 5% the γ- fiber 10% and the S texture an 8% volume fraction and they were comparable with those of the as-received alloy. Within the larger ε L regime, however, most of the above mentioned components showed volume fractions less than 2%. Apart from the Goss component which is also taken as RX texture component and the τ−fiber which is dominated by the Cube and Goss components, the rest not only have volume fractions comparable and/or lower than the P, Q and R components, but they also exhibited lower volume fraction change with increasing ε L . Consequently, only the Cube and Goss components will be considered henceforth.

4.4 Deformation, recovery, and recrystallization. Figure 4.4, shows the Inverse Pole Figure (Auto IPF) from a microstructure obtained from the RD specimen at 200% ε L . At each value of ε L images such as these were collected and analyzed by calculating two important parameters, the Grain Orientation Spread (GOS) and the Grain Average Misorientation (GAM). The former is defined as the average deviation between the orientation of each point in the grain from

80

Deformation Processes

the average orientation of the grain, whereas the latter is the average misorientation angle between all neighboring pairs of points inside the grain. Since each dislocation introduces a rotation in the lattice, severely deformed grains with high dislocation densities will demonstrate high GOS values. Recovery only partially reduces the dislocation density, since it comprises all the annealing processes occurring within deformed grains without the migration high-angle grain boundaries [9]. Grains demonstrating recovery, therefore, will have lower GOS values than the deformed grains but will exhibit significantly high values of GAM. Finally, grains that have experienced recrystallization will exhibit low GOS, as well as low GAM values. It has been reported that grains with internal misorientation larger than 2° (i.e. having GOS > 2°) could be considered as deformed grains [10]. Among the grains with GOS ≤ 2°, those with GAM larger than the average GAM (GAMAVR) were considered as those undergone recovery. Finally, the grains with GOS ≤ 2°, as well as GAM ≤ GAMAVR were considered as recrystallized (RX) grains. In summary each OIM map obtained from each area with a specific ε L was partitioned in grains according to the criteria shown below: I)

Deformed grains: GOS >2°.

II)

Recovered grains: GOS ≤2° and GAM > GAMAVR.

III)

Recrystallized grains: GOS ≤2° and GAM≤GAMAVR. Figure 4.5, shows the above mentioned analysis performed on the OIM map of

Figure 4.4. The map containing all grain data was partitioned in deformed (Fig. 4.5a,b), recovered (RV) (Fig. 4.5c,d) and recrystallized grains (Fig. 4.5e,f). The deformed grains have many sub-grain boundaries (i.e. blue colored) and are separated by several lowangle grain boundaries (i.e. green colored) and a couple high-angle grain boundaries (i.e.

81

Chapter 4

red colored). On the other hand the recrystallized grains are separated mostly by highangle grain boundaries. There are

Figure 4.4. Auto IPF from the RD specimen with ε L equal to 200%.

still a few low-angle grain boundaries and sub-grain boundaries but they are observed mostly adjacent to the high-angle grain boundaries as if they are about to be absorbed by their migration. Finally, the recovered (RV) grains exhibit still sub-grain boundaries (most of them located in their middle) but also well and distinctly shaped low-angle grain boundaries and high-angle grain boundaries. The grayscale depicting the grain surface is a measure of the Average Image Quality (AIQ) of the pattern. The lighter the grains appear, the less the distortions and the higher the average image quality. It is obvious that the deformed grains appear darker because their microstructure is highly distorted (due to dislocations). Some of the grains in the recovered partition appear lighter whereas in the recrystallized partition, nearly all of them show very high values of average image quality demonstrating a microstructure with a comparatively very low dislocation density. There are, however, two grains in the recovered (RV) inset (Figure 4.5(d)), that appear almost black indicating an unusually high amount of distortions (denoted by “BG” for black grains in the middle and top left). Some small dark non-indexed areas can be seen located primarily at the grain 82

Deformation Processes

Figure 4.5. (a) The deformed, (c), recovered, (e) and recrystallized grain partitions (d). The insets (b), (d) and (f) show the high-angle grain boundaries, low-angle grain boundaries and sub-grain boundaries in the grain microstructure of the deformed, recovered and recrystallized partitions, respectively. The lighter grey grains have better average image quality i.e. less distortions.

83

Chapter 4

boundaries which appear smaller and more numerous between the deformed grains (Figure 4.5(b)) slightly fewer but larger between recovered grains (Figure 4.5(d)) and are absent in the recrystallized grains (the large black inclusion at the top of the image is actually a small recovered grain). These non-indexed areas correspond to precipitates, they appear as if they pin the newly developed sub-grain boundaries and/or low-angle grain boundaries between deformed grains (Figure 4.5(b)) and are mostly located at the high-angle grain boundaries between recovered grains (Figure 4.5(d)). Figure 4.6 shows the grain size evolution of the deformed, recovered and recrystallized grains as a function of ε L for the RD (Figure 4.6(a)) and the TD specimen (Figure 4.6(b)). Apart from the region at the failure point, both specimens exhibit the same behaviour, i.e. the size of the deformed grains is larger than that of the recovered grains and that of the latter is larger than the size of the recrystallized grains. Furthermore, the size difference between the deformed and recovered grains is much larger than that between the recovered and recrystallized grains. It is clearly evident, however, that above 200% ε L and up to 1015%, all types of grains in the TD specimen are smaller than those in the RD specimen and in this ε L regime the grain size of the former decreases continuously with small fluctuations. At the failure point the grain size of the recovered grains seems to exceed that of the deformed grains in the RD specimen, whereas in the TD specimen at ε L equal to 2400%, the recrystallized grains were the largest and the recovered grains, despite being smaller, were substantially larger than the deformed grains. At the failure point (i.e. at ε L equal to 6830%) the recovered grains where slightly larger than the recrystallized ones and both were much larger than the deformed grains. In summary, close to the failure point the recovered grains become the largest, whereas in

84

Deformation Processes

the TD specimen, the recrystallized grains are much larger at the onset of local necking

60 55 50 45 40 35 30 25 20 15 10 5

Deformed

RV

RX Grain size (µm)

Grain size (µm)

and slightly smaller than the recovered grains at the point of failure.

0

100

200

300

400

500

7000

60 55 50 45 40 35 30 25 20 15 10 5

Deformed

0

200

RV

400

600

800

RX

1000

7000

ε L (%)

ε L (%)

(a) (b) Figure 4.6. Grain size evolution of the deformed, recovered (RV), and recrystallized (RX) grains in the RD (a) and in the TD specimen, (b). 0.35

0.35

SGBs

HAGBs -1

0.25 0.20 0.15 0.10 0.05 0.00

100

200

300

εL (%)

(a)

400

500

7000

LAGBs

HAGBs

0.25 0.20 0.15 0.10 0.05 0.00

0

SGBs

0.30

GB Density/ ( µm )

-1

GB Density/ (µm )

0.30

LAGBs

0

200

400

600

ε L (%)

800

1000

7000

(b)

Figure 4.7. Grain boundary density variations as a function of ε L in the RD (a) and TD (b) specimens. In this chapter, different deformed areas but only of one specimen was characterized, thereby no error bars are measured. But the uncertainty value can be estimated at the level of within 5% for this figure and the figures hereafter.

A comparison between Figure 4.6a and 4.6b with those of the average grain size in Figure 4.3a and 4.3d show, furthermore, that the variations with ε L , observed for the deformed grains resemble more those of the average grain size. The resemblance of the variations of the recovered grains with that of the average grain size is considerably lower 85

Chapter 4

and that of the recrystallized even lower indicating that the size of the deformed grains is actually the largest in the microstructure, i.e. deformation dominates in the largest grains at all ε whereas at the local necking RX leads to large grain growth and recovered apparently dominates at the larger grains at the failure point. Figure 4.7a and 4.7b show the variations of the density of the sub-grain boundaries, low-angle grain boundaries and high-angle grain boundaries with increasing ε L . A decrease in the sub-grain boundary and high-angle grain boundary density is

associated with grain growth, whereas the opposite behaviour characterizes grain refinement. It is evident therefore that the fluctuations observed in the average grain size in the RD specimen will be reflected in the variations of the grain boundary density with ε L . Indeed, the average grain size increases from 0% and 105% ε L is reflected by local minima in the variations of the grain boundary density. The continuously increasing trend in the grain boundary density in the TD specimen from 145 up to 1015% ε L is indicative of the slow but continuous grain refinement in the microstructure. The resemblance of the variations between the sub-grain-boudnaries and high-angle grain boundaries in the RD specimen and up to ε L equal to 1415% is remarkably obvious. At the failure point (RD specimen) the density of the sub-grain boundary and high-angle grain boundaries still behave the same, but the much sudden increase of the latter compared with that the former suggests the occurrence of extensive dynamic recrystallization processes. The density of low-angle grain boundaries with increasing ε L remains constant up to 100% and then slightly increases up to 450% ε L . It increases suddenly at 1415%, where extreme grain refinement occurs and then decreases at the failure point where they were converted into Hags. The resemblance of the variations of

86

Deformation Processes

the sub-angle grain boundary and high-angle grain boundary density with ε L in the TD specimen become apparent only above 350%. The sudden drop in the high-angle grain boundary density at ε L equal to 120% which is associated with the large growth of predominantly the deformed grains is subsequently followed by a steep and almost linear increase of the SBS from 120 up to 250% ε L . Their density, however, decreases between 250 and 350% without any notable increase in the high-angle grain boundary and/or lowangle grain boundary density. This may be perhaps indicative of a small fraction of the large deformed grains becoming adjusted to more favourable deformation orientations by a process that involves extensive sub-grain formation and their subsequent coalescence along these orientations. A similar feature was observed at the early stages of deformation in few grains of polycrystalline 1050 aluminum alloy [11]. With increasing ε , sub-grain boundaries inside some grains disappeared without motion of the high-angle

grain boundaries and without the formation of low-angle grain boundaries. It was considered that this occurred because the distorted lattice remaining after the annealing of the sub-grain boundaries was rearranged by lattice rotation due to the primary slip operating in each sub-grain. In our case, this may characterize large portions of the grain microstructure since we have extended the specimen in large elongations. This, however, needs to be investigated further also by examination of the microstructure of TD specimens strained up to 150%. For both specimens, however, the apparent resemblance of the variation of the sub-grain boundary density with that of the high-angle grain boundaries over a large range of ε L values, seems to indicate that the formation rate of new sub-grain boundaries by the continuous accumulation of dislocation arrangements is comparable with that of the formation of new high-angle grain boundaries by the increase

87

Chapter 4

in the misorientation of low-angle grain boundaries. Since the sub-grain boundaries are most frequently absorbed and/or annihilated by the high-angle grain boundary migration, the density of the latter ends up being much higher than that of the former. Figure 4.8, shows the variation of the deformed, recovered and recrystallized grain volume fractions as functions of the ε L , for both the RD (Figure 4.8(a)) and the TD (Figure 4.8(b)), specimen. For both specimens the volume fraction of the small recrystallized grains is over 60% and thus larger than that of the recovered grains and considerably larger than that of the large deformed grains. For comparison the asreceived microstructures showed a recrystallized grain volume fraction exceeding that of the recovered grains only by approximately 20%. For the RD specimen, the variation of the recrystallized grain volume fraction with ε L appears to be complementary that of the recovered grain volume fraction. That is, with increasing ε L when the volume fraction of the recovered grains decreases, that of the recrystallized grains increases and microstructures with high volume fraction of recovered have relatively low volume fraction of recrystallized grains. The large deformed grains, on the other hand do not demonstrate a large variations in their volume fraction apart from a moderate increase at ε L between 200 and 400% (in the RD specimen). The same trend is also observed for the

TD specimen with one major exception, i.e. the variation of the grain volume fraction between 90 and 320% ε L . In this regime, the variation of the recrystallized grain volume fraction seems to be complementary to that of the deformed grains, whereas the volume fraction of the recovered grains decreases, slightly between 90 and 120% ε L , suddenly increases to 25% at 145% ε L and then decreases slightly at the beginning and more drastically at the end. This decrease in the volume fraction of the recovered grains,

88

Deformation Processes

between 145 and 320% ε L combined with a significant increase in the sub-grain boundaries density at the onset of this process and the decrease in the size of the recovered grains (Figure 4.7(b)) may be a consequence of the rapid reconstruction of the large deformed, recovered and recrystallized grains observed at 120% ε L and their alignment along favorable deformation orientations (i.e. having more slip systems available to carry out the deformation). Past 320% ε L , the relatively low volume fractions of recrystallized grains (with the simultaneous high volume fractions of the recovered grains) and the increase of all types of grain boundaries until the region of local necking is reached, indicates that TD specimens undergo continuous and gradual grain refinement (see Figure 4.6b and 4.7b). This refinement gives way to intense dynamic recrystallization upon the first stages of the development of local necking, leading to its rapid development and the premature failure of the specimen. Indeed, statistically, specimens having the TD geometry exhibited lower maximum elongations prior to failure compared with those having the RD geometry [6]. 100

Deformed

RV

RX

80 70 60 40 30 20 10 0

100

Grain Volume fraction (%)

Grain Volume Fraction (%)

90

100

200

300

εL (%)

400

500

7000

RV

RX

80 70 60 40 30 20 10 0

0

Deformed

90

0

200

400

600

800

1000

7000

εL (%)

(b) (a) Figure 4.8. Grain fractions in the RD (a) and the TD (b) specimen of the deformed, recovered, (RV) and recrystallized grains, (RX).

89

Chapter 4

Figure 4.9, shows the variation of the volume fractions of the Cube (4.9a and c) and Goss (4.9b and d) texture components with ε L for the deformed, recovered and recrystallized grains, of the RD and TD specimen, respectively. For both specimens, in the deformed, recovered and recrystallized grains, Cube is the most pronounced texture with volume fractions often exceeding 30%. For the RD specimen the volume fraction of the Cube component is larger in the recrystallized grains up to 90% ε L . From this value and up to 535% ε L the deformed grains retain the largest volume fraction of the Cube component. From 90% and up to 400% ε L between the recrystallized and the recovered grains, the Cube texture is more prominent in the former compared to the latter, whereas at ε L between 450 and 535% ε L , the situation is reversed, as the volume faction of recovered grains that demonstrate the Cube orientation is slightly higher than that of the recrystallized grains. Immediately afterwards, the Cube texture reaches a minimum upon extreme grain refinement, before a slight increase up to 5% on all grains which characterizes the more random texture at the failure point. For the TD specimen, the Cube texture dominates in the larger deformed grains at all values above 55% ε L ; it increases showing severe fluctuations up to 440% ε L and then decreases along with the grain refinement. At the region of the failure point the Cube component reaches a volume fraction of 5% as the texture becomes more random. Between the recovered and recrystallized grains, the Cube volume fraction in the former exceeds slightly that of the latter especially above 250% ε L . Despite significant fluctuations the Cube volume fraction in both the recovered and the recrystallized grains remains between 17 and 27% up to 1015% and then decreases down to a 5% volume fraction at the failure point, initially more, and finally less rapidly and than that observed for the deformed grains.

90

Deformation Processes

The variation of the Goss texture is considerably different from the Cube texture. In the RD specimen and for the deformed and recovered grains, it increases with significant fluctuations up to 200% ε L and keeps almost constant until ε L 400%. For the recrystallized grains, however, it increases (showing still fluctuations) almost

50

Deformed

45

RV

Cube Texture Volume Fraction (%)

Cube Texture Volume Fraction (%)

continuously until 400% ε L . RX

40 35 30 25 20 15 10 5 0 0

100

200

300

400

500

50

Deformed

45 35 30 25 20 15 10 5 0

7000

0

200

400

RV

RX

40 35 30 25 20 15 10 5 0 0

100

200

300

εL (%)

400

500

7000

Goss Texture Volume Fraction (%)

Goss Texture Volume Fraction (%)

Deformed

600

εL (%)

800

1000

7000

(c)

(a) 45

RX

40

εL (%) 50

RV

50

Deformed

45

RV

RX

40 35 30 25 20 15 10 5 0 0

200

400

600

800

1000

7000

εL (%)

(b) (d) Figure 4.9. (a, c) Cube and (b, d) Goss texture volume fractions in the deformed, recovered (RV) and recrystallized (RX) grains of the RD and TD specimens, respectively. At ε L above 400% the volume fraction of the Goss texture slightly increases mostly in the deformed grains before decreasing to a minimum at 1415% where the microstructure exhibits significant grain refinement. At the point of failure, however, a significant

91

Chapter 4 volume fraction of Goss texture seems to characterize many deformed grains (with a volume fraction 25%). The total volume fraction of the deformed grains in failure point, however, is essentially insignificant since most grains demonstrate recrystallization. In the TD specimen, simultaneously in the deformed, recovered, and grains, the Goss texture increases with severe fluctuations up to 250% ε L and from then stays within volume fractions varying between 15 and 30%. It demonstrates a slight increasing trend, despite the gradual grain refinement. At 2400% ε L , i.e. at the onset of the local necking, it decreased and, finally, reached a minimum at the failure point for the deformed and the recovered grains. In this specimen, many recrystallized grains, at the failure point seem to retain a Goss texture with a volume fraction approximately equal to 10%. The large volume fraction of these grains suggests that, at least in this specimen, many grains seem to favour a Goss orientation. It is worth noting finally, that the Goss component increased in the deformed grains prior to extreme grain refinement (in the RD specimen) and before the onset of local necking (in the TD specimen).

4.5 Discussion The superplastic deformation of coarse grained AA5182 aluminum alloys exhibits considerable differences in the maximum elongation to failure when they are extended at the 425°C and at 10-2 s-1 along and perpendicular to the rolling direction. These variations can be explained in terms of the microstructure features presented in our current investigation, i.e. grain size evolution, grain boundary character and density and texture development.

92

Deformation Processes Influence of grain size and texture on failure mechanism All specimens failed by the development of local necking. As mentioned previously, the specimens having TD geometry failed systematically much sooner than those having RD geometry. The difference in the grain size variation as a function of ε L between the RD and TD specimen provides some evidence for this behaviour. The RD specimen exhibited fluctuations in the grain size over a large range of ε L . The grain size was more or less stable (fluctuating between 22.5 and 29.8 µm) within most of the gauge length including the region of the secondary instability (at ε L ≈ 535%). Only at the very at 1415% ε L close to the failure point did the grain size decrease dramatically. In many cases, despite the appearance of secondary instabilities, the grain size increased again reaching similar values, past the instability, as the failure point was approached [6]. Here this behavior cannot be depicted as in this region, i.e. past the necking instability, ε L had the same value as before the development of the secondary instability. On the other hand, the TD specimen showed not only lower values above 150% ε L , the average grain size exhibited a continuous decrease with practically no variations. Consequently, the region where the microstructure can no longer resist the occurrence of severe dynamic recrystallization is reached most likely sooner for these specimens. Furthermore, dynamic recrystallization does not occur only at the failure point but at the onset of local necking i.e. it may well be the cause for the necking development. Thus continuous grain refinement does not result in large elongations. Maintaining a stable grain size, despite the small variations, for a large regime of ε L seems to be more critical in obtaining large elongations prior to failure.

93

Chapter 4 Throughout the entire ε L regime the Cube texture was stronger for the TD specimen, whereas the Goss component was weaker than for the RD specimen. It is not presumptuous, therefore, to assume that a microstructure with more evenly balanced Cube and Goss components results in larger elongations prior to failure. Prior to the region where grain refinement was observed at the RD specimen the volume fraction of the Goss component of the deformed grains exceeded 25%. In the TD specimen the deformed grains exhibited similar volume fractions prior to the onset of local necking. Grains of material with fcc lattice which possess Cube and Goss orientation have both eight active slip systems. The lattice of the grains with Goss orientation, however, is rotated by 45° around an axis parallel to the rolling direction with respect to that of grains with Cube orientation. Thus, grain coherency, may probably favour the rotation of deformed grains into alternating Cube and Goss orientations, so as to allow for the operation of multiple slip systems across the grain boundaries at large values of ε L . If that is true, then an evenly balanced Cube and Goss texture is prone to produce larger elongations prior to failure. At the onset of the local necking and at the failure points, large grain growth is accompanied with the RX textures decreasing to volume fractions between 4 and 8% (i.e. Cube, Goss and R, for the RD specimen and Cube, Goss, P and R for the TD specimen). The calculation of the rolling and fiber texture components showed that in the RD specimen the S texture reached a volume fraction of 10% and the brass texture a volume fraction 4% at the failure point, whereas for the TD specimen the α- fiber exhibited a volume fraction of 5%, the γ- fiber 10% and the S texture an 8% volume fraction. This extensive grain growth, presumably due to long range motions of RX fronts, as well as

94

Deformation Processes the texture randomization (weakening) that occurred due to the minority texture components acting most likely as the source of new nuclei, is characteristic of discontinuous dynamic recrystallization (DDRX). The microstructure consists of recrystallized grains with a volume fraction exceeding 80%, large densities of high-angle grain boundaries, average grain size of 40 and 30 µm for the RD and TD specimens, respectively and a small volume fraction of recovered grains.

Extended Recovery Based on the features observed in the grain partitions, the deformed grains contained a large density of dislocations (poor average image quality). Randomly placed sub-grain boundaries could be discerned in their microstructure but their most remarkable feature was that their continuous grain boundaries (GBs) were essentially alternating lines of low-angle grain boundaries and sub-grain boundaries. Only a few high-angle grain boundaries were observed. Deformation therefore is envisaged as a process where random

dislocations

rearrange

themselves

into

sub-grain

boundaries

whose

misorientation continuously increases and where the sub-grain boundaries are converted into low-angle grain boundaries. This process, however, appears to be quite slow. The main feature of the deformed grains is therefore their high dislocation density. They have the larger grain size and the smallest volume fraction. In terms of dislocation density, the recovered grains appear to have much larger differences. Some were observed almost black suggesting unusually high dislocation content, whereas most appeared to be significantly less distorted than those of the deformed grains i.e. they had higher average image quality values. The sub-grain boundary density does not appear to change notably, but these sub-grain boundaries

95

Chapter 4 appear significantly longer and very often span the entire width in the middle of the grains. The low-angle grain boundary density, however, appears to be significantly lower, whereas the high-angle grain boundary density is higher. What is very remarkable in the grain boundary structure of those grains is that a few of these boundaries present themselves as long low-angle grain boundaries, which become high-angle grain boundaries especially when they reach a triple junction (see at the lower right side of Figure 4.5 (d)) or when they intersect an unindexed region (i.e. a precipitate). The few precipitates that were observed, were quite fine, more numerous in the deformed grains and fewer and larger in the recovered grains. They seem to facilitate pinning of the newly formed sub-grain boundaries and low-angle grain boundaries in the deformed microstructure and are observed to intersect many high-angle grain boundaries in the recovered microstructure. The recovered grains have significantly higher volume fraction and a grain size much smaller than that of the deformed grains. The grains with higher volume fraction are produced after recrystallization. They have the smaller size and the highest average image quality, i.e. the lower amount of distortions due to dislocations. They are separated mostly by high-angle grain boundaries. Well developed low-angle grain boundaries and a few sub-grain boundaries could be distinguished in their microstructure but nearly all of them were observed close to highangle grain boundaries. This suggests that if the recrystallization process in the microstructure progressed a little further, the migration of the high-angle grain boundaries would, most likely consume most of the sub-grain boundaries and/or lowangle grain boundaries, or the misorientation of the latter would most likely increase and thus they would be converted into high-angle grain boundaries.

96

Deformation Processes In summary, the separation of the EBSD patterns into partitions, according to the separate processes responsible for the evolution of the microstructure provides a quite clear picture for the overall mechanism which leads homogeneously to grain refinement. This transformation has been termed as “extended recovery”, continuous recrystallization, or continuous dynamic recrystallization (CDRX) [12,13] and based on the current description is clearly a dislocation controlled mechanism. In large highly distorted grains having a relatively low volume fraction, the dislocations arrange themselves so as to produce random sub-grain boundaries. The continuous accumulation of dislocations into some sub-grain boundaries gradually increases their misorientation converting them to low-angle grain boundaries. This stage, which ends with a microstructure having a high density of sub-grain boundaries and low-angle grain boundaries, could be envisaged as a relatively slow process, since in grains with homogenous distribution of dislocations their absorption by pre-existing boundaries and/or their mutual annihilation is quite frequent. Thus only a small percentage of dislocations will be most likely arranged so as to produce sub-grain boundaries and/or low-angle grain boundaries, preferably close to precipitates which pin and/or arrest the dislocation motion. This is supported by the fact that the most distorted grains had significantly fewer sub-grain boundaries in their microstructure, whereas those with many sub-grain boundaries had a much better average image quality and were bordering precipitates. Even though the density of the sub-grain boundaries does not change notably, the density of the high-angle grain boundaries increases, while that of the low-angle grain boundaries apparently decreases in the areas that show recovery. One possible mechanism for this increase in the high-angle grain boundary density could be produced

97

Chapter 4 by the merging of low-angle grain boundaries during sub-grain coalescence [14,15]. The almost marginal increase of the low-angle grain boundary density with ε L seems to suggest that indeed their merging during sub-grain coalescence towards increasing the high-angle grain boundary density may be a probable process. Furthermore, in special cases where the grain size increases and/or decreases rapidly this process may be favored (e.g. the complementary variations of the deformed and the recrystallized grain volume fractions in the TD specimen and at ε L between 90 and 250%) but results apparently in refined grains with recrystallization character (i.e. a low amount of distortion). Finally the presence of highly distorted grains and their complete absence in the recrystallization partition may indicate that sub-grain coalescence is a relatively fast process which occurs widely in grains during recovery. The end product of this microstructural transformation is a high volume fraction of small grains with a comparatively very low dislocation density, which are separated mostly by high-angle grain boundaries. During that stage extensive migration of selected high-angle grain boundaries (i.e. those with the highest misorientation) absorbs some of the sub-grain boundaries. It is interesting to see that the size difference of the recovered and the recrystallized grains is much smaller than that between the deformed with the recovered grains. Furthermore, the increase in the volume fraction of the recrystallized grains is synonymous with a decrease of the volume fraction of the recovered grains. It is conceivable, therefore, to regard recovery and recrystallization as successive processes where the interval of the transition of the microstructure from the former to the latter is relatively brief. The apparent resemblance of the variation of the sub-angle grain

98

Deformation Processes boundary density with that of the high-angle grain boundaries over a large range of ε L seems to indicate that the rate of formation of new sub-grain boundaries within the large deformed grains with increasing ε L , is comparable with the formation of new high-angle grain boundaries that separate the small recrystallized grains. Due to the sub-angle grain boundary absorption by the high-angle grain boundary migration during recrystallization the sub-grain boundaries density is substantially lower than that of the high-angle grain boundaries. The low values of the low-angle grain boundary density and their slow increase with increasing ε L demonstrates that its formation by increasing sub-angle grain boundary misorientation is comparable with their conversion to high-angle grain boundaries. Upon grain refinement, their density continuously increases. The recovery processes outlined above result in a fine-grained microstructure with reduced lattice strain, since the dislocation structures within the grain interiors are gradually eliminated. This results in texture sharpening, especially in specimens, where this process results in progressively finer grains (e.g. in the TD specimen). The extended recovery can continue refining further the grains until the combination of strain and heat renders the pinning action of the precipitates ineffective to prevent long range motion of the grain boundaries. Discontinuous dynamic recrystallization then takes place leading to rapid necking and specimen failure.

4.6 Conclusions On the deformation, recovery and recrystallization in the superplastic deformation of the coarse-grained aluminum alloy AA5182 the following conclusions are drawn:

99

Chapter 4 •

Continuous grain refinement does not result in large elongations. Maintaining a stable grain size, for a large regime of ε L seems to be more critical in obtaining large elongations prior to failure.

•

Grain coherency promotes the rotation of deformed grains into alternating Cube and Goss orientations, so as to allow for the operation of multiple slip systems across the grain boundaries at large values of ε L . Consequently, an evenly balanced Cube and Goss texture is prone to produce larger elongations prior to failure.

•

Extended recovery or continuous dynamic recrystallization leads to homogeneous grain refinement.

•

In the low volume fraction of the large deformed grains a relatively slow dislocation controlled process rearranges the microstructure forming many subgrain boundaries and low-angle grain boundaries which are often pinned by small precipitates.

•

During recovery low-angle grain boundaries are converted into high-angle grain boundaries which reduce the number of grains showing large amount of dislocation induced distortions.

•

The end product of the transformation is a large volume fraction of small recrystallized grains separated by high-angle grain boundaries whose migration absorbs any remaining sub-grain boundaries and/or low-angle grain boundaries.

•

The extended recovery can continue refining the grains further until the combination of strain and heat renders the pinning action of the precipitates ineffective in preventing long range motion of the grain boundaries. Discontinuous dynamic recrystallization then takes place leading to rapid necking.

100

Deformation Processes •

Texture randomization, with a volume fraction of recrystallized grains exceeding 80%, large densities of high-angle grain boundaries, and large grain sizes are consequences of discontinuous dynamic recrystallization occurring at the failure point.

References [1]

Hurley PJ, Humphreys FJ. Acta Mater 2003;51:1087.

[2]

Xun Y, Tan MJ, Nieh TG. Mater Sci Tech 2004;20:173.

[3]

Bate PS, Humphreys FJ, Ridley N, Zhang B. Acta Mater 2005;53:3059.

[4]

Agarwal S, Briant CL, Krajewski PE, Bower AF, Taleff EM. J Mater Eng Perform 2007;16:170.

[5]

Smallman RE, Bishop RJ. Metals and Materials: science processes and applications. Oxford: Butterworth – Heinemann; 1995. p. 206.

[6]

Kazantzis AV, Chen Z, Kroezen HJ, Bijlsma JA, De Hosson JThM. Acta Mater (to be submitted).

[7]

Langdon TG. Metal Sci 1982;16:175.

[8]

Xu c, Furukawa M, Horita Z, Langdon TG. Acta Mater 2003;51:6139.

[9]

Doherty RD, Hughes DA, Humphreys FJ, Jonas JJ, Jensen DJ, Kassner ME, King WE, McNelley TR, McQueen HJ, Rollett AD. Mater Sci Eng A 1997;238:219.

[10]

Sitdikov O, Sakai T, Avtokratova E, Kaibyshev R, Tsuzaki K, Watanabe Y. Acta Mater 2008;56:821.

[11]

Han JH, Jee KK, Oh KH. Int J Mech Sci 2003;45:1613.

[12]

Hornbogen E, Köster U. In: Haessner F, Riederer Dr, editors. Recrystallization of Metalic Materials, Berlin:Verlag; 1978 p.159.

[13]

McQueen HJ, Jonas JJ. Treatise on Materials Science and Technology, vol. 6. New York: Academic Press; 1975. p. 393.

[14]

Hornbogen E. Metall Trans A 1979:10A:947.

[15]

Nes E. In: Baudelet B, Surey M, editors. Superplasticité, Centre National de la Recherche Scientifique, Paris; 1985. pp. 7.1.

101

5 Deformation Mechanisms In comparison to steel aluminum alloy sheet materials have a lower formability in cold stamping processes [1-5]. The 5000 series aluminum alloy (AA 5xxxx) offers an alternative approach since it can be deformed to quite a high percentage at elevated temperature by the so-called superplastic forming process [6-19]. Because the deformation mechanisms fall into the grain boundary sliding (GBS) regime [20-25], fine grain size of 10 µm, high operation temperature and slow strain rate are required as for the processing of typical AA5083 material. Quick plastic forming technique is a promising replacement of superplastic forming, which can be achieved at lower temperatures and higher strain rate for time- and cost-effective considerations but still with fine-grained AA5083 material [15, 19, 26-32]. As far as the deformation mechanisms are considered, the grain boundary sliding and solute drag creep contribute together [19]. A further decrease in the cost of the primary material was found to be possible with the use of coarse-grained AA5083 [33-35]. These alloys with an initial grain size of 70 µm, exhibited a maximum elongation to failure in excess of 300% at 440°C and at 10-2 s-1. The behavior of the material in the solute drag creep regime was found to promote dynamic recovery in regions of the gauge. This type of dynamic reconstruction led to a significant grain refinement of the microstructure (with an average grain size equal to 43 µm), resulting in an enhancement of plasticity. Since coarse-grained materials may well fulfill the industrial requirements, the application of materials of low purity, such as the coarse-grained AA5182, if successful, will constitute a further step towards cost reduction.

102

Deformation Mechanisms

5.1 Strain Rate Change Test Results Two commercial Aluminum alloy AA5182 materials were used with composition of Al-5.0% Mg-0.3% Mn-0.1% Cu (wt.%) and minor impurities of Si, Fe, and Cr. The Material B indicates the alloy sheet with thickness of 2 mm and average grain size of 21µm, and the Materials C has thickness of 1.5 mm and average grain size of 37µm. Dog-bone specimens were cut by Electrical Discharge Machining. The gauge directions have different orientations i.e. along, diagonal- and perpendicular-to-the rolling direction (RD) which will be noted for Materials B and C as Ba, Bd, Bp and Ca, Cd, Cp. The gauge width is 4 mm and the length is 16 mm with a curvature section at each end connected to the grip having a radius of 2 mm. In addition, the specimens were tested using shoulder-loading rigid grips. The mass flow from the specimen grip to the gauge can be neglected. Moreover, the rigid grip design makes it possible to calculate the true strain of the gauge from the grip displacement. This accuracy increases with the specimen gauge length. Strain Rate Change (SRC) tests were conducted at elevated temperatures for each material at temperature 400°C to 500°C and strain rate from 5x10-4s-1 to 2x10-1s-1. The temperatures were controlled within 3°C with thermal couples attached on the specimen and the extension grips in a three-zone furnace. Specimens were tested at true strain rate in a hydraulic-driven, computer-controlled, electromechanical testing machine to impose prescribed strain rates. For each single specimen the strain rate change tests use a series of strain rates imposed for increasing strain rate changes (5x10-4s-1 to 2x10-1s-1) or decreasing strain rate changes (1x10-1s-1 to 5x10-4s-1), with each rate held for a minimum of 2 pct engineering strain, to produce data for flow stress as a function of temperature,

103

Chapter 5 strain, and strain rate. “Steady-state” flow stress measurements were made after each rate change by evaluating the stress transient following the rate change and measuring the stress at which the transient had fully decayed. Thereafter, when the specimen was tested using increasing or decreasing strain rate change, it will be described as Up or Down specimen, for example Ba-Up or Ba-Down specimen.

5.2 Strain Rate Sensitivity, m a)

Strain Rate Change Test Results

Representative data of strain rate change tests on specimens Ba-UP, Ba-Down, Ca-Up and Ca-Down at temperatures from 400°C to 475°C for material B and 425°C to 500°C for material C are shown in Figure 5.1. These data are plotted as the logarithm of true-strain rate , ε& , vs true steady-state flow stress, σ, compensated by the dynamic unrelaxed, Young’s elastic modulus E. Temperature-dependent elastic (dynamic) modulus values are calculated from the following fit to the data of Köster for pure Al: [36] E = 77.630 + 12.98T − 0.03084T 2

(1)

where E is expressed in MPa and T is absolute temperature in Kelvin. The slope on the plots is equivalent to the stress exponent, n, from the phenomenological equation for creep, which can be written as [37] :

Q b σ ε& = A( ) p ( )n exp(− c ) d

E

RT

(2)

where A is a constant that depends on the material and the dominant deformation mechanism, b is the magnitude of the Burgers vector, d is the grain size, p is the grainsize exponent, Qc is the activation energy for creep, R is the universal gas constant, and the remaining terms are as abovementioned. 104

Deformation Mechanisms The stress exponent values represented by the data in Figures 1 were measured at temperatures 425°C and 450°C for all four specimens. The calculation method is as follows: for a given specimen at a constant temperature, flow stress data at three consecutive strain rates were taken from the strain rate change test data, these three data were used for linear regression on the dual-logarithmic scales as shown in Figure 1, while the slope of the resulting line was taken as the stress exponent n for the central strain rate. Thereby the stress components could be calculated at all intermediate stain rates except the lowest and the highest ones. The resulting results of the stress exponent are plotted in Figure 5.2 against the logarithm of strain rate at (a) 425°C and (b) 450°C. 0

-1

10

-2

10

-3

10

-4

10

5

10

-1

10

-2

10

-3

10

-4

4

2 Ba-UP 10

0

a

-1

10

450癈 425癈 400癈

ε (s )

-1

ε (s )

10

b

o

450 C o 425 C o 400 C

5 3

5 Ba-DOWN

-3

10

σ/E

σ/E 0

-1

10

-2

10

-3

10

-4

5 4

-1

10

10

c

o

500 C o 475 C o 450 C o 425 C

ε (s )

-1

ε (s )

10

3

Ca-UP

10 σ/E

-3

-3

0

10

-1

10

-2

10

-3

10

-4

d

o

500 C o 475 C o 450 C o 425 C

5 4

3

Ca-DOWN

10 σ/E

-3

Figure 5.1: The true-strain rate ε& vs compensated true steady-state flow stress σ/E curves for specimens a) Ba-Up, b) Ba-Down, c) Ca-Up and d) Ca-Down. (The symbol ε should be ε& at y-axis. The error bars are just as small as the scatters. This applies to all the subsequent figures.)

105

Chapter 5 At 425°C, Figure 5.2(a) shows the following phenomena. The Ba-Up specimen exhibits n values increasing from 2.3 to 3.3 and to 5.4 with increasing strain rate. The n value of 2.3 at the lowest strain rate is consistent with that observed in conventional superplastic material AA5083 with fine grain size of less than 10 µm which can be deformed by the Grain Boundary Sliding mechanism, which is enabled at very low strain rate typically with ε& 5). The resulting deformation mechanism map for the specimen Ba-Up is shown in Figures 5. The regions over which strain rate change experiments were conducted are indicated in Figure 5.5 by dashed lines.

5.5 Conclusion Strain rate change mechanical tests were conducted to characterize the deformation behavior of four AA5182 specimens from two alloy materials. The Ba-Up specimen with grain size of 21 µm demonstrated stress exponents and activation energies characteristic of grain boundary sliding, solute drag creep and dislocation glide creep at low, intermediate and high strain rates. Solute drag creep and dislocation glide creep are the dominating mechanisms governing the deformation and contributing to the high strain

112

Deformation Mechanisms rate superplasticity. It is also demonstrated that dislocation glide creep is the only dominating mechanism responsible for superplasticity for specimens Ba-Down, Ca-Up and Ca-Down. For the Ba-Up specimen, a deformation mechanism map was constructed highlighting the dominant mechanism in regions with different temperatures and stresses level.

References [1] Hecker SS. Formability of aluminium alloy sheets. Transactions of the ASME. Series H, Journal of Engineering Materials and Technology|Transactions of the ASME. Series H, Journal of Engineering Materials and Technology 1975;97:66. [2] Golovashchenko SF, Mamutov V, Dmitriev VV, Sherman AM. Formability of sheet metal with pulsed electromagnetic and electrohydraulic technologies. Aluminum 2003 2003:99. [3] Miki Y, Koyama K, Noguchi O, Ueno Y, Komatsubara T. Increase of Lankford value of AI-Mg-Si sheets for automotive panel produced by asymmetric warm rolling. THERMEC 2006, Pts 1-5 2007;539543:333. [4] Sousa LC, Castro CF, Antonio CC. Optimization of forming processes with different sheet metal alloys. NUMIFORM '07: Materials Processing and Design: Modeling, Simulation and Applications, Pts I and II 2007;908:467. [5] Zhongqi Y, Zhongqin L, Yuying Y. Prediction of fracture in square-cup forming of aluminium alloy. Materials Science Forum 2007:703. [6] Kulas MA, Green WP, Pettengill EC, Krajewski PE, Taleff EM. Superplastic failure mechanisms and ductility of AA5083. Advances in Superplasticity and Superplastic Forming 2004:127. [7] Valiev RZ, Salimonenko DA, Tsenev NK, Berbon PB, Langdon TG. Observations of high strain rate superplasticity in commercial aluminum alloys with ultrafine grain sizes. Scripta Materialia 1997;37:1945. [8] Langdon TG. Recent developments in high strain rate superplasticity. Materials Transactions Jim 1999;40:716. [9] Horita Z, Komura S, Berbon PB, Utsunomiya A, Furukawa M, Nemoto M, Langdon TG. Superplasticity of ultrafine-grained aluminum alloys processed by equal-channel angular pressing. Towards Innovation in Superplasticity Ii 1999;304-3:91. [10] Akamatsu H, Fujinami T, Horita Z, Langdon TG. Influence of rolling on the superplastic behaviour of an Al-Mg-Sc alloy after ECAP. Scripta Materialia 2001;44:759. [11] Furukawa M, Utsunomiya A, Matsubara K, Horita Z, Langdon TG. Influence of magnesium on grain refinement and ductility in a dilute Al-Sc alloy. Acta Materialia 2001;49:3829. [12] Lee S, Utsunomiya A, Akamatsu H, Neishi K, Furukawa M, Horita Z, Langdon TG. Influence of scandium and zirconium on grain stability and superplastic ductilities in ultrafine-grained Al-Mg alloys. Acta Materialia 2002;50:553. [13] Taleff EM, Nagao M, Higashi K, Sherby OD. High-strain-rate superplasticity in ultrahigh-carbon steel containing 10wt.%Al (UHCS-10Al). Scripta Materialia 1996;34:1919.

113

Chapter 5 [14] Taleff EM, Nevland PJ, Krajewski PE. Tensile ductility of several commercial aluminum alloys at elevated temperatures. Metallurgical and Materials Transactions a-Physical Metallurgy and Materials Science 2001;32:1119. [15] Kulas MA, Green WP, Taleff EM, Krajewski PE, McNelley TR. Failure mechanisms in superplastic AA5083 materials. Metallurgical and Materials Transactions a-Physical Metallurgy and Materials Science 2006;37A:645. [16] Taleff EM, Hector LG, Bradley JR, Verma R, Krajewski PE. The effect of stress state on hightemperature deformation of fine-grained aluminum-magnesium alloy AA5083 sheet. Acta Materialia 2009;57:2812. [17] Figueiredo RB, Kawasaki M, Langdon TG. Developing Superplasticity in Metallic Alloys through the Application of Severe Plastic Deformation. Recent Developments in the Processing and Applications of Structural Metals and Alloys 2009;604-605:97. [18] Figueiredo RB, Kawasaki M, Langdon TG. The Mechanical Properties of Ultrafine-Grained Metals at Elevated Temperatures. Reviews on Advanced Materials Science 2009;19:1. [19] McNelley TR, Oh-Ishi K, Zhilyaev AP, Swaminathan S, Krajewski PE, Taleff EM. Characteristics of the transition from grain boundary sliding to solute drag creep in superplastic AA5083. Metallurgical and Materials Transactions a-Physical Metallurgy and Materials Science 2008;39A:50. [20]

Ball A, Hutchison MM. Superplasticity in the aluminium-zinc eutectoid. Met. Sci. J. 1969;3:1.

[21] Langdon TG. The Mechanical-Properties of Superplastic Materials. Metallurgical Transactions aPhysical Metallurgy and Materials Science 1982;13:689. [22] Arieli A, Mukherjee AK. The Rate-Controlling Deformation Mechanisms in Superplasticity - a Critical-Assessment. Metallurgical Transactions a-Physical Metallurgy and Materials Science 1982;13:717. [23] Edington JW. Microstructural Aspects of Superplasticity. Metallurgical Transactions a-Physical Metallurgy and Materials Science 1982;13:703. [24] Murty GS, Koczak MJ. Investigation of Region-I of a Superplastic Al-Zn-Mg-Cu-Mn Alloy. Materials Science and Engineering 1987;96:117. [25] Wadsworth J, Ruano OA, Sherby OD. Denuded zones, diffusional creep, and grain boundary sliding. Metallurgical and Materials Transactions a-Physical Metallurgy and Materials Science 2002;33:219. [26] Schroth JG. General motors' quick plastic forming process. Advances in Superplasticity and Superplastic Forming 2004:9. [27] Montgomery GP. Effect of the constitutive equation on MARC analysis of quick plastic forming. Advances in Superplasticity and Superplastic Forming 2004:323. [28] Krajewski PE, Morales AT. Tribological issues during quick plastic forming. Journal of Materials Engineering and Performance 2004;13:700. [29] Kullas MA, Paul Green W, Taleff EM, Krajewski PE, McNelley TR. Deformation mechanisms in superplastic AA5083 materials. Metallurgical and Materials Transactions A (Physical Metallurgy and Materials Science) 2005;36A:1249. [30] Krajewski PE, Schroth JG. Overview of Quick Plastic Forming technology. Superplasticity in Advanced Materials 2007;551-552:3. [31] Kulas MA, Krajewski PE, Bradley JR, Taleff EM. Forming limit diagrams for AA5083 under SPF and QPF conditions. Superplasticity in Advanced Materials 2007;551-552:129. [32] Kulas MA, Krajewski PE, Bradley JR, Taleff EM. Forming-limit diagrams for hot-forming of AA5083 aluminum sheet: Continuously cast material. Journal of Materials Engineering and Performance 2007;16:308. [33] Soer WA, Chezan AR, De Hosson JTM. Deformation and reconstruction mechanisms in coarsegrained superplastic Al-Mg alloys. Acta Materialia 2006;54:3827.

114

Deformation Mechanisms [34] Chezan AR, De Hosson JTM. Superplastic behaviour of coarse-grained aluminum alloys. Materials Science and Engineering a-Structural Materials Properties Microstructure and Processing 2005;410:120. [35] Chezan AR, De Hosson JTM. The role of microstructural aspects on the performance of coarsegrained superplastic Al alloys. Icotom 14: Textures of Materials, Pts 1and 2 2005;495-497:883. [36] Köster W. The Temperature Dependence of the Elasticity Modulus of Pure Metals. Zeitschrift fur Metallkunde 1948;39:1. [37] Sherby OD, Wadsworth J. Superplasticity. Recent advances and future directions. Progress in Materials Science 1989;33:169. [38] Hsiao IC, Huang JC. Deformation mechanisms during low- and high-temperature superplasticity in 5083 Al-Mg alloy. Metallurgical and Materials Transactions a-Physical Metallurgy and Materials Science 2002;33:1373. [39] Taleff EM, Henshall GA, Nieh TG, Lesuer DR, Wadsworth J. Warm-temperature tensile ductility in Al-Mg alloys. Metallurgical and Materials Transactions a-Physical Metallurgy and Materials Science 1998;29:1081. [40] Lee SW, Yeh JW. Superplasticity of 5083 alloys with Zr and Mn additions produced by reciprocating extrusion. Materials Science and Engineering a-Structural Materials Properties Microstructure and Processing 2007;460:409. [41] Jien-Wei Y, Shih-Wei L. Superplasticity of 5083 alloys with Zr and Mn additions produced by reciprocating extrusion. Materials Science & Engineering A (Structural Materials: Properties, Microstructure and Processing) 2007;460-461:409. [42]

Barrett CR, Ardell AJ, Sherby OD. Trans. AIME 1964;230:200.

[43] Nieh TG, Wadsworth J, Sherby OD. Superplasticity in Metals and Ceramics. Cambridge University Press, Cambridge, United Kingdom 1997:40.

115

6 Summary and Outlook Summary Aluminum alloys may fulfill the increasing demands in automotive industry to reduce weight so as to enhance performance and to reduce fuel consumption, provided the formability could be improved to a level comparable with that of their steel counterparts. Aluminum alloy AA5083 is currently the referred material for the so-called Superplastic Forming (SPF) process, alternatively employed in industrial application. Affected by the deformation mechanism of Grain boundary sliding (GBS), this technique has the disadvantage of slow forming rates together with the requirement of high forming temperature and small grain size. The patented Quick Plastic Forming (QPF®) technique is developed at significantly higher strain rate than SPF, capable of producing aluminum parts at relatively high volumes and extremely complex shapes. As the deformation mechanism switches to solute drag creep (SDC), coarse grained AA5182 plays the important role putting the cost-effective efforts even forward.

Mechanical Properties toward Superplasticity Uniaxial tensile tests of AA5182 sheet material were conducted over a reach of e conditions of temperature, strain rate and specimen geometry. The coarse-grained AA5182 with grain size of 21 µm and 37 µm (denoted by 21G and 37G), exhibited optimum deformation conditions at ε& 10-2 s-1 and at T 425°C and above 475°C, respectively, with maximum elongations to failure between 300 and 400% along, and at

116

Summary and Outlook 45° to the rolling direction and approximately equal to 300% perpendicularly to the rolling direction. The alloy 21G is essentially isotropic, exhibiting consistently uniform deformation at all orientations up to an elongation of 250%, at 425°C and at 10-2 s-1, whereas the 37G is slightly anisotropic and demonstrated significant deviations from uniformity at 200%, as well as significant necking and numerous secondary necking instabilities at 250%. Secondary necking instabilities were most likely to appear at the optimum deformation conditions, especially when elongations in excess of 400% were achieved, but their development could not be predicted. For both alloys these instabilities were most likely to appear in specimens along the rolling direction; their presence in specimens at 45° was less frequent and it was rare in those perpendicularly to the RD. These secondary necking instabilities are associated with regions of the gauge that contain a large volume fraction of soft grains and produce microstructures that exhibit maxima in the Cube and Goss component of the deformed grains. They exhibit slight grain refinement compared with the adjoining thicker and more uniformly deformed regions.

Deformation, Recovery and Recrystallization processes The Electron Backscatter Diffraction (EBSD) maps were initially partitioned into deformed, recovered (RV) and recrystallized (RX) portions. The microstructure evolution of the texture, the grain size, the grain volume fraction, the sub-grain, low-angle and high-angle grain boundary density were investigated using predefined, sequentially increasing incremental values of the local strain, ε L , along the post-mortem gauge of

117

Chapter 6 tensile specimens. This new approach permitted the comparative evaluation of the progress of these three processes, and thus produced important conclusions with respect to the total mechanism responsible for the microstructure modification along the gauge of specimens from AA5182 aluminum alloy upon “superplastic” extension. From the analysis of grain size evolution, it is shown that continuous grain refinement does not result in large elongations. Maintaining a stable grain size, for a large regime of ε L seems to be more critical in obtaining large elongations prior to failure. Grain coherency promotes the rotation of deformed grains into alternating Cube and Goss orientations, so as to allow for the operation of multiple slip systems across the grain boundaries at large values of ε L . Consequently, an evenly balanced Cube and Goss texture is prone to produce larger elongations prior to failure. Extended recovery or continuous dynamic recrystallization leads to homogeneous grain refinement. During recovery low-angle grain boundaries are converted into highangle grain boundaries and the process reduces the number of grains showing large amount of dislocation induced distortions. The extended recovery can continue refining grains further until the combination of strain and heat renders the pinning action of the precipitates ineffective to prevent long range motion of the grain boundaries. Discontinuous dynamic recrystallization takes place leading to rapid necking.

Deformation mechanisms Strain rate change mechanical tests were conducted to characterize the deformation behavior of four AA5182 specimens from two alloy materials. The activation energy for creep behavior, Qc, was calculated to determine the deformation mechanisms. At very low flow stress region, the Qc has a low value around 110 kJ/mol 118

Summary and Outlook indicating the grain boundary diffusion actively participates during deformation by grain boundary sliding creep. At intermediate flow stress region Qc increases to 138±2 kJ/mole supposing a solute drag creep deformation mechanism corresponding to the diffusion of Mg solute in the Al matrix (Qc= 136 kJ/mol). While the flow stress goes to the highest region, the Qc value increases to more than 180 kJ/mol. This value is higher than self-diffusion in Al (Qc= 142 kJ/mol) and lower than solute diffusion energy in Al (Qc,Mn = 220 kJ/mol, Qc,Fe = 213 kJ/mol, Qc,Cr = 252 kJ/mol). Therefore, it is concluded that this phenomenon is attributed to the Dislocation Glide Creep mechanism in high-stress region, which can be explained as the dislocation/dislocation interactions or the dislocation glide limited by dispersed particles.

Outlook The conventional superplastic material of AA5083 has relatively high purity and fine grain size requiring significant thermo-mechanical processing. As a consequence the material production cost will be considerably high. Although various deformation mechanisms have been studied and deformation models were proposed, there are some intrinsic characteristics of the material that hamper industrial application. Because grain boundary sliding is the dominating controlling mechanism, the deformation strain rate can only typically be of the order of 10-4s-1. Subsequently the forming time of a component will be too long for the consideration of cost control in industry. Also, due to the grain boundary sliding mechanism, the deformation temperature has to be relatively high, namely close to the melting point of the material. None of the abovementioned

119

Chapter 6 factors is cost-effectively favorable for the material to scale production as in industrial application. As proved in this thesis work, our Aluminum alloy is demonstrated to be the perfect candidate to the new technique known as Quick Plastic Forming (QPF®). Compared to AA5083, the coarse-grained material studied in this thesis has bigger grain size and lower purity, making it easier and cheaper to produce. Above all, the deformation mechanisms not only rely on grain boundary sliding creep, but the combination of grain boundary sliding and/(or just) other creeps of solute drag creep and dislocation glide creep. These have the significant importance for reducing the forming time of a component as the strain rate is typically high at 10-2s-1. The deformation temperature is thereafter also much lower, 425°C compared to the 500°C employed for AA5083. The deformation mechanisms are less established yet for the AA5182 material. A slight change in composition or different processing method applied by different industries may cause significant differences in mechanical performance. This is mainly due to the virginal starting microstructures and grain size. The distributions of dispersoids are different to hinder dislocation motion during the deformation, giving varied deformation performance and ductility. Hence, microstructural characterizations before and after the deformation should be conducted in order to formulate appropriate evolution models. Detailed investigation of the deformation mechanism is recommended for the design of new coarse grained alloys.

120

Acknowledgements Four years’ study as a PhD student in the MK-groep has brought me tremendous success and pleasure mostly but not least summarized in this thesis. I would like to thank the people who supported me academically but also the good friends I have made, who helped me directly or indirectly to make this happen. Fist of all, I would like to thank my promotor, Professor Jeff Th.M. De Hosson, for giving me the opportunity to come to The Netherlands to join a magnificent research team, for giving me the freedom to explore my scientific interests, for providing direct access to such a wide array of equipments, for supporting me go to international conferences and, most of all, for maintaining such an inspiring and enjoyable working environment. I am very thankful for the scientific guidance and all the scientific discussions you provided, not only this but also I am extremely grateful for your encouragement during the research project and your help with finding for me a challenging postdoc position in the USA. Dr. Tony Kazantzis, thank you for your patient carrying out the tensile tests and excellent accomplishment on time consuming and tedious analysis of these experimental data. I have benefited a lot from your knowledge in the field along with our pleasant discussions. What a profound friendship we have which will not fade away as time passes. Dr. Lin Zhuang and Dr. Toni Chezan from Corus® for your assistance providing the state-of-the-art material, and I am very thankful for your time and giving valuable advices at project progress meetings. Dr. Yutao Pei, interacting with you has been an inspiring and joyous time. I am grateful that your door is always open to my trifle questions, not only to this but also I am extremely thankful to your warm-hearted help on non-work related matters. I have learned so much from you not only the results we have achieved but also the way how to achieve them. I am truly indebted to you for your advice, support and motivation. Equally, my thanks are due to Dr. David Vainshtein and Dr. Václav Ocelík: many thanks for your scientific support, assistance and help. Dr. Paul Bronsveld, thank you for your genuine

121

interests in my work and your inspiring discussions. Ms. Elly Eekhof our MK secretary for your well organized administration, making life easy for me. Members of my reading committee, Profs. L. Katgerman, P. Rudolf, H. A. De Raedt– thank you all for your careful reading and for your valuable comments and suggestions. My paranymphs, Kalpak Shaha and Huajie Yang thank you for your great help to make this happen in time. My roommates, Kalpak Shaha, Xuan Lam Bui, ChangQiang Chen and Willem van Dorp, thank you all for the enjoyable discussion and pleasant time we spent together. I thank all other MK-groep members, including Uazir O. B. de Oliveira, Dave Matthews, Jiancun Rao, Sascha Fedorov, Emiel Amsterdam, Damiano Galvan, Maarten Hilgenga, Alessio Morelli, Anton Sugonyako, Willem-Pier Vellinga, Sriram Venkatesan, Gopi Krishnan, Eric Detsi, Diego Martinez-Martinez, Anatoly Turkin, Jasper Oosthoek, Alexander Kilimovitskiy, Enne Faber, Ismail Hemmati, Jozef Vincenc Obona, Oleksii Kuzmin, Sergey Punzhin, Ivan Furár, Peter van Zwol, Ramanathaswamy Pandian. Dad and Mum, thank you so much for your support, your belief and your love. You made everything possible. My brother Yonggang, sister in law Bin and my cousin Zi’ang, thank you for your concern and loving care. My wife, Lei, you make me laugh all the time, spent the delighted time with me and support me always. It does not go unnoticed. My dear babies, Feiyu and Pengyuan, you bring me the wonderful like I never had before.

Zhenguo th

Groningen 30 July 2010

122