Ultrafine-grained Metal Sheets produced using the Accumulative Roll Bonding Process for Light

Ultrafine‐grained Metal Sheets produced using the Accumulative Roll Bonding Process

for Light‐Weight Structures

Der Technischen Fakultät der Universität Erlangen‐Nürnberg zur Erlangung des Grades

DOKTOR‐INGENIEUR

vorgelegt von

Irena Topić

Erlangen 2008

Herstellung von ultrafeinkörnigen Blechen mittels des kumulativen Walzprozesses

für den Leichtbau

Der Technischen Fakultät der Universität Erlangen‐Nürnberg zur Erlangung des Grades

DOKTOR‐INGENIEUR

vorgelegt von

Irena Topić

Erlangen 2008

Als Dissertation genehmigt von der Technischen Fakultät der Universität Erlangen‐Nürnberg

Tag der Einreichung: 24.11.2008 Tag der Promotion: 15.04.2009 Dekan: Prof. Dr.‐Ing. Johannes Huber Berichterstatter: Prof. Dr. rer. nat. Mathias Göken Prof. Dr.‐Ing. Marion Merklein

ABSTRACT Over the last decade, nanocrystalline and ultrafine-grained (UFG) materials with a grain size of less than 1 µm have aroused considerable interest due to their superior mechanical properties in terms of strength and/or ductility compared to conventionally grained materials. As such, they have a strong potential for prospective engineering applications for structural, high durability components in automobile, aerospace and medical industry. In this work, different materials such as commercial purity aluminium AA1050, aluminium alloy AA6016, oxygen free copper, titanium and niobium were processed by the Severe Plastic Deformation (SPD) technique called Accumulative Roll Bonding (ARB) in order to produce an ultrafine-grained microstructure and improve the mechanical properties. One of the biggest advantages of the ARB process in comparison to other SPD methods such as Equal Channel Angular Pressing (ECAP) or High Pressure Torsion (HPT) is that it is a continuous process, which can be incorporated in industry to produce large scale UFG metal sheets. During the ARB process, the metal sheet surfaces are wire brushed in order to remove the oxide layer, stacked on top of each other and rolled together with a thickness reduction of 50 %. The metals sheets bond together during rolling and the procedure can then be repeated any number of times. The material is subjected to very high plastic, shear deformation and the UFG microstructure starts to develop after approximately 4 ARB cycles. This study focuses primarily on the ARB processed commercial purity aluminium AA1050 and the technically relevant aluminium alloy AA6016. Ultrafine-grained Al metal sheets are especially interesting for light weight construction in the automobile industry due to their high specific strength. In order to qualify the accumulative roll bonding process for these purposes detailed investigations on microstructural evolution, mechanical properties and sheet metal forming using bulge tests and cup drawing tests have been carried out and investigated. Sheet metal joining is one further technologically important issue, which places a challenge upon UFG aluminium sheet materials. Friction Stir Welding (FSW) was found to be a desirable joining technique for UFG materials, since it provides excellent mechanical properties and retains the fine grained microstructure. During the course of this study, the ARB process was adapted and optimised for every materials system and the quality of the sheets was improved. The ARB process was significantly shortened and it became more robust. The deformation during rolling became more homogeneous, cracking of the edges was eliminated and crack propagation was suppressed. These factors cumulatively contributed to less material waste during the process. The quality of the surface was considerably improved and the sheet thickness became more homogeneous. The contribution of a four-high rolling mill was especially manifested in terms of the final width of metal sheets. Irrespective of the process parameters, rapid grain refinement and significantly higher hardness and strength with increasing number of ARB cycles, in comparison to the CG counterpart were observed for all materials. The ARB processed materials are microstructurally anisotropic and they develop a characteristic ß-fibre texture with a Cu component. UFG Al sheets showed promising sheet metal forming potential under biaxial stress state conditions and under tension-compression conditions, which occur during cup drawing experiments. Generally, UFG aluminium samples rolled up to 4 ARB cycles showed a good compromise between strength, elongation to failure, minimal sheet thinning and earing during deep drawing cup tests. However, deep drawing cup tests showed that the metal sheet formability significantly increases at elevated temperatures. Furthermore, the UFG materials confirmed that their enhanced strain rate sensitivity can be advantageously used in order to achieve higher formability. The UFG AA1050 and AA6016 sheets were successfully friction stir welded. Although a drop in hardness is measured in the nugget for both materials, the hardness is comparable to

that of the CG counterparts and is not considered to be a technological limitation. However, bulge tests and cup drawing tests both confirmed limited formability, which appears to be governed by the amount of deformation and strength of the nugget. Even though there is still very limited amount of research regarding the formability and direct applications of UFG sheets, their potential should not be underestimated. The production of UFG materials can become commercialised and cost effective, and it could become possible to control the mechanical properties of materials by processing rather than by alloying. In the meantime, the big technical potential of ARB processed materials was also recognised by the aluminium manufacturers. Future interests are closely related to superplastic forming, accumulative roll bonding of magnesium alloys for lightweight structural components, as well as accumulative roll bonding of IF-steel. Thus, the innovation potential of the UFG materials for advanced applications in engineering is high, and the requirements for producing such materials are becoming more and more economically feasible.

KURZFASSUNG In den letzten Jahren ist das Interesse an nanokristallinen und ultrafeinkörnigen Werkstoffen enorm gestiegen. Ultrafeinkörnige (engl.: ultrafine-grained UFG) Werkstoffe mit einer Korngröße von etwa 100 nm bis 1000 nm besitzen außergewöhnliche und für die technische Anwendung vielversprechende mechanische Eigenschaften im Vergleich zu Werkstoffen mit konventioneller Korngröße. Durch die hohe spezifische Festigkeit haben UFG-Werkstoffe insbesondere im Bereich der Konstruktionswerkstoffe und des Leichtbaus ein großes Anwendungspotenzial. Im Rahmen dieser Arbeit wurden verschiedene Werkstoffen konventioneller Korngröße wie technisch reines Aluminium (AA1050), die Aluminiumlegierung AA6016, hochreines Kupfer, Titan und Niob mit dem sogenannten kumulativen Walzprozess (engl.: Accumulative Roll Bonding ARB) umgeformt. Mit diesem Verfahren lassen sich hauptsächlich flächige Bauteile mit ultrafeinkörnigen Mikrostrukturen erzeugen. Einer der größten Vorteile dieses Prozesses im Vergleicht zu den anderen Hochverformungsprozessen wie z.B. Equal Channel Angular Pressing (ECAP) oder High Pressure Torsion (HPT), ist dass er verhältnismäßig leicht in bestehende Walzanlagen integriert werden kann und so großflächige ultrafeinkörnige Bauteile hergestellt werden können. Grundlage des ARB-Prozesses ist es, dass ein Blechwerkstoff wiederholt einer Walzverformung (Scherverformung) unterzogen wird. Um den Prozess hinreichend häufig durchführen zu können, wird nach jedem Walzdurchgang mit einer Stichabnahme von 50 % das Blech in der Länge halbiert. Nach dem Drahtbürsten zur Beseitigung der Oberflächenoxide und Erzeugung einer entsprechend aufgerauten Oberfläche werden beide Blechstreifen wieder aufeinander gelegt und fixiert. Der Walzschritt wird dann von neuem ausgeführt. Während des Walzvorgangs erfolgt dabei ein Verbinden der beiden Blechlagen. Durch Wiederholung dieses Vorgangs kann nach einer hinreichend großen Anzahl an ARB-Zyklen eine ultrafeinkörnige Mikrostruktur mit einer Korngröße im Bereich von einigen hundert Nanometern erreicht werden. Der Schwerpunkt dieser Arbeit lag auf der Herstellung von UFG technisch reinem Aluminium (AA1050) und Aluminiumlegierung AA6016. Die ultrafeinkörnigen Aluminiumbleche sind auf Grund ihrer hohen spezifischen Festigkeit von besonderem Interesse im Bereich des Leichtbaus. Um den ARB Prozess als ein modernes Herstellungsverfahren für die UFG Bleche zu qualifizieren, wurden detaillierte Untersuchungen der Mikrostruktur, mechanischen Eigenschaften und Umformverhalten wie z.B. Tiefungsversuche und Napfziehversuche durchgeführt. Zur Herstellung großflächiger Bauteile in der technologischen Anwendung war es außerdem notwendig, geeignete Fügeverfahren hinsichtlich ihrer Auswirkungen auf die mechanischen Eigenschaften zu untersuchen. Im laufe der Arbeit hat sich gezeigt, dass das Reibrührschweißen ein geeignetes Verfahren für das Fügen von UFG Aluminium-Blechwerkstoffe darstellt. Die feinkörnige Mikrostruktur wurde während des Reibrührschweißens beibehalten und der Prozess führte zu ausgezeichnete mechanische Eigenschaften der UFG-Bleche. Durch gezielte Untersuchungen gelang es zunächst, sowohl die ARB-Prozesskette zu optimieren, als auch die erzielbare Blechqualität deutlich zu verbessern. Mit der Anschaffung eines Quartowalzgerüstes, konnte im Vergleich zur Duowalze eine homogenere Materialverformung bei gleichzeitig geringerer Rißanfälligkeit im Randbereich erreicht werden. Insgesamt führte dies zu weniger Materialverlust. Der ARB-Prozess wurde dadurch deutlich verkürzt und robuster. Gleichzeitig wurden Grundlagen für eine direkte Integration des ARB-Prozesses in die Fertigungsabläufe bei der Herstellung von Feinblechen erarbeitet. Es zeigte sich, dass als wichtigste Voraussetzungen für eine gute Bindelagenfestigkeit, eine rauhe Blechoberfläche und erhöhte Prozesstemperaturen von Nöten sind. Die Anschaffung des Quartowalzgerüsts und die Optimierung der Prozessparameter führten zu einem robusten Prozessfenster und eröffneten die Möglichkeit für eine Integration des ARBs in die

industrielle Prozesskette. Unabhängig von der Prozesstemperatur und den Prozessparametern, konnte im Allgemeinen eine rapide Kornfeinung von der konventionellen zur UFG Mikrostruktur beobachtet werden. Gleichzeitig konnte auch eine deutliche Steigerung der Härte mit zunehmender ARB-Zyklenzahl bei allen erfolgreich gewalzten Werkstoffen nachgewiesen werden. Die ARB-Werkstoffe sind stark anisotrop und entwickeln eine charakteristische ß-Fiber Walztextur mit Cu-Komponente. Die erreichten mechanischen und Umformeigenschaften zeigen deutlich, dass auch bei technologisch interessanter Legierung im UFG-Zustand ein hohes Potenzial für den technischen Einsatz vorliegt. Dieses wurde anhand von den hydraulischen Tiefungsversuchen und Napfziehversuchen beobachtet. Generell ergab sich bei den UFG-Proben nach etwa 4 ARB-Zyklen ein guter Kompromiss zwischen Festigkeit, ausreichender Bruchdehnung, einer kleinen Änderung der Blechdicke und kleiner Zipfelbildung. Es konnte auch gezeigt werden, dass das Umformpotenzial bei erhöhten Temperaturen deutlich gesteigert werden kann. Darüber hinaus haben die untersuchten UFG-Werkstoffe bestätigt, dass ihre Dehnratenabhängigkeit eine positive Auswirkung auf die Duktilität aufweisen kann. Ferner wurden beide UFG Aluminiumlegierungen, AA1050 und AA6016, erfolgreich reibrührgeschweißt. Der beobachtete Abfall der Härte an der Schweißnaht konnte mit einer während des Schweißprozesses ausgelösten Rekristallisation der Mikrostruktur erklärt werden. Da die Härtewerte an der Schweißnaht der UFG-Werkstoffe ein ähnliches Niveau wie die verschweißten CG Bleche im Ausgangszustand zeigen, besteht in dieser Hinsicht keine Begrenzung für die technische Anwendung von UFG-Werkstoffen. Die Tiefungs- und die Napfziehversuche an ARB hergestellten und reibrührgeschweißten UFG AA6016-Blechen zeigten eine begrenzte Umformbarkeit im Vergleich zu konventionellen Al-Blechen. Dies ist vermutlich auf eine starke plastische Verformung und niedrige Festigkeit der Schweißzone zurückzuführen. Obwohl noch keine ausführlichen Untersuchungen über die Umformbarkeit und eine direkte Anwendung von UFG-Werkstoffen vorhanden sind, sollte ihr Potenzial nicht unterschätzt werden. Die Herstellung von UFG-Werkstoffen kann kommerzialisiert werden und die Verbesserung der mechanischen Eigenschaften wäre durch einen Umformprozess wie das Walzen, statt durch eine Zugabe der Legierungselemente realisierbar. Das große technische Potenzial der ARB Al-Werkstoffe wurde mittlerweile auch von den Aluminiumhalbzeugherstellern erkannt. Die zukünftigen Vorschungsinteressen an UFG-Werkstoffen liegen im Bereich der Superplastizität, kumulatives Walzen von Magnesiumlegierungen und IF-Stählen. Folglich ist das Innovationspotenzial der UFG-Werkstoffe für die erweiterte technische Anwendung sehr hoch, und die Voraussetzungen für die Herstellung solche Werkstoffe werden immer mehr wirtschaftlich plausibel.

i

TABLE OF CONTENTS

1 INTRODUCTION………………………………………………………………………... 1

2 LITERATURE REVIEW………………………………………………………………... 42.1 Severe Plastic Deformation (SPD)…………………………………………………… 4

2.1.1 General Aspects of Severe Plastic Deformation (SPD) 72.1.2 Strengthening Mechanisms 8

2.2 Accumulative Roll Bonding (ARB)………………………………………….............. 112.2.1 Conventional Sheet Metal Rolling 112.2.2 Accumulative Roll Bonding Process 152.2.3 Microstructural Evolution of Accumulative Roll Bonded Metal Sheets 182.2.4 Accumulative Roll Bonding Parameters 232.2.5 Mechanical Properties of Accumulative Roll Bonded Materials 292.2.6 Strain rate sensitivity of UFG Materials 322.2.7 Thermal Stability of UFG Materials 332.2.8 Texture Formation in ARB materials 34

2.3 Sheet Metal Forming…………………………………………………......................... 352.3.1 Hydraulic Bulge Test 362.3.2 Deep Drawing 39

2.4 Friction Stir Welding (FSW)…………………………………………………………. 442.4.1 Friction stir welding of conventionally grained aluminium sheets 442.4.2 Friction stir welding of ultrafine-grained aluminium sheets 48

2.5 Corrosion of aluminium and aluminium alloys………………………………………. 512.5.1 Corrosion of conventional aluminium and aluminium alloys 512.5.2 Corrosion of ultrafine-grained aluminium and aluminium alloys 522.5.3 Corrosion of friction stir welded conventional aluminium and aluminium alloys 53

3 EXPERIMENTAL PROCEDURE………………………………………........................ 543.1 Materials Investigated………………………………………………………………… 543.2 Accumulative Roll Bonding (ARB) Process…………………………………………. 573.3 Microstructural Investigation………………………………………………………… 593.4 Surface Roughness…………………………………………………………………… 613.5 Heat Treatment……………………………………………………….......................... 62

3.5.1 Precipitation Development in AA6016 623.5.2 Thermal Stability 633.5.3 Post-roll Annealing 63

3.6 Mechanical Characterisation…………………………………………………………. 643.6.1 Hardness Testing 643.6.2 Tensile Testing 65

3.7 Sheet Metal Forming…………………………………………………......................... 693.7.1 Hydraulic Bulge Testing 693.7.2 Cup Drawing Tests 73

3.8 Friction Stir Welding (FSW)…………………………………………………………. 753.9 Corrosion Measurements……………………………………………........................... 76

4 RESULTS…………………………………………………………………………………. 784.1 ARB Process Optimisation……………………………………………........................ 784.2 Transfer of the ARB process on other technologically relevant materials…................ 82

ii

4.3 Mechanical Characterisation ………………...………………………………………. 854.3.1 Hardness Tests 854.3.2 Uniaxial Tensile Tests 89

4.4 Microstructural Characterisation …………………………………………….............. 1044.4.1 As-received and ARB processed materials 1044.4.2 Grain Size Refinement of ARB processed Aluminium Sheets 1094.4.3 Fracture and fracture surfaces of ARB samples after Tensile Testing 113

4.5 Microstructural Stability during Annealing of CG and UFG Materials……………… 1164.5.1 Annealing Response of CG Aluminium Alloy AA6016 1164.5.2 Annealing Response of ARB processed Aluminium Alloy AA6016 1174.5.3 DSC Measurements of CG and UFG Aluminium Alloy AA6016 1184.5.4 Thermal Stability of ARB processed Materials 120

4.6 Formability of CG and UFG Aluminium Sheets……………………………………... 1234.6.1 Hydraulic Bulge Tests 1234.6.2 Cup Drawing Tests 130

4.7 Friction Stir Welding…………………………………………………………………. 1364.7.1 Microstructure of friction stir welded aluminium sheets 1364.7.2 Through Thickness Hardness Distribution of FSW sheets 1384.7.3 Deformation behaviour of ARB and FSW aluminium alloys 140

4.8 Corrosion……………………………………………………………………………... 1474.8.1 Electrochemical (EC) Pen Measurements 1474.8.2 Polarisation Curves 150

5 SUMMARY AND DISCUSSION OF RESULTS ……………………………………… 1525.1 Introduction…………………………………………………………………............... 1525.2 ARB Process Optimisation and Transfer of the ARB Process on other

Technologically Relevant Materials...………………………………………............. 1535.3 Mechanical Properties of ARB Materials…………….………………………………. 1545.4 Microstructural Development during the ARB Process and Thermal Stability of

ARB materials…………………………………………………………….................. 1585.5 Formability of CG and UFG Aluminium Sheets……………………………………... 1595.6 Friction stir welding of CG and UFG Aluminium Sheets……………………………. 1615.7 Corrosion of CG and UFG Aluminium Sheets……………………………….............. 1625.8 Recommended ARB process parameters for optimal material properties……………. 163

6 CONCLUSIONS……………………………………………………………..................... 165

7 OUTLOOK…………………………………………………………………...................... 1707.1 Further improvement of ARB process………………………………………………... 1707.2 Potential applications of ARB processed aluminium sheets…………………………. 171

8 APPENDIX…………………………………………………………………...................... 1738.1 Experimental Procedure................................................................................................ 1738.2 Materials……………………………………………………………………………… 1748.3 Results…………………………………………........................................................... 175

9 REFERENCES……………………………………………………………........................ 179

CHAPTER 1 Introduction 1

1 INTRODUCTION

The main objective of the transportation and especially the automotive industry is aimed at weight and fuel reduction, and hence energy saving. Additionally, the requirements placed on today’s modern vehicles must also meet the consumer demands concerning safety and comfort. Components such as air conditioning systems or electronics meet the consumer demand, but they also contribute to the additional weight of the vehicle. The aim is therefore to reduce the overall vehicle body weight by using lighter materials or thinner metal sheets and exploiting the material diversity in order to compensate for the extra weight, while at the same time not compromising the strength. Light weight construction can be achieved by utilising lightweight materials and by designing and manufacturing the components with less weight, lower cost, improved functionality and optimised properties (Fig. 1-1). This is only possible by fully using the potential of the material, the design of load-optimised component geometries, and by using the adequate and economical manufacturing procedures [Kle06].

aluminium space frame

tailored blank

Figure 1-1: Audi A8 with an aluminium space frame (top right) and a tailored blank door structure (bottom right) for weight reduction as some of the ways of increasing the light weight construction potential of the auto body [Aud08, Hei08, Thy08]. Today’s requirements aimed at weight reduction and energy saving have positively influenced the development of new materials and the material variety (Fig. 1-2). The new material trends in the automobile industry can be divided into three main groups including high strength steels, aluminium and magnesium. Recently, the fraction of light weight metal components has drastically increased due to improvements in forming, joining, machining and strength [Mer06]. Aluminium alloys have many advantages and excellent properties, which makes them one of the most important technologically used materials. They have low density, good ductility and formability, as well as good strength, crash behaviour and corrosion resistance. Having a density of 2.8 g/cm³, which is one third of that of steel, aluminium


alloys play one of the major roles in the automotive or in general the transportation industry. Depending on the component design it is possible to achieve 30-50 % weight reduction, compared to the conventional steel components [Alu08]. The first auto body for mass production, which was completely made of aluminium and incorporated the so-called aluminium space frame (Fig. 1-1) was developed by Audi for the A2 and A8 series [Kog02]. The advantages of aluminium in terms of weight reductions was also recognised later by the BMW group with the development of the current 5th series [Zei03]. From the design point of view, it is also important to mention the tailored-blanks (Fig. 1-1). Tailored blanks are made out of two or more metal sheets of unequal thicknesses or materials, which are welded together to form a single blank (tailor welded-blank) and which can subsequently be used for sheet metal operations such as bending, stretching, deep drawing, etc. Tailor blanks have a loading optimised design and hence increase the light weight construction potential of components [Mer06].

Figure 1-2: Mechanical properties of materials used for light weight construction [Thy08]. Over the passed couple of years nanocrystalline or ultrafine-grained (UFG) materials with an average grain size below 1 µm, have received considerable attention due to their superior mechanical properties in terms of strength and/or elongation to failure in comparison to the conventionally grained (CG) materials, see for example [Hor05]. The ultrafine-grained aluminium sheets have especially raised interest for potential engineering applications as structural, high durability components in the automotive, aerospace and medical industry due to their high specific strength [Sai98]. In addition, UFG materials are also known to be strain rate sensitive and have also shown evidence of superplastic behaviour. With the above mentioned properties of the UFG metal sheets, a potential industrials application concerns structural components and tailored blanks. The advantage of the


UFG sheets is that they would provide the adequate strength in certain parts such as tailored blanks, while other parts could be formed at lower strain rates or elevated temperatures in order to achieve better formability and complex geometries. Another possibility is to use the UFG aluminium sheets for tailor heat treated blanks. In this case, the metal sheet would be heat treated locally in order to obtain e.g. better ductility, whereas the other regions would remain unaffected and retain high strength. In this way, one can tailor the mechanical properties of sheets according to the requirements. Even though nanostructured and ultrafine-grained materials are still a relatively new class of materials their interesting and superior properties in terms of high strength, ductility, strain rate sensitivity and superplasticity compared to their CG counterparts should not be neglected, and their potential should be explored further in order to successfully use them for future industrial applications as structural components, tailored blanks, or specialised applications like superplastic forming. This work focuses predominantly on commercial purity aluminium AA1050 and hardenable aluminium alloy AA6016, although other materials of technological interest such as titanium, copper, magnesium and niobium will also be mentioned. The aims of the project can be summarised and divided into five main points: 1. Manufacturing of UFG aluminium sheets by the ARB process

• Commercial purity aluminium AA1050 2. ARB process optimisation

• Evaluation of process parameters such as process temperature, surface quality, percentage reduction and etc. for easier process implementation and higher process efficiency

3. Transfer of the ARB process on other technologically relevant materials and

adaptation of process parameters according to each material • Aluminium alloy AA6016 • Pure titanium grade 2 • High purity oxygen free copper (99.99% Cu) • Niobium (99.8% purity) • AZ31 Magnesium alloy

4. Optimisation of the deformation behaviour of the ARB processed UFG materials

• Influence of number of ARB cycles and process temperature on the microstructure and mechanical properties

• Investigation of sheet metal forming potential using hydraulic bulge testing and deep drawing cup tests

• Improvement of deformation behaviour of UFG aluminium and investigation of strain rate sensitivity of UFG sheets

5. Adaptation of friction stir welding to UFG materials

• Implementing the friction stir welding process on UFG aluminium sheets • Investigation of the mechanical properties, microstructure, sheet metal forming

potential and corrosion behaviour of the welded sheets


CHAPTER 2 Literature Review

5

2 LITERATURE REVIEW

2.1 Severe Plastic Deformation (SPD) Over the last decade, nanocrystalline and ultrafine-grained (UFG) materials with an average grain size of less than ~1 µm have aroused considerable interest due to their superior mechanical properties in terms of strength and/or ductility compared to conventionally grained (CG) materials, as can be seen in figure 2-1 [Val04]. As such, they have a strong potential for prospective engineering applications as structural, high durability components in automobile, aerospace and medical industry. The ultrafine-grained sheet materials can be effectively produced by a relatively new process called Accumulative Roll Bonding (ARB). This process belongs to the group of Severe Plastic Deformation (SPD) techniques, and it was originally developed and firstly introduced by Japanese researchers Saito et al [Sai98]. The ARB process was initially performed on commercial purity aluminium. It entails repeated stacking and rolling of the conventionally grained (10 - 50 µm grain size) metal sheets in order to develop an ultrafine-grained structure.

Figure 2-1: Strength and ductility of nanostructured metals compared to the coarse grained metals. Conventional cold rolling of Cu and Al increases the yield strength, but decreases the ductility. In contrast, the extraordinary high strength and ductility of nanostructured Cu and Ti clearly set them apart from coarse grained metals [Val04]. Besides the ARB process, there are a number of other methods all belonging to the severe plastic deformation (SPD) techniques, which can be used to obtain an UFG microstructure and hence achieve an enhancement in material strength and/or ductility;


6

these include: Equal Channel Angular Pressing (ECAP) firstly introduced by Segal in the beginning of 80s [Seg77, Seg81] and later commercialised by Valiev and co-workers [Val97, Yam01, Seg95], High Pressure Torsion (HPT) [Smi86], Folding and Rolling (F&R) [Sag98], etc. However, the biggest contemporary limitation of most SPD techniques like ECAP and HPT is the up-scaling of the process to mass production for commercial applications. On the other hand, the ARB process is an adequate method for practical applications since it can be easily scaled up and adapted to a conventional rolling process to produce large scale UFG metal sheets [Sai98, Sai99, Tsu99a, Lee02a]. During the ARB process, the surfaces of two aluminium sheets are wire brushed, stacked on top of each other and rolled together to a 50 % thickness reduction. The two aluminium sheets bond during rolling and can then be halved and roll bonded again. The process is usually repeated up to 8 times. Increasing the number of ARB cycles leads to an increase in yield strength, tensile strength and/or ductility. Thus, UFG aluminium and aluminium alloy sheets with high specific strength are of technological interest and have received considerable attention as potential candidates for modern lightweight and high durability construction materials in the automotive industry. In order to qualify the ARB process for the production of high strength UFG metal sheets, detailed investigations on microstructural evolution, mechanical properties, sheet metal forming potential and joinability must be carried out. The sheet forming capability of the ARB processed materials is very crucial for a successful use of these modern classes of materials in technical applications. Typical tests which can be used to evaluate the sheet metal forming capability are hydraulic bulge test [Gol75, Ree95] and cup drawing tests. These testing techniques are important because they can characterise the sheet metal formability and compare it to technologically similar processes like hydroforming, stretching, drawing and etc [Nov03; Hei99]. However, up to now only limited amount of data are available in the literature on the sheet forming capability of the UFG sheet materials, which is mostly due to limited ARB sample size [Lap07a, Top08a]. In the aforementioned field, sheet metal joining is a technologically important issue, which places a further challenge upon UFG aluminium sheets. Originally developed by Thomas et al. [Tho91, Tho97] in 1991, the friction stir welding (FSW) process has become a highly recognised joining technique due to numerous advantages, including higher weight reduction potential compared to standard mechanical fasteners, easier joining of the hard-to-weld materials such as aluminium alloys and a lower heat input compared to fusion welding [Rho97, Cab07, Sua05, Fon04; Daw96]. Friction stir welding is found to be a desirable joining technique for UFG materials, since it provides excellent mechanical properties and retains the fine grained microstructure [Sat04, Sat06, Fuj06]. Furthermore, the ultrafine-grained materials are also known for their enhanced strain rate sensitivity in comparison to the CG materials [May05, Dal04]. An increase in strain rate causes an increase in strength and a decrease in elongation to failure, while a decrease in strain rate results in lower strength and significantly higher elongation to failure (see Fig. 2-2).


7

Figure 2-2: Stress vs. strain for the as-received, cold rolled aluminium and for the UFG aluminium with 5 ARB cycles. The experiments were performed at room temperature [Höp04]. This effect can be favourably used for sheet metal forming where the strain rate sensitivity influences the deformation capability and higher elongation to failure can be obtained at lower strain rates. However, in order to increase the efficiency, it is important to reduce the time of the overall metal forming process. In this case, performing the sheet metal forming of UFG materials at elevated temperatures may be a more desirable method for achieving higher deformation and at the same time better process efficiency. Additionally, higher strain rates are typically associated with crash events and consequently higher energy absorption of a given component. From this point of view, strain rate sensitive materials would also be advantageous, because they show an increase in strength with increasing strain rates. At this stage, the ductility of UFG aluminium sheets is still a limitation, which places another important challenge on ARB processed sheets.

Figure 2-3: View of the A1 1420 alloy samples after tensile tests. Superplastic behaviour of the Al 1420 alloy at very high strain rates, up to 5×10-1 s-1 and at a relatively low temperature of 300°C [Val00].


8

Furthermore, there have also been some reports on low temperature superplasticity of the UFG metal sheets produced by the ARB process [Tsu99b, Tsu99c] and friction stir welding [Ma05]. The superplastic behaviour of UFG sheets (Fig. 2-3) has a potential for more specialised applications and/or hard to form components with complex geometries, due to the excellent ductility which plays a crucial role in metal forming processes. 2.1.1 General Aspects of Severe Plastic Deformation Severe plastic deformation is a promising technique for producing bulk metallic materials with an ultrafine grain structure by introducing very large plastic strain (i.e. ε > 7), while at the same time maintaining the original geometry [Val06, Tsu03a]. The most important strengthening mechanisms, which lead to high strength of materials processed by any SPD technique, are work hardening and grain refinement. These two strengthening mechanisms will be mentioned in the chapter 2.1.2. By employing the conventional processing techniques such as cold rolling, even large percentages of thickness reduction between 60 - 80 % result in relatively large grain sizes of approximately 10 µm. Thus, heavy deformation i.e. very high plastic strain of more than 4 is necessary in order to obtain an UFG microstructure from the conventionally grained materials. As mentioned earlier, the most successful severe plastic deformation processes include Equal Channel Angular Pressing (ECAP), High Pressure Torsion (HPT) and Accumulative Roll Bonding (ARB) used for sheet metal manufacturing. Equal channel angular pressing is a non-continuous process where a sample of circular or rectangular cross section passes through a die having two channels of equal diameters with an intersecting angle of 90°. The material is subjected to high shear strains and little to no change in the cross sectional geometry. The process can therefore be repeated a number of times in order to achieve the UFG microstructure [Val00]. During high pressure torsion a disk like sample is placed between two anvils. The lower anvil rotates and induces high torsional strain on the sample, while the upper anvil applies a pressure of several GPa. Very large plastic strain can be achieved at the periphery of the sample, since no change in dimension occurs during the process. The main disadvantage of the HPT technique is that the sample size is limited to small dimensions and that the samples are characterised by a microstructural gradient from the periphery to the centre [Hor96]. As a result of the fact that most SPD techniques are costly and laborious, have low production efficiency, limited sample size and are discontinuous processes, new trends of bulk ultrafine-grained or nanocrystalline materials processing have emerged. Some of these include the ECAP conform process [Sai00], continuous confined strip shearing [Lee02c], continuous cyclic bending (CCB) [Tak99] and repetitive corrugation and strengthening (RCS) process [Hua01, Zhu01]. Regardless of the type of SPD process used, numerous researchers confirmed that it is possible to obtain a micron or sub-micron grain structure and a good combination of mechanical properties by introducing large strains [Val00]. If the size of these processes can be increased to an industrial scale such that the production of UFG


9

materials becomes commercialised and cost effective, it would be possible to control the mechanical properties of materials by processing rather than by alloying. Thus, the innovation potential of the UFG materials for advanced applications in engineering and medicine is high, and the requirements for producing such materials are becoming more and more economically feasible. 2.1.2 Strengthening Mechanisms In order to produce materials with desirable mechanical properties and considerable strength, various strengthening mechanisms such as solid solution hardening, grain refinement, dislocation hardening (strain/work hardening) and precipitation hardening can be employed. The main mechanisms responsible for material strengthening during SPD processes are dislocation hardening (strain/work hardening) and grain refinement. Each one of these mechanisms is based on different principles and is shortly described in this section. Solid Solution Strengthening Pure metals can be very effectively strengthened by additions of solute atoms into a solid solution of the solvent atom lattice. Solid solution strengthening is based on the interaction of solute atoms with dislocations and the resulting resistance to dislocation motion. Depending on the atomic size of the solute atoms, two types of solid solutions can be differentiated: substitutional (solute and solvent atoms are similar in size and solute atoms occupy the lattice points of the solvent atom) and interstitial (solute atoms are significantly smaller than the solvent atoms and occupy interstitial points of solvent lattice) solid solution. Correlation between the concentration of solute atoms (c) and the contribution to strength increase (ΔσSS) can be determined from the following equation:

nSS c≈Δσ , Eq. 2-1

where n is approximately 0.5 [Rös03]. There are a number of possible ways in which solute atoms can interact with dislocations, including paraelastic (lattice parameter effect), dielastic (shear modulus effect) and chemical interaction. In case of the paraelastic interaction, solute atoms differ in size compared to the matrix atoms. Depending whether they are smaller or bigger than the matrix atoms, they cause either tensile or compressive stresses. The lattice distortion energy caused by the solute atom can be decreased if the solute atom segregates to a dislocation. However, if this dislocation moves, additional forces must be applied in order to overcome the back stress from the solute atom and separate the dislocation from it. Therefore, the critical resolved shear stress of the solid solution is bigger than that of the pure matrix. Dielastic interaction is based on the fact that the solute atoms have a different shear modulus than the matrix, thus increasing the resistance to dislocation motion. Stacking fault energy (SFE) is dependent on the concentration of solute atoms and decreases with an increase in solute concentration. Low SFE leads to bigger dislocation distances and lowers the overall dislocation energy. Therefore, solute atoms favourably segregate to dislocations in order to lower the SFE. If a dislocation moves, the


10

concentration ratio changes and the dislocation energy increases. The dragging force impedes the dislocation motion, leading to material strengthening. This is called the chemical interaction. Precipitation Hardening Second phase particles form if the solubility limit of the solute atoms in the solvent matrix is exceeded. In this way, many alloy systems can be strengthened by precipitation of small second phase particles due to their interaction with dislocations. Precipitation or age hardening can be achieved by solutionising, quenching and annealing of the alloy. During solutionising the second phase particles will dissolve into the metal matrix and during the consequent annealing, they will precipitate out of the solution. The main requirement for the precipitation hardening is that the precipitates must be soluble in the metal matrix at elevated temperature and have a limited solubility at low temperatures. The degree of strengthening depends on the average precipitation size and shape, volume fraction as well as the mean precipitation spacing. They also have a certain mismatch together with the surrounding matrix. If the particles are coherent, the lattice planes of the matrix and the particles coincide with one another and the particles can be cut by dislocations. The lattice planes of the incoherent particles however mismatch those of the matrix, causing stresses at the particle-matrix interface. Since the dislocation can usually not cut this type of particles, it moves around them. In both cases, additional force needs to be applied for the dislocations to overcome the particles. The cutting of coherent particles by the dislocations requires the formation of new interfaces, while the movement of dislocations around the incoherent particles leaves a dislocation ring, which creates a stress field making it more difficult for the next dislocation to move past. The shear stress (Δτc) for cutting the particles is proportional to the square root of the particle radius (r) and volume fraction (ƒ):

ƒrc ⋅∝τΔ , Eq. 2-2 The Orowan stress for circumventing of particles can be expressed as follows:

rƒbG

OR ⋅π⋅

=τΔ , Eq. 2-3

where G is the shear modulus and b the Burgers vector [Vol89]. The optimal precipitate size can be obtained when the shear stress for particle cutting and Orowan stress for particle circumventing equal each other. It is well known that the second phase particles such as the Mg2Si particles present in the aluminium alloy AA6016, investigated in this study, affect the natural aging behaviour of the material, as well as its microstructure and the mechanical properties. The great increase in strength of AA6016 is caused by approximately 1 wt% of magnesium and silicon that precipitate out of the solution and form Mg2Si precipitates [Aut08]. From the silicon and magnesium cluster, the coherent spherical GPI zones of 1-3 nm are formed. With time diffusional processes enable the transformation to semi-


11

coherent GPII zones or β˝ needles approximately 50 nm in size and then to hexagonal β’ rods of 500 nm. Finally, these transform into the equilibrium phase β - Mg2Si plates of a cubic crystal structure and a few microns in size. Therefore, the precipitation sequence is as follows: Mg, Si clusters → GPI zones→ GPII (β”) → β’→ β (Mg2Si). The transition from incoherent β” to coherent β’ precipitation type has the highest strengthening effect [And98, Dut91]. Thereafter, the precipitates overage and the strengthening effect is lost. Strain Hardening Strain or work hardening is an important technique to strengthen the metallic materials using commercial industrial processes like cold rolling, extrusion, drawing and etc. During all these processes the materials are plastically deformed into a desired shape and at the same time strengthened. The principle of strain hardening is based on dislocation multiplication, dislocation motion and mutual dislocation interaction. During strain hardening, the dislocation sources (Frank-Read sources) become activated and the number of dislocations increases. Since the dislocations incorporate strain fields due to lattice distortion, the dislocation strain fields can interact with each other if they are within proximity, making it more difficult for the next dislocation to move through the strain field. Therefore, by increasing the amount of plastic deformation, the number of dislocations rises, leading to a significant increase of the yield strength. The increase in yield strength (ΔσSH) is proportional to the dislocation density (ρ):

ρ⋅⋅⋅⋅=σΔ bGM)2.0...1.0(SH , Eq. 2-4 where M represents the Taylor factor, G the shear modulus and b the Burgers vector [Rös03]. One of the disadvantages of this strengthening method is that it leads to lower elongation to failure and that its strengthening effect can be lost if a high temperature process such as welding is subsequently conducted. Grain Boundary Strengthening It is well known that the mechanical properties of metallic materials are strongly affected by the grain size. The increase in yield strength with a decrease in grain size was found to be applicable to a wide variety of metals and obeys the Hall-Petch relationship:

21

0

−⋅+= dky σσ ,

Eq. 2-5

where σy is the yield strength, σ0 the intrinsic yield stress, k is a constant for a given material and d grain size [Hal51]. If a dislocation source within one grain produces dislocations which pile-up at the next grain boundary, the stress at the tip of the pile-up must exceed some critical stress in order for the slip to continue into the neighbouring grain. Therefore, grain boundaries act as efficient barriers to dislocation motion. Because of this it is often desirable to produce materials with small grain size, due to a


12

consequently higher number of grain boundaries acting as obstacles to dislocation glide. 2.2 Accumulative Roll Bonding (ARB) 2.2.1 Conventional Sheet Metal Rolling In order to understand the main differences between the conventional sheet metal rolling and the accumulative roll bonding process used in this work for ultrafine-grained metal sheet processing, the most important fundaments regarding conventional rolling are shortly described in this chapter. The process of plastically deforming a metal slab or sheet by passing it between rolls is known as rolling [Die88]. During the process the metal sheet is exposed to high compressive stresses due to squeezing action of the rolls and high surface shearing stresses arising from friction between the sheet and the rolls. The main objective of rolling is to reduce the cross-section of the incoming material as well as obtain good mechanical properties of the outgoing product.

Figure 2-4: Sketch of a conventional rolling process after G. E. Dieter [Die88] During rolling the metal sheet with an initial thickness t0 enters the rolls with an initial velocity v0, which is smaller than the constant surface velocity of the rolls. The metal sheet is pulled through the rolls by the surface frictional force of the rolls and increases its speed throughout the deformation zone. At some point within the deformation zone, the metal sheet speed reaches the roll velocity. This point is called the neutral point. On the exit side of the neutral point, the sheet moves faster than the rolls and the direction of frictional force is reversed and now acts as to oppose the movement of the


13

material flow. Figure 2-4 illustrates some of the most important geometrical parameters such as: t0: initial sheet thickness R: roll radius tf: final sheet thickness ω: rotational roll speed v0: initial sheet velocity α: angle of contact vf: final sheet velocity Lp: projected length of contact (deformation

zone) b: sheet width.

The fundamental terms and parameters such as the thickness reduction, amount of deformation in terms of strain, the sheet width increase, velocity of the sheet and the length of the deformation zone, can all be takes out of the following few equations: Thickness reduction f0 ttt −=Δ Eq. 2-6

Deformation strain 0

h ttΔ

=ε Eq. 2-7

If the metal sheets are relatively wide, it can be assumed that there is only an increase in length and no increase in the width of a sheet during rolling [Lan74]. The increase in width can therefore be considered as negligible when the width is much bigger than the length of the deformation zone, as seen in equation 2-8. 20L

bp≥ Eq. 2-8

The velocity of the sheet can be derived from the equal volume of material entering and leaving the rolls, assuming that the width is equal before and after rolling. Constant material volume fff btvbtv ⋅⋅=⋅⋅ 000 Eq. 2-9 Equal width fbb =0 Eq. 2-10

Velocity of the sheet )x(ttv)x(v 0

0 ⋅= Eq. 2-11

The length of the deformation zone (projected length of arc of contact) can be calculated from a simple relation between the roll radius and the initial and final sheet thickness. Deformation zone [ ] 2

1f0p )tt(RL −⋅= Eq. 2-12


14

Before the sheet can be rolled, there are two conditions, which have to be fulfilled. The rolls must be able to grip the incoming sheet and than draw it through the rolls. The two relationships can be derived from figure 2-5. Figure 2-5: Sketch of forces during conventional rolling after [Gei01] Sheet can be pulled in when 12 FF ≥ Eq. 2-13where αsin1 ⋅= NFF Eq. 2-14and αμα coscos2 ⋅⋅=⋅= NR FFF Eq. 2-15Therefore: αμ tan≥ Eq. 2-16Grip condition: αμ ≥ Eq. 2-17Pull through condition: αμ ≥⋅2 Eq. 2-18

There are a number of classifications of rolling one of which is the rolling temperature. Hot rolling is generally performed above the recrystallisation temperature of the metal and therefore requires low flow stresses. Due to the increased ductility of the sheet at higher temperatures, high deformation or thickness reduction can be achieved within one pass. The most important feature for the microstructural development is that the material simultaneously strain hardens and is at the same time softened or recovered. Cold rolling on the other hand is usually conducted at room temperature after hot rolling. The material strain hardens significantly and higher stress is required for the deformation. The advantages of cold rolling include better thickness tolerances and surface finish, no oxide layer formation and a possibility to achieve thinner sheets. A number of defects in rolled products may arise due to a complex interaction between the rolls and the metal sheet. Some of these include: sheet bending in the rolling plane, non-uniform thickness in width or length, waviness, edge cracking and etc. Generally, problems with shape and flatness take place due to inhomogeneities in deformation in the rolling direction. If the roll gap is not perfectly parallel, one side of the sheet is deformed and elongated more than the other, resulting in sheet bowing as well as non-uniform thickness. In case of roll deflection, the edges of the sheet experience higher


15

compressive stresses in the rolling direction and the centre portion of the sheet is stretched in tension. The consequence of this stress state is edge waviness. a) b) c)

Figure 2-6: Defects which may occur as a result of lateral spread during rolling [Die88] Another commonly encountered problem is edge cracking of sheets. There is a natural tendency of sheets to expand in the transverse direction during rolling, but the material flow is opposed by the frictional forces. The frictional forces are higher at the centre and most of the material flow occurs in the rolling direction, while the edges can also spread laterally. Due to the material continuity, the centre of the material is under compression and the edges under tension (Fig. 2-6 a), leading to edge cracking (Fig. 2-6 b). In extreme cases of strain distribution, the sheets may even split in the centre parallel to the rolling direction (Fig. 2-6 c) [Die88].


16

2.2.2 Accumulative Roll Bonding Process Accumulative roll bonding is a relatively new severe plastic deformation (SPD) process, which was originally introduced and developed by Saito et al. in 1998 [Sai98]. The ARB process shown in figure 2-7 involves wire brushing of metal sheet surfaces in order to remove the oxide layer, stacking of two sheets on top of each other and roll bonding them together. The two sheets are generally rolled to 50 % thickness reduction and therefore leave the rolls with the original sheet thickness. During rolling the two metal sheets join together to form a solid body and can once again be halved, wire brushed and roll bonded. The process can be repeated many number of times. In most cases, the process is repeated up to 10 times.

Figure 2-7: Schematic illustration showing the principle of accumulative roll bonding (ARB) process after Saito et al. [Sai98]. The metal sheet deformation takes place predominantly in the rolling direction. Any increase in sheet width can be neglected since the sheets are wide enough and the broadening is prevented by high frictional resistance. This can be described by equation 2-8. During roll bonding the number of individual layers within 1 mm thick sheet increases according to the power law equation (equation 2-19) and the thickness of the individual layers can be calculated from equation 2-20,

Nn 2= Eq. 2-19


17

N

tt2

0= Eq. 2-20

where n = number of individual layers N = number of ARB cycles t0 = initial sheet thickness t = final thickness of the individual layer. Therefore, the total reduction rtot after N cycles can be calculated from equation 2-21.

Ntot ttr

2111

0

−=−= Eq. 2-21

After 10 ARB cycles with 50 % thickness reduction per cycle, the 1 mm thick sheet develops 1024 individual layers, each having a theoretical thickness of 1 µm, while the total reduction approaches 100 %. Assuming von Mises yield criterion and plane strain condition i.e. no lateral spreading, the total equivalent strain after N cycles can be calculated according to equation 2-22 [Tsu03a].

N80.0N21ln

32

tot ⋅=⋅⎭⎬⎫

⎩⎨⎧

⎟⎠⎞

⎜⎝⎛⋅=ε Eq. 2-22

Figures 2-8 a) and b) summarise the relevant geometrical changes taking place during accumulative roll bonding of two 1 mm thick sheets, with a thickness reduction of 50 % per cycle. The graphs were derived from equations 2-19 to 2-22 [Tsu03a]. Using the ARB process it was possible to produce a wide range of different UFG materials including commercial purity aluminium, aluminium alloys, copper, titanium and interstitial free (IF) steels [Sai98, Kim06, Lee02b, Top07, Tak07, Ter07,Tsu02a]. However, most authors concentrate on the microstructural evolution and the mechanical properties, while the actual ARB process development and/or improvement are usually not discussed.


18

a)

b)

Figure 2-8: Geometrical changes of the materials during ARB where 2 metals sheets 1 mm thick are roll bonded by 50 % thickness reduction per cycle [Tsu03a].


19

2.2.3 Microstructural Evolution of Accumulative Roll Bonded Metal Sheets

The following process parameters may be considered as relevant for the microstructural evolution during rolling and the resulting mechanical properties of the ultrafine-grained metal sheets. The most important ones will be discussed in this section with reference to various materials. Formation of ultrafine-grained microstructure during ARB The ultrafine-grained microstructure develops after repeated rolling deformation of a conventionally grain sized metallic material. Irrespective of the type of material being rolled, the grain size reaches a submicron scale, ranging from 100 nm to 1000 nm. During the ARB process, the grains become strongly elongated in the rolling direction. This was reported by Tsuji et al. [Tsu02a] for ultra-low-carbon interstitial free (IF) steel rolled at 500 °C as well as aluminium alloys [Hua03]. The microstructure has elongated grains or lamellar boundaries, which are typical for heavily rolled materials [Han99]. It consists of a high fraction of high angle grain boundaries (15 - 20 °), confirmed using an Electron Back Scatter Diffraction (EBSD) technique [Tsu02a] and dislocation substructures inside the grains. Even though the exact grain refining mechanisms during the ARB process are still not completely understood, there have been some reports on grain subdivision and continuous (or in-situ) recrystallisation. Based on the findings that heavily cold worked metals are subdivided by grain boundaries and dislocation boundaries which are arranged in a lamellar or subgrain structure, Hughes and Hansen [Hug97, Ros95] suggested a possible mechanism for high angle boundary formation involving grain subdivision (Fig. 2-9). Initially, at low strains, the dislocations may be freely standing and may form tangles or cells. The cells can be separated by dislocation boundaries such as dense dislocation walls (DDWs), microbands and lamellar boundaries, as indicated in figure 2-9 a. The subgrain structure develops within the initial HAGBs. By increasing the strain level, the distance between the HAGBs decreases in the normal direction, i.e. they are pushed together and they are separated by fewer cells/subgrains. There is an increasing tendency for the dislocations to be stored within the subgrains, thus subdividing the microstructure. The subgrain size decreases to submicrometer range and the dislocation boundary misorientation is widely distributed. Eventually, the microstructure evolves into a lamellar structure sandwiching in small dislocation boundaries i.e. cells/subgrains, and is subdivided into high and low angle grain boundaries (Fig. 2-9 b). The grain misorientation increases and as a consequence the number of HAGBs increases, indicating that the grain size reduction was induced by high plastic deformation [Hug97].


20

a) b)

Figure 2-9: Schematic drawing of deformation microstructures and grain subdivision for a) small strains (εvM = 0.06 - 0.8) with long microbands and dense dislocation walls (DDW) surrounding groups of cells in a cell block and b) large strains (εvM > 1) with lamellar boundaries (LBs) parallel to the deformation direction, sandwiching in narrow slabs of cells or equiaxed subgrains [Hug97]. Another theory regarding the development of ultrafine-grained HAGB is based on recrystallisation processes rather than the previously proposed grain subdivision and increase in grain misorientation. Humphreys and Hatherly [Hum95] suggest that the ultrafine-grained HAGB microstructure forms by the mechanism of continuous recrystallisation rather than the conventional discontinuous recrystallisation. During conventional discontinuous recrystallisation a small grain size can be achieved by deformation to large strains and subsequent annealing to stimulate recrystallisation [Hum95]. However, if a metal undergoes large strain at intermediate or high temperatures, a microstructure containing predominantly high angle grain boundaries may also evolve with little or no further annealing. At large strains the HAGBs are pushed together and they are separated by subgrains. Once the separation of HAGB equals the subgrain size, the HAGB impinge upon one another and a microstructure of small and equiaxed grains is formed [Hug97]. The continuous recrystallisation usually occurs after annealing of heavily deformed materials or in the case where the grain boundary migration is strongly pinned and impeded by the second phase particles. This leads to strongly recovered microstructure of low angle grain boundaries (LAGB) as well as high angle grain boundaries (HAGB). In this case the new microstructure forms without the migration of HAGB and the process is called continuous or in-situ recrystallisation [Got98]. Taking both theories into account it can be said that the evolution of ultrafine-grained HAGBs in heavily deformed materials is still a strongly disputed topic and does not occur in a conventional nucleation and growth manner. As a consequence, the mechanism of HAGB evolution in strongly deformed materials remains to be clarified.


21

Influence of Second Phase Particles Until now, the influence of second phase particles on the microstructural evolution during accumulative roll bonding has not been investigated in detail. Authors Pérez-Prado et al. [Per04] and Slamova et al. [Sla07] investigated the influence of precipitates in the accumulative roll bonded magnesium alloy Mg-Al-Zn and the accumulative roll bonded aluminium alloy AA8006 (Al-Fe-Mn-Si), respectively. Both authors reported that the small particles exert substantial pinning on grain boundaries, stabilise the deformed structure and contribute to material strengthening. Conventional vs. ARB Rolling The conventional rolling and the accumulative roll bonding process are very similar, but due to the fact that considerably smaller grain sizes can be achieved by the ARB process, it is worth pointing out the main differences. The most important characteristics of the ARB process include:

• higher percentage reduction per rolling cycles • smaller achievable grain size • additional shear stresses at the surface and complex shear strain distribution • higher friction of high angle grain boundaries (HAGB) and • grain misorientation.

In the conventional industrial rolling process, the total thickness reduction varies between 60 - 80 % (1.06 - 1.86 von Mises true strain). After subsequent annealing or recrystallisation, the final average grain size reaches approximately 10 µm [Tsu03a]. Contrary to the conventional rolling, the ARB process accumulates large plastic strains during each cycle and averages 50 %, while the total percentage reduction can reach over 90 % (see Fig. 2-8 b). Another significant difference is the effect of friction. In a conventional rolling process, it is generally aimed to reduce the amount of friction between the metal and the rolls due to the energy considerations. However, the presence of friction in the ARB process was found to be advantageous and desirable, due to the additional shear deformation introduced in the subsurface region during roll bonding without lubrication, leading to faster grain refinement [Li06a, May04]. It also results in a rapid build up of an oxide layer on the rolls, which was found to increase the hardness of the roll bonded sheets [May04]. Lee et al. [Lee02a] investigated the change in shear strain across the thickness of the sheet of the commercial purity aluminium AA1100 by embedding a pin and measuring its inclination after every cycle (Fig. 2-10). During the first few ARB cycles, the distribution of shear strain through the thickness of the sheet is inhomogeneous due to repeated stacking and rolling of the two sheets. The surfaces with the highest shear strain migrate to the middle during the second cycle. Depending on the location and the number of ARB cycles, a complex combination of plane strain and shear deformation can develop and affect the final texture of the sheets. After performing more than 8 cycles, the shear strain distribution as well as the grain size become more homogeneous.


22

Figure 2-10: Distribution of shear strain through thickness of ARB processed AA1100 after a) one, b) two, c) four and d) eight cycles [Lee02a]. The grain misorientation of high purity cold rolled aluminium and nickel processed by conventional rolling to 20 % reduction was measured by Hughes [Hug03], who observed that significant angles of misorientation were already present at relatively low strains. At higher strain levels the dislocations form lamellar structures and develop into HAGBs. Similar observations regarding the UFG microstructural evolution and grain misorientation were also reported by Huang et al. [Hua03] for the commercial purity aluminium AA1100 processed by accumulative roll bonding. From figure 2-11 a), it can be seen that the boundary spacing in terms of lamellar boundaries and interconnecting boundaries (cells) decrease with an increase in von Mises strain or deformation. At higher strains the lamellar structure is dominated by high angle boundaries. The misorientation angle increases with strain in the low strain range, but reaches a saturation value of about 36° at a strain of 3.2. During cold rolling, no saturation was observed (Fig. 2-11 b). Therefore, the ARB process seems to be a more effective grain refining method than the conventional rolling process for the same percentage of reduction.

a) b)

Figure 2-11: a) Boundary spacing as a function of von Mises strain and b) mean misorientation angle across the lamellar boundaries as a function of von Mises strain, for aluminium processed by the ARB technique (triangles) and by the conventional cold rolling [Hua03].


23

As described earlier, the ARB processed materials have a strongly elongated microstructure with lamellar boundaries in the rolling direction and short, transverse boundaries inter-connecting the lamellar boundaries [Hua03], while the fraction of high angle boundaries is much higher than in the conventionally heavily deformed materials [Han99]. However, the morphology and the subgrain or grain size may differ depending on the type of material being processed. Li et al. investigated four different materials, namely the commercial purity aluminium AA1100, oxygen free high conductivity (OFHC) copper, austenitic steel and ultra-low carbon interstitial free (IF) ferritic steel [Li06a]. OFHC-Cu showed a coarser grain structure and a significantly smaller fraction of HAGBs in comparison to other materials that showed a fraction of HAGBs of more than > 70 %. This was attributed to the recovery and recrystallisation, which took place during the accumulative roll bonding and retarded the formation of the high angle grain boundaries. The new, strain free grains were easily formed due to the high accumulated strain, adiabatic heating and the high purity of the material. Similar observation were also reported in another publication of Li. et al. [Li06b] On the other hand, accumulative roll bonded austenitic steel showed smaller lamellar boundary spacing and straight lamellar boundaries, indicating that recovery is more difficult in this case due to higher melting point, larger amount of alloying elements and medium stacking fault energy. Solid State Bonding of Metals Solid state bonding of metals can be divided in two types, namely diffusion bonding and mechanical bonding. Diffusion bonding is a joining process where the principal mechanism is interdiffusion of atoms across the interface. It requires considerable amount of time and the application of temperature and pressure. Mechanical bonding occurs instantaneously or over a very short time and depends among other factors on the forces of attraction between the atoms [Wu98]. Roll bonding is a mechanical solid state bonding process, based on cold welding and it is important to emphasise some of its features and link its importance to the overall process of accumulative roll bonding. Laminates or accumulative roll bonded metal sheets strongly rely on good lamellar bonding for successful sheet metal forming. The shear strength of the lamellar bond determines the mechanical performance especially in processes where bending takes place, such as deep drawing or metal sheet stretching. In 1950 F. P. Bowden and D. Tabor [Bow50] suggested that cold welding process depends on adhesion caused by intermolecular forces between the interfaces of materials. The main requirement for good adhesion is the contamination free surfaces at an interatomic distance to each other. This type of cold welding based on adhesion of the metal sheets also takes place during the ARB process. The interlamellar bonding (sheet metal bonding during ARB) is one of the most important microstructural features affecting the mechanical properties of the final accumulative roll bonded metal sheets, which has been the least investigated. Until now numerous factors which influence the success of the process and the quality of the bond have been considered; these include: oxide layer, surface roughness, normal pressure, percentage reduction, process temperature, annealing temperature and rolling speed. These parameters are explained more thoroughly in the following chapter.


24

2.2.4 Accumulative Roll Bonding Parameters Various parameters influence the success of the ARB process, the quality of the bond as well as the microstructure and the mechanical properties of the ARB processed sheets. The important ARB process parameters include: process temperature, percentage reduction per cycle, number of ARB cycles, roll diameter and roll speed, physical properties of the material (e.g. crystal structure), initial material condition before rolling (e.g. solution treatment), surface roughness of the sheets as well as the oxide layer. Process Temperature Process temperature is one of the most important factors, which influences the microstructure, thermal stability, mechanical properties as well as the quality of the metal sheet bonding. An optimal process temperature usually has to be determined for every material. It is a compromise between good bonding between the sheets and a good thermal stability. This was confirmed by Slamova et al., who investigated thermal stability of the accumulative roll bonded twin-roll cast Al–Fe–Mn–Si sheets between 200 °C and 450 °C [Sla07, Hom06]. The highest hardness was reached when the rolling was performed at the lowest temperature, although better bonding was obtained at temperatures higher than 200 °C. At higher temperatures work hardening takes place only during the first cycle and the material softens after the subsequent cycles. The reasons for softening include dynamic recovery and/or partial recrystallisation during pre-heating and rolling. Higher temperatures also decrease the potential for a rapid grain refinement. Thermal stability of the ARB processed sheets was found to be higher than for the as-received cold rolled sheets. The cold rolled and the ARB specimens were fully softened at 300 °C and 350 °C, respectively, while the differences in the annealing response results from the different amount of stored energy in both materials. Authors Yan and Lenard [Yan04] have conducted a number of experiments on the aluminium alloy AA6111 investigating the effect of temperature on the shear strength of the metal sheet bonding (Fig 2-12 a). The entry temperature (or process temperature) affects bonding by influencing the adhesion process and the break-up of the oxide layer. At higher temperatures the strength of the oxide layer of the parent metal, and the bond between the oxide and the strip decrease. This can enhance the possibility of two surfaces conforming to each another, leading to higher bond strength. Heat treatment performed prior to rolling (here referred to as annealing temperature) also showed a significant influence on the strength of the bond (Fig. 2-12 b). Annealing at elevated temperate generally created the strongest bonds, since the previous cold work was eliminated and the material softened. The deformation of the surface asperities became easier and the true contact area larger, collectively leading to stronger interlamellar bonds [Yan04].


25

a) b)

Figure 2-12: Shear strength of the interlamellar bond of the AA6111 as a function of a) entry temperature and b) annealing temperature and the percentage reduction [Yan04]. Percentage Reduction per Rolling Cycle Generally, 50 % thickness reduction per cycles (von Mises equivalent strain, εvM=0.8) is necessary for a good quality bond of the aluminium and aluminium alloy sheets, as well as other materials such as titanium, interstitial free steel, or even aluminium-magnesium composites [Hua03, Ter07, Cos05, Che06a]. Kralliks et al. [Kra04] found that roll bonding below 50 % thickness reduction results in low bonding strength or unsuccessful bonding between the sheets and that bonding above 50 % leads to edge cracking. Thus, the optimal thickness reduction seems to be a compromise between the good quality of the bond and the minimal crack development at the edges. Until now, the effect of the degree of reduction on the microstructure was not investigated. Yan and Lenard [Yan04] observed that the interlamellar bond strength increases with a higher percentage of reduction for the aluminium alloy AA6111 (Fig. 2-13 a). Higher reduction and lower roll speed can be sufficient to create a bond whose shear strength is near 230 MPa, approximately equivalent to the shear strength of the original material. Therefore, the time of contact also seems to play a very important role, since adhesion and perhaps some limited amount of diffusion appear to be the dominating interlamellar bonding mechanisms.


26

a) b)

Figure 2-13: Shear strength of interlamellar bonding of the AA6111 as a function of a) rolling speed and percentage reduction and b) roll pressure [Yan04]. Number of Rolling Cycles Increasing number of ARB cycles results in the hardness, yield and tensile strength increase. The strengthening mechanisms active in aluminium alloys include: solid solution strengthening, precipitation srengtening, grain boundary strengthening and work hardening. During the fist few cycles, the dominating strengthening mechanism is work hardening. Later, the contribution of grain boundary strengthening becomes a more dominating strengthening mechanism due to the formation UFG microstructure. Krallics and Lenard [Kra04] roll bonded ultra low carbon steel containing 0.002 % C with a surface roughness between 1.5 – 1.8 µm Ra achieved by wire brushing. They performed three point bending tests in order to observe the multi layered strip behaviour in a potential sheet forming operation. As the number of bonded layers increased fewer cases of delamination were observed, indicating that the bond strength increases after repeated rolling. Roll Diameter Roll diameter directly influences the angular velocity of the rolls and the deformation area of the sheets [Top07]. Smaller roll diameters lead to higher rolling speed, smaller deformation area and therefore higher roll pressures. The effect of rolling speed has also been investigated with regard to the sheet metal bonding. From figure 2-13 a), it can be seen that the strength of the bond decreases by increasing the rolling speed. This is due to the faster strain hardening of the surface asperities and shorter contact times between the asperities of the two surfaces to be joined [Yan04]. Tsuji et al. deformed ultrafine-grained aluminium sheets at various strain rates ranging from 2.0×10-4 s-1 to 6.0×10-4 s-1 by conventional rolling, ultra-high-speed rolling, and impact compression. Initially, it was expected that the higher strain


27

rate would result in a more effective grain subdivision process due to the higher dislocation density. However, irrespective of the deformation mode used, the grain size increased with an increase in strain rate. This was a consequence of the large amount of heat generation during the process. Authors believe that the grain size increased by a mechanism of short-range grain boundary migration, which was attributed to the heat generation during plastic deformation. On the other hand, it was also suggested that the high speed plastic deformation is potentially a more effective grain subdivision process if the effect of heat was to be minimised [Tsu03b]. The effect of applied roll pressure is similar to the degree of reduction. Higher roll pressures generally lead to better interlamellar bonding between the sheets (Fig. 2-13 b). It is hypothesized that the shear strength of the bond approaches that of the parent metal when the sufficient amount of energy – the activation energy to initiate the bonding process – is provided for the two components to be joined. This energy may be provided by heating and/or by mechanical means [Yan04]. Surface Roughness By wire brushing the surfaces before stacking and rolling, plastic deformation as well as surface roughness is introduced. Sufficient surface roughness is another parameter influencing the strength of the interlamellar bonds, although it is generally a difficult parameter to quantify and interpret. A direct effect of surface roughness on the interlamellar bond strength has still not been largely investigated. In the Diploma Thesis, Klösters [Klö06] compares the surface roughness values of the as-received aluminium sheets with the ones which have been treated using different surface techniques, including: wire brushing, polishing and sand blasting (Fig. 2-14). The as-received material has the smoothest and the wire brushed sheets the roughest surfaces. The roughness values of surfaces, which have been mechanically polished or sand blasted lie in the middle of the as-received and the wire brushed sheets. It was shown that the highest surface roughness achieved by wire brushing provided the strongest interlamellar bonding between the metal sheets (Fig. 2-15 a). In order to correlate the interlamellar strength with the surface roughness values, a number of peel tests were carried out. The ARB sheets were manufactured such that one part of the sheet was able to be pulled apart from the other. The two sheets were fixed by a pin and one half of the sheet was pulled over a rotating roll by the cross head of the tensile testing machine. The test set-up can be seen in figure 2-15 b.


28

Figure 2-14: Surface roughness after different surface treatments [Klö06]; (*Rz: average roughness i.e. arithmetical mean of five values; Ra: arithmetical mean from the measured average profile). The roughest wire brushed surface provided the highest interlamellar bond strength, while the polished surfaces with the lowest surface roughness resulted in poor bond strength. Even though this was not a standardised testing technique, it can be concluded, that there is a definite correlation between the surface roughness and the quality of the metal sheet bonding. a) b)

Figure 2-15: Force-distance curves measured during the peel tests of samples with different surface treatments (the force was measured over the width of the sample) and b) sketch of the peel test set-up [Klö06].


29

Oxide Layer Researchers have recognised that the removal of the oxide layer before roll bonding may be important for good interlamellar bond strength between the sheets and the oxide particles may act as obstacles to grain growth [Sai99]. Early work of Milner and Rowe [Mil62, Le04] showed that aluminium oxide breaks up into small particles as it is elongated during conventional cold rolling and leaves behind a newly generated metal surface. The break up of the oxide layer was later also confirmed by Barlow et al. [Bar04] for the roll bonding process of aluminium foils. Barlow et al. investigated the surfaces of aluminium AA1200 (commercial purity) and found that the oxide particles average up to 10 nm in thickness and 60-100 nm in length. They are aligned parallel to the rolling direction and form stringers outlining the original position of the foil surfaces. Authors believe that the aluminium oxide particles lead to dispersion strengthening and increase the material flow stress by approximately 20 MPa [Bar04]. However, the dominant fraction of material strengthening comes from the fine grain structure. Authors also suggest that the presence of fine oxide particles may stabilise the microstructure, and increase thermal stability. At this point, it MUST be emphasised that the aluminium foils processed in the work of Barlow et al. have not been surface treated, as it is usually the case for the ARB process. During ARB, the oxide layer is generally removed mechanically using a rough wire brush, leaving behind only a very thin oxide layer. The removal of an oxide layer is especially difficult for aluminium alloys because of the high aluminium affinity to oxygen and the very rapidly forming aluminium oxide layer when exposed to air. Its thickness also varies with the environmental temperature. Based on the results of Barlow et al. it can be stated that the thin oxide layer which is still present after wire brushing, will be broken up during ARB processing. Friction The success of metal rolling also depends strongly on the friction between the rolls and the metal sheet. In order for the sheet to be taken in by the rolls as well as pulled through the rolls, equations 2-17 and 2-18 have to be obeyed. Although the influence of friction with respect to the ARB process was not investigated or reported in any publications until now, it was shown in the Diploma Thesis of J. May [May04] that the amount of oxide layer on the rolls and therefore the friction between the rolls and the metal sheet, can affect the final mechanical properties of the ARB processed material.


30

2.2.5 Mechanical Properties of Accumulative Roll Bonded Materials It is well known that the reduction of grain size increases the material strength. This concept was initially introduced by Hall [Hal51] and Petch [Pet53], who developed the so called Hall-Petch relationship between the grain size and strength (see Eq. 2-5). Based on this idea, the aim to produce nanocrystalline and ultrafine-grained materials has drastically increased over the last 20 years [Gle89, Koc99]. Ultrafine-grained materials have especially enhanced strength in comparison to the conventionally grain sized materials, which was confirmed by many researchers [Low00, Fur01]. In some cases the ductility as well as the strength can be increased. This was observed by Valiev et al. on nanocrystalline titanium and copper processed by ECAP, and reported as the “Paradox of strength and ductility” [Val02]. A similar observation was also reported by Hoeppel et al. [Höp04] who processed commercial purity aluminium AA1050 using accumulative roll bonding (Fig. 2-16). It can clearly be seen that it is possible to obtain a consecutive increase of elongation to failure and strength with increasing number of ARB cycles compared to the cold rolled, conventionally grained state.

Figure 2-16: Stress vs. strain for UFG aluminium at different numbers N of ARB cycles compared to the cold rolled state with approximately 3 µm grain size. Tensile tests were performed at 1·10-4 s-1 and room temperature [Höp04].


31

Most researchers, who investigated the manufacturing of UFG aluminium and aluminium alloys using the ARB method, generally observed the increase in hardness and yield or tensile strength with increasing number of ARB cycles. However, the previously motioned “Paradox of strength and ductility” was not always as pronounced as in case of [Höp04]. A typical curve of hardness vs. strain can be seen in figure 2-17 a) for the commercial purity aluminium AA1100 and aluminium alloy AA8011 processed at 200 °C. After 3 - 4 ARB cycles, the hardness generally increases by up to 2-3 times in comparison to the as-received material. Thereafter, the hardness reaches a saturation level and does not increase any further [Xin01, Par01, Höp04]. However, depending on the type of alloy and processing parameters such as temperature, roll speed or roll diameter, it may come to softening after the maximum hardness level has been reached (Fig. 2-17 a). Xing et al. [Xin01] argue that second phase particles e.g. MgsSi, common in hardenable aluminium alloys increase the hardness by means of precipitation strengthening, but decrease the solid solution strengthening contribution, which eventually leads to softening.

a) b)

Figure 2-17: Variation of hardness with strain of a) AA1100 commercial purity aluminium and AA8011 aluminium alloys processed at 200°C [Xin01] and b) mechanical properties of AA1100 and AA6061 roll bonded at 200 °C and room temperature, respectively [Lee02b]. An increase in the yield and tensile strength with an increase in number of ARB cycles or strain level was observed for most ARB produced aluminium alloys. A typical example can be seen in figure 2-17 b) for the commercial purity aluminium AA1100 [Sai99] and the AA6061 aluminium alloy [Lee02b], processed at 200 °C and room temperature, respectively. Tensile strength increased by a factor of 3 to 3.5 in both cases. Similar results were also reported by various other researchers [Xin01, Sla06a]. The mechanical properties of nanostructured copper produced by the ARB process were investigated by Takata et al. [Tak07]. The tensile strength of samples processed up to 8 ARB cycles (the equivalent strain of 6.4) reached 520 MPa, which was about


32

three times higher than that of the as-received, conventionally grained material. On the other hand, the total elongation decreased significantly after the first ARB cycle and averaged only 3 %. However, the total elongation increased again with increasing number of the ARB cycles and reached 10 % after 8 ARB cycles. The microstructure and mechanical properties of commercial purity titanium severely deformed by the ARB process was investigated by Terada et al [Ter07]. In this case, it was observed that with increasing number of ARB cycles, the tensile strength continuously increased, while the total elongation to failure decreased. After 6 ARB cycles, the tensile strength increased by a factor of 2.2 in comparison to the as-received state and the total elongation to failure decreased by a factor of 3. It was interesting to see that the tensile strength of commercial purity titanium after 6 ARB cycles corresponds to that of titanium alloy Ti-6Al-4V. In spite of the excellent strength, ARB aluminium sheets as well as other nanostructured or UFG materials have lower ductility compared to the as-received, usually annealed materials. In most cases their ductility is limited to approximately 5 % to 10 % [Sai99, Lee02b, Xin01]. There are two major sources of limited ductility in nanocrystalline materials, namely: artefacts from processing like pores and low work hardening. The absence of work hardening causes localised deformation leading to early tensile instability and thus low ductility. As the grain size is reduced to nanoscale, the dislocation mean free path and the number of dislocations that contribute to work hardening is strongly reduced and the traditional concept of pile-up at grain boundaries becomes very limited or it does not occur at all. The dislocations are produced during deformation, but they can also be stored or annihilated at the grain boundaries. The grain boundaries can therefore act as sources as well as sinks for dislocations. Thus, room temperature dynamic recovery becomes very common in nanocrystalline/ultrafine-grained materials and the competition between dislocation generation during plastic deformation and dislocation annihilation during recovery determines the steady state dislocation density or grain size [Mey06, Blu04]. This is usually reflected in the mechanical properties of these materials, namely at some point there is a saturation in hardness and strength. In order to qualify the nanocrystalline or ultrafine-grained materials for technical applications, increasing the ductility is crucial, since it governs the material formability. This is especially important for components with complex geometries. There are a number of methods currently investigated, which can be used to increase the ductility of ultrafine-grained materials processed by the SPD techniques. These methods include: post process annealing, introducing a bimodal microstructure, processing at cryogenic temperatures or using the effect of enhanced strain rate sensitivity generally found in most UFG materials. Kamikawa et al. [Kam06] annealed the accumulative roll bonded high purity aluminium (Al 99.99 %) and confirmed that a two-step annealing process can effectively produce uniform, equiaxed grain structure throughout the thickness of a sample, previously deformed to large plastic strain. The annealed samples contained predominantly high angle grain boundaries, but regions with a high concentration of


33

low angle grain boundaries were also observed. The mechanical properties of the annealed samples were not mentioned. Some studies indicated that a duplex microstructure enhances ductility. The small grains can contribute to an increase in strength, while the bigger grains allow for the plastic deformation to take place. An optimal ratio of the volume fraction between the small and the big grains could therefore lead to an enhanced work hardening rate. Duplex grain sizes in nanocrystalline copper [Wan02] or aluminium alloys [Wit03] resulted in a decrease in yield strength, but a significant increase in the tensile elongation. Jin and Lloyd [Jin04] also showed that the UFG aluminium alloy AA5754 with a duplex grain structure produced by asymmetric rolling and annealing results in an increase in elongation to approximately 20 %, but the strength decreases by a factor of 2 in comparison to the originally rolled sample. The two main advantages of rolling at cryogenic temperatures include: a) suppressing the dynamic recovery during deformation and accumulating a large number of defects, which can act as potential recrystallisation sites and b) formation of an UFG microstructure at lower strains at room or elevated temperatures. Lee et al. [Lee04] investigated the effect of annealing on the microstructure and mechanical properties of the cryogenically rolled AA5083. They showed that the yield and tensile strength can be enhanced by more than 10 % in comparison to the alloys rolled at room temperature for the same percentage of reduction. The formation of a duplex grain structure led to a good combination of strength and ductility. 2.2.6 Strain rate sensitivity of UFG Materials Several publications have reported enhanced strain rate sensitivity (SRS) of UFG materials [Höp04, May05, Dal04]. Strain rate sensitivity is an important characteristic since it increases the material ductility and stabilises necking. Hart [Har67] defined strain rate sensitivity as the change in stress divided by the change in strain rate for a constant microstructure using the following equation.

..ln

ln

sconstd

dm ⋅=ε

σ Eq. 2-23

The transition between strengthening and softening of UFG materials, which becomes especially evident at lower strain rates or at elevated temperatures was observed by a number of researchers [Li04, Höp04, May05]. Li et al. investigated 99.99% purity Cu (oxygen-free, high conductivity grade) and introduced an important microstructural length scale, which can be used to explain strain rate sensitivity of UFG materials. As the grain size d decreases below the stress-dependent steady-state (dynamic equilibrium between strengthening and softening) grain size w, the mean-free path of dislocation becomes limited by the grain size rather than by the dislocation substructure. The steady-state subgrain size is shown in equation 2-24,


34

σ

bGkw w⋅

⋅= Eq. 2-24

where, kw is the numerical factor, σ is the applied stress, G is the shear modulus and b is the length of the Burgers vector. This means that fewer dislocations would be stored in the grain interior and more dislocations would be stored at grain boundaries. Therefore, the situation becomes strongly dependent on the grain boundary characteristic. If the dislocations can be annihilated more easily at grain boundaries by thermally activated processes, the material would soften and if the annihilation is not possible, the dislocations would be stored within grain boundaries and the material would harden [Li04]. Because of the high dislocation density in UFG material, which was reported to average approximately 1015 - 1016 m-2, and because of the dislocation storage at grain boundaries, it is quite likely for the thermally activated dislocation annihilation to take place even at low homologous temperatures. On the other hand, there have also been some suggestions regarding grain boundary sliding as a possible mechanism to explain the enhanced SRS in UFG materials, which should not be neglected [Sch03]. 2.2.7 Thermal Stability of UFG Materials Ultra-fine grained bulk materials which are produced by severe plastic deformation methods are characterised by a submicron scale grain size and a strongly deformed microstructure. They are not in thermal equilibrium and are therefore unstable. As previously mentioned, ultrafine-grained materials are also usually associated with relatively low elongation to failure owing to pre-work hardening during the deformation process. It is therefore important to study the annealing behaviour of heavily deformed materials with the aim to improve the ductility. Another important factor also includes determining thermal stability and with that the important material specific process frame. Thermal stability of UFG materials was investigated by a number of researchers [Mol07, Cao03, Mor02, Sla07, Par01]. It was reported that thermal stability of UFG materials is influenced by the number of ARB or ECAP cycles as well as by the material process temperature. Thermal stability of pure copper varies between 150-230°C [Mol07] and that of aluminium alloys between 300-350°C [Sla07, Par01]. There is a current dispute regarding the microstructural changes which occur during annealing. Until now, it is not clear whether grain coarsening or discontinuous recrystallisation is responsible for the appearance of a duplex microstructure. However, the ultrafine-grained materials contain high dislocation density and a high fraction of low angle grain boundaries. Moldova et al. [Mol07] argue that such structures undergo recrystallisation which is distinctly different from grain growth since it is driven by a volume energy differential caused by the dislocation structure and not by boundary curvature. Therefore, according to the essential features of the deformed microstructure, these materials undergo primary static recrystallisation during annealing and not discontinuous grain growth.


35

2.2.8 Texture Formation in ARB materials As a result of repeated rolling and high plastic strain, it is expected that the ARB materials develop certain texture, which will differ compared to texture of the as-received cold rolled or the solution treated sheets. Texture found in most aluminium and aluminium alloys corresponds to a typical rolling texture with a β-fibre component [Sla06b, Cho06]. The ARB sample texture may vary strongly through the thickness of the sheet. However, texture investigations on the precipitation strengthened Al-2.3 %Li alloy (AA8090) revealed that the degree of symmetry increased with increasing number of ARB cycles and the texture after the third cycle corresponded to a typical uniform β-fibre texture [Cho06]. Kamikawa et al. [Kam07] investigated the distribution of strain and the development of texture in accumulative roll bonded IF-steel during the process performed with or without lubrication. They observed that the shear strain significantly differs with and without lubrication. Conducting the ARB process with lubrication produces a homogeneously distributed microstructure, while rolling without lubrication results in high friction at the surface leading to inhomogeneously distributed microstructure, which strongly depends on the thickness location. The ARB samples rolled under lubrication had extremely elongated lamellar boundaries and sharp rolling texture, but the non-lubricated samples had relatively equiaxed pancake-shaped structures and a complicated textural distribution. Authors believe that the redundant shear deformation in the ARB process is necessary and effective in producing the equiaxed grain structure and random texture.


36

2.3 Sheet Metal Forming The ability to produce variety of shapes from flat sheets of metal at high rates of production has been a real technological advancement of the twentieth century. In essence, a shape is produced by stretching and shrinking the dimensions of a flat blank in the three mutually perpendicular principal directions [Die88]. Sheet metal forming can be classified into bending, shearing, deep drawing, stretching, bulging and etc. Due to today’s increasing demands towards energy consumption, economical savings and hence lightweight constructions, the challenges especially in the automotive industry are not only placed on the materials, but also on the metal sheet forming processes. Introduction of tailored blanks was one of the most important developments in the lightweight construction. Tailored blanks are made out of two or more metal sheets of unequal thicknesses or materials, which can be welded together to form a single blank (tailor welded-blank) and subsequently be used for bending, stretching, deep drawing, etc [Mer06]. In comparison to tailored blanks made out of steel, tailored blanking with aluminium presents more difficulties. Especially the hardenable aluminium alloys often used for drawing the outer body panels generally show lower formability compared to steel [Mer02]. Additionally, increasing the deformation rates also reduces the formability of aluminium. For this reason, new forming techniques and new technologies need to be employed in order to overcome the difficulties associated with aluminium. Locally heat treated tailored blanks offer an alternative possibility of achieving improved formability of hardenable aluminium alloys. Local laser heat treatment of aluminium sheets leads to local microstructural changes and hence different mechanical properties [Hof97, Mer01]. It is possible to obtain complete dissolution of the precipitates and therefore lower yield and tensile strength compared to the as-received material. In case of deep drawing, local heat treatment would decrease the flow stresses within the flange area and insure better formability. In this manner, it is possible to increase the deep drawing ratio by 30 % in comparison to the conventional non-heat treated material, while at the same time decreasing the punch force by approximately 10% [Mer02]. Over the last couple of years it was shown that hydroforming has a high potential with regard to lightweight construction, especially in the production of tubes for exhaust systems, chassis parts and frame structures. Hydroforming uses fluid pressure instead of a punch in order to obtain a desired shape of the sheet. The increasing use of hydroforming is due to a number of advantages of the process such as: easier forming of complex shapes, high form accuracy, high tensional and bending stiffness of the component and avoidance of fracture during forming of thin sheets. However, modified processing techniques need to be developed due to the previously mentioned difficulties regarding aluminium forming. A promising method for enhancing the formability is by increasing the process temperature below the recrystallisation temperature of the material [Kei05]. In this context, the formability potential of the newly developed UFG materials is going to be investigated using the conventional forming techniques such as hydraulic bulge testing and deep drawing at room and at elevated temperatures. However, for future references one should also bear in mind that advanced concepts and forming technologies such as tailored blanks and hydroforming could also be interesting in association with the UFG aluminium sheets.


37

2.3.1 Hydraulic Bulge Test In order to qualify a sheet metal material for industrial applications, it is important to characterise the strength and formability under the multiaxial stress state conditions, which usually occur during sheet metal forming such as bending, drawing or stretching. Although the most common and the most efficient type of mechanical testing is still the uniaxial tensile test, it nevertheless has two main disadvantages. In uniaxial tensile tests, the materials can only reach a small maximum uniform strain due to the plastic instability, which normally occurs as localised necking. Another disadvantage is that the flow curves can only be determined during uniform elongation i.e. before necking starts to occur. As a result, a hydraulic bulge test is often used to investigate the material behaviour under the biaxial tensile stress state. The bulge test represents a more realistic loading condition and leads to a more accurate representation of the structural behaviour under the industrial loading conditions [Han08]. Furthermore, since the loads are simultaneously applied in all directions, the material anisotropy is also taken into consideration and there is no bias to a particular direction. In a hydraulic bulge test, a circular sheet metal specimen is clamped under a drawing ring (upper die) of a cirtain diameter. In this manner the material flow between the drawing ring and the lower die is prevented. The movement of the punch applies the pressure on the liquid situated in a reservoir below the specimen. The specimen bulges and the deformation takes place in the sheet plane and in the sheet thickness (Fig. 2-18). In this manner, the specimen is deformed by stretching [Lan74].

Figure 2-18: Principle and geometry of the bulge test after [Gut04]. The biaxial tensile stress state results in radial and tangential stresses σr (corresponds to principal stress σ1) and σt (corresponds to principal stress σ2), which are equal to each other due to the axial symmetry of the circular specimen. On the other hand, the


38

deformation state is triaxial. The deformation in the radial and the tangential direction are positive due to the positive tensile stresses, while the deformation through the thickness of the sheet decreases leading to negative deformation values. The so-called membrane theory is commonly used in order to determine the flow curves obtained from the bulge tests [Gol75]. The membrane theory neglects bending and therefore it can only be applied to thin sheets. The ratio between the diameter of the drawing ring and the thickness of the sheet is a crucial parameter which must be optimised in order to avoid bending at the filler or transition radius of the drawing ring. Therefore, it is suggested to use a ratio of at least 1:100 between the sheet thickness and the diameter of the drawing ring [Mer01]. From the membrane theory, a number of equations concerning stresses, specimen geometry and bulge pressure have been derived (see Eq. 2-25 – 2-29). If a circular die is used, the specimen is evenly strained in all directions and the principal stresses (σ = σ1 = σ2) are equal to each other. The effective stress can be calculated by Tresca’s plastic flow criteria, from the principal stresses and the normal stress acting on the internal sheet surface. The effective stress can be calculated using the sheet thickness (Eq. 2-29).

tp

RR=+

2

2

1

1 σσ Eq. 2-25 [Gut04]

Eq. 2-26 [Gut04] due to symmetry: σ = σ1 = σ2

d

d

t2Rp⋅⋅

=σ Eq. 2-27 [Gut04]

( )p21

n −=σ Eq. 2-28 [Gut04]

⎟⎟⎠

⎞⎜⎜⎝

⎛+=−= 1

2minmaxd

d

tRpσσσ Eq. 2-29 [Gut04]

where, σ1, σ2: principal stresses, corresponding to radial σr and tangential σt stresses

average stress in the sheet normal σn: effective stress σ :

Rd: radius of the dome p: pressure t0, td: initial sheet thickness and thickness at the apex of the dome.

In order to determine the flow curves using the bulge test, one needs to determine four variables: instantaneous radius of curvature Rd, instantaneous wall thickness td, instantaneous dome height hd and pressure p. Depending on the equipment used, the determination of various parameters may be difficult and complicated to measure.


39

There have been a number of studies concerning the behaviour of aluminium or aluminium alloys and other materials under the biaxial loading condition [Nov03, Gut04]. However, only a few studies deal with the behaviour of the ultrafine-grained materials under the biaxial stress conditions [Lap07b, Lau04, Neu04]. One of the most frequently encountered problems concerns the production of relatively large sheets needed to perform a bulge test. For this reason, one of the earlier studies [ Neu04] on the formability of accumulative roll bonded UFG commercial purity aluminium AA1050 was performed using the Erichsen test (DIN 50 101), with the specimen diameter of approximately 10 mm. In an Erichsen test, the specimens are deformed using a circular punch. In this case, the frictional forces must firstly be overcome in order to draw the sheet. The conventionally grained, solution treated sheets with low dislocation density and good ductility, showed the best formability i.e. the highest displacement of the punch. The plastic deformation of the material increases with an increase in the number of ARB cycles, leading to reduced drawability of the UFG sheets. However, the drawability of the UFG sheets was not significantly lower than that of the solution treated sheets, although the force required for the drawability was much higher for the UFG sheets (6 kN) than for the conventionally grained ones (3 kN). In some tests, the UFG sheets have unfortunately fractured at the rim of the die. The possible reasons for this problem include: the somewhat higher notch sensitivity and bending stiffness of the sheets, the small radius of the die or the effect of friction [Neu04]. The Diploma thesis of T. Laumann [Lau04] focused on the manufacturing of the UFG aluminium AA6016 sheets using the ARB process, subsequently joining the sheets with the process of friction stir welding and at the end investigating the behaviour of the sheets using the bulge tests. It was shown that the sheets deform predominantly in the nugget area. The fracture strain and the maximal pole height of the ultrafine-grained (N8) friction stir welded sheets and the conventionally grained friction stir welded ones were comparable to each other. The strain and the maximal pole height of the UFG N8-FSW and the CG N0-FSW sheets averaged 10.8 % and 11.6 %, and 3.5 mm and 4.1 mm. Some lightweight materials like aluminium and magnesium alloys offer a high technological potential as lightweight construction components and can be used in processes like hydroforming in order to shorten the number of production steps. However, the high percentage of alloying elements in aluminium and the hexagonal crystal structure of magnesium lead to a relatively low formability of the sheets at room temperature. A promising strategy for the formability enhancement is forming at elevated temperatures. Testing at elevated temperatures results in lower flow stresses, and requires lower forming pressure as well as lower tool clamping forces of the sheet. The studies on formability of the conventionally grained 6xxx and 5xxx series indicated that the appropriate testing temperature must be determined for each material, but that the elevated testing temperature generally leads to better formability [Nov03]. The same strategy will be used for the ARB aluminium sheets, which also show limited formability at room temperature. It is aimed to increase the formability of the UFG sheets, while at the same time not significantly sacrificing the strength.


40

2.3.2 Deep Drawing Deep drawing is the metal forming process used for shaping flat sheets into cup-shaped articles such as automobile panels, bathtubs, kitchen sinks, shell cases and etc. This is done by clamping the initially flat sheet material, the blank, between a draw ring (die) and a blank holder (Fig. 2-19). The punch is pushed into the die cavity, while the metal sheet is deformed and takes the specific shape of the die and the punch. The blank holder exerts a force onto the blank in order to control the metal flow into the cavity and most importantly in order to prevent wrinkling [Die88].

Figure 2-19: Principle and geometry of deep drawing after [Lan74]. It can generally be said that the deep drawing process is characterised by tensile and compressive stresses (see figure 2-20). The main part of the deformation takes place in the flange i.e. at the transition radius of the drawing ring. At this point the metal sheet is drawn inward towards the throat of the die and it is also bent over the drawing ring. The flange area is exposed to radial tensile stresses and tangential or hoop compressive stresses. If the compressive stresses exceed the bending stiffness, the flange zone of the blank wrinkles. In order to prevent wrinkling a blank holder which exerts axial compressive stress is normally used in flange area. During forming, the cup wall is subjected to biaxial tensile stresses in the axial and tangential direction. The cup wall deforms in the direction of the stress, leading to continuous cup wall thinning. On the other hand, the cup base maintains its original sheet thickness, because it does not deform during the process. It is solely transported by the punch and


41

it is only elastically deformed. The actual transfer of the load takes place from the cup base to the cup wall, over the transition radius of the punch (i.e. transition base-wall). This zone is subjected to radial tensile stresses and the maximal drawing force is limited by the tensile strength of the material [Lan74, Die88].

Figure 2-20: Stresses in various sections of the drawn cup after [Die88]. The punch force is applied to the bottom of the cup and it is then transmitted to the sidewalls of the cup over the punch transition radius. The drawing force (FP) required to produce the cup is the summation of the deformation force (including the bending force at the radius of the punch) and the frictional force which comes from the hold-down pressure. The drawing force can be expressed as follows:

⎥⎦

⎤⎢⎣

⎡

⋅⋅+⎟⎟

⎠

⎞⎜⎜⎝

⎛⋅⋅⋅⋅

+⋅⋅⋅⋅⋅⋅=⋅

dd

NfP r

ttd

FddetdF

22ln1.1 0

,0011

0,0

201 σ

πμσπ

πμ Eq. 2-30 [Lan74]

where, d0 blank diameter

punch diameter d1t0: sheet thickness rd: die transition radius µ: coefficient of friction

average flow stress in the flange σ0,f : average flow stress in the area of the die σ0,d:

FP: drawing force and FN: blank holder load.

The failure of the material usually occurs in the cup wall just above the radius of the punch once the critical drawing force is exceeded. This area undergoes no drawing or bending, but is essentially subjected to tensile stresses. The stability criteria and the fracture force can be expressed in the equation below,


42

UTSFP tdFF σπ ⋅⋅⋅=≥ 01 Eq. 2-31 [Lan74]

where, FF: tearing or fracture force and

ultimate tensile strength. σUTS: In order to determine the drawing limit ratio, a series of circular blanks with a gradually increasing diameter have to be drawn to cylindrical cups using a punch with a constant diameter. The drawability or the drawing ratio of the metal sheet having a specific thickness can be determined from the ratio of the initial blank diameter to the diameter of the cup drawn from the blank.

1

0

dd

=β Eq. 2-32 [Lan74]

For every material there is a limiting drawing ratio (LDR, also known as βmax.), representing the largest blank diameter that can be drawn without fracturing.

1

max,0max d

d=β Eq. 2-33 [Lan74]

The limiting draw ratio strongly depends on friction between the flange and the blank holder. The effect of friction becomes greater when deforming the blanks with larger diameters. Other parameters, which also influence the limiting draw ratio, include the strain hardening exponent and the anisotropy values. The Schmidt-Kapfenberg [Lan74] method is a very useful method for determining the biggest possible blank diameter which can be used to evaluate the limiting drawing ratio (βmax). Firstly, the maximum drawing force has to be determined for two blanks having different diameters. The two chosen blank diameters should be such so that no fracture occurs. These two values (point 1 and point 2 in Fig. 2-21) must be plotted on a logarithmic scale, together with the fracture force of the third blank (point 3, in Fig. 2-21). The fracture force can be calculated by choosing the biggest possible blank diameter, which will insure fracture. Depending on the material, sheet thickness, the geometry of the toll and the lubrications, typical limiting drawing ratio for aluminium and aluminium alloys ranges between 1.85 and 2.05.


43

Figure 2-21: Schmidt-Kapfenberg method for determining the drawing limit ratio. Anisotropy can be defined as the dependence of mechanical properties on orientation. In many wrought metal products and especially in products deformed by processes such as rolling or extrusion, the tensile properties are frequently not the same in all directions. Crystallographic anisotropy arises as a result of the deformation process and the preferred orientation of grains. A practical manifestation of crystallographic anisotropy is the formation of “ears” (deutsch: Zipfelbildung) during a deep drawing process (Fig. 2-22). Earing is the formation of wavy edges at the top of the drawing cup and it can be directly correlated with the planar anisotropy, measured by Δr [Die88].

Figure 2-22: Influence of metal sheet anisotropy [Alu08].


44

The anisotropy coefficient or the so-called Lankford parameter r is a measure of plastic anisotropy and can be defined as:

z

y

thickness

width

dd

ddr

εε

εε

== Eq. 2-34 [Lan74]

where, dεy and dεz are incremental strains in y and z direction, i.e. within and normal to the sheet plane, respectively. Assuming volume constancy (dεx + dεy + dεz = 0) the Lankford parameter becomes:

qqr−

=1

Eq. 2-35 [Lan74]

with q = -dεy/dεx being the contraction ratio. Having determined the values of r for three directions in the plane of the metal sheet (0°, 45° and 90°), the normal anisotropy or the mean Lankford parameter can be obtained from the following equation.

4

2 90450 rrrr +⋅+=⟩⟨ Eq. 2-36 [Lan74]

Later, the anisotropy values in three different directions will be referred as rRD, r45° and rTD. The variation of the normal anisotropy or the mean Lankford parameter can then be determined from the equation 2-37.

2

2 45900 rrrr ⋅−+=Δ Eq. 2-37 [Lan74]

For good deep drawing, it is generally desirable to obtain large <r> values, which would lead to only small thickness reduction. On the other hand, Δr value should be as small as possible, aiming to approach zero. This would insure isotropic drawing conditions within the sheet plane and minimise earing.


45

2.4 Friction Stir Welding (FSW) 2.4.1 Friction stir welding of conventionally grained aluminium sheets Friction stir welding (FSW) was invented at The Welding Institute (TWI) in UK in 1991 as a solid-state joining technique, and it was initially applied to aluminium alloys [Tho91, Daw91]. In particular, it can be used to join high-strength aerospace aluminium alloys and other metallic alloys that are hard to weld by conventional fusion welding. Other advantages include, higher weight reduction potential compared to standard mechanical fasteners, no melting of the base material due to a solid state nature of the process, lower heat input compared to fusion welding, low distortion of the workpiece and good reproducibility.

Figure 2-23: Schematic drawing of friction stir welding after Mishra and Ma [Mis05]. A rotating tool with a specially designed pin and shoulder is inserted into the abutting edges of sheets to be joined and traversed along the joint line (Fig. 2-23). During FSW the material undergoes intense plastic deformation at elevated temperatures. The reported temperatures in the weld zone or the so-called nugget vary between 400 °C and 500 °C [Rho97, Liu97]. The material is mixed or stirred by the rotational motion of the pin, while the friction between the pin and the material results in heating of the workpiece. The localised heating softens the material around the pin and the combination of tool rotation and translation leads to material movement from the front to the back of the pin. The weld zone is generally left with fine, equiaxed and recrystallised grains. Friction stir welding involves complex material movement and plastic deformation. Welding parameters, tool geometry, and joint design exert significant effect on the material flow pattern and temperature distribution, thereby influencing the microstructural evolution of material. The geometry of the tool can affect the amount


46

of displaced material volume, welding force, material flow, the interface between the pin and the plasticised material and etc. The main welding parameters include tool rotation rate (ω, rpm), tool transverse speed (v, mm/min.) and tool tilt angle. Higher tool rotation rates generate higher temperature, because of higher frictional heating and result in more intense stirring and mixing of material. The tool tilt angle ensures that the material is efficiently stirred from the front to the back of the pin. The insertion depth of pin is associated with the pin height. If the insertion of the pin is too shallow the welding proceeds only along the surface and the bottom part of the material is not stirred. However, if the insertion depth is too deep, it leads to local thinning of the workpiece. Microstructural Changes Due to the intense plastic deformation and exposure to elevated temperatures during the process, a number of important microstructural changes take place. Researchers have confirmed recrystallisation and development of texture within the stirred zone as well as precipitation dissolution and coarsening with and around the stirred zone [Liu97, Sua05; Su03]. Generally four different areas after FSW can be differentiated: the base material, essentially unaffected by the process, the thermo-mechanically-affected-zone (TMAZ), the heat affected zone (HAZ) and the nugget (Fig. 2-24 a). • Nugget zone The nugget zone typically comprises of small, equiaxed and recrystallised grains [Ben99, Liu97]. The interior of the recrystallised grains generally contains low dislocation density. However, some researchers have also reported high density of subgrains and dislocations [Hei02, Jat00]. Irrespective of the process parameters used, most researchers were able to achieve a fine-grained microstructure ranging from 1 µm to 10 µm. However, there are different methods, which can be employed in order to control grain growth during the process. One of these includes cooling of the material with liquid nitrogen or decreasing the tool rotational speed and thus minimising friction. There is also some grain size variation within the nugget. The variation in grain size from the bottom to the top of the nugget is associated with differences in temperature profiles and heat dissipation. A typical equiaxed and recrystallised grain structure in the nugget of the aluminium alloy AA7050 can be seen in figure 2-24 b) [Bow90]. One of the most disputed topics with regard to friction stir welding is whether discontinuous or continuous dynamic recrystallisation mechanism is the cause of dynamic recrystallisation of the nugget. Some researchers suggested that the discontinuous dynamic recrystallisation is highly unlikely in aluminium alloys due to aluminium’s high stacking fault energy, which makes dislocation climb and cross slip easy. As a consequence, it is difficult to achieve sufficient dislocation density in order to initiate the discontinuous dynamic recrystallisation. Others believe that the second phase particles can initiate the nucleation of new grains and result in discontinuous dynamic recrystallisation. Heinz and Skrotzki [Hei02] proposed the mechanism of continuous dynamic recrystallisation, which was based on a gradual rotation and transformation of low angle grain boundaries into high angle grain boundaries.


47

Initially, Su et al. [Su03] argue for the continuous dynamic recrystallisation, but claim that this it is based on a dynamic recovery process where dislocations are absorbed into the grain boundaries thus increasing the misorientation between adjacent subgrains. However, more recent results have shown that it is likely that the high-angle boundaries were created directly from the heavily deformed substructure by moderate to high levels of plastic deformation. As the FSW process continues, the strain, strain rate and temperature gradually decrease behind the pin and a number of changes occur in the recrystallised grains, including grain growth, dislocation introduction, dynamic recovery and continuous dynamic recrystallisation. Since the deformation is non-uniform, different grains may be affected by different mechanisms. Su et al. [Su03] believe that the final FSW microstructure consists of some grains formed as a result of growth of the initially recrystallised grains, while others are formed from subgrains via continuous dynamic recrystallisation. The grains have different densities of dislocations and are in various degrees of recovery. On the other hand, researchers Cabibbo et al. [Cab07] observed that there are alternating layers of crystallites and elongated grains with cells, and that there is no evidence of growth by boundary migration from lower density nugget layers to more deformed cell layers after FSW, and argue that discontinuous recrystallisation does not occur. a)

b) c)

Figure 2-24: a) Typical microstructure showing various zones in FSW 7075Al-T651 [Mis05], b) zoomed in areas of the equiaxed and recrystallised grains in the nugget of AA7050 [Bow90] and c) thermo-mechanically affected zone (TMAZ) of the AA7075 [Ma02]. In the age hardenable aluminium alloys, it is possible that the precipitates coarsen and dissolve in the matrix due to the high temperature in the nugget. Liu et al. [Liu97] investigated the microstructure of a friction stir welded AA6061-T6. The precipitates in the workpiece were small and homogeneously distributed, while the nugget showed


48

fewer larger precipitates. This implies the occurrence of dissolution and coarsening of precipitates during FSW. Some researchers like Cabibbo et al. [Cab07] observed complete precipitation dissolution and re-precipitation within the nugget of AA6056 aluminium alloy, resulting in a slightly higher hardness in the nugget than in the TMEZ. On the other hand, Chen et al. [Che06b] argue that the process temperature within the nugget is not sufficient enough to dissolve the stable precipitates, although it may be sufficient for the dissolution of the metastable ones. Whatever the case may be it seems that the precipitation morphology in the nugget zone or other zones next to the nugget is strongly affected by: the alloy type, the alloy state, e.g. T4 or T6, process parameters e.g. rotational and linear speed of the tool, temperature etc., and the tool geometry. The overall response includes a combination of coarsening, dissolution and reprecipitation of precipitates during FSW. • Thermo-mechanically affected zone (TMAZ) Just like the nugget, the thermo-mechanically affected zone is influenced by the plastic deformation and elevated temperature. This zone is characterised by a strongly deformed structure in the direction of pin rotation and strongly elongated grains (Fig. 2-24 c) which may contain high density of subgrains. Recrystallisation does not take place due to lower strain and temperature in comparison to the nugget. Precipitation coarsening and even dissolution has been reported by Sato et al. [Sat99b] and Su et al [Su03]. • Heat affected zone (HAZ) The heat affected zone (HAZ) is situated next to the thermo-mechanically affected zone (TMEZ). It is affected by temperature, but does not undergo any plastic deformation. In the heat-treatable aluminium alloys the temperature in the HAZ may rise above 250 °C [Mah98] and the precipitates generally tend to coarsen [Su03, Hei02]. Mechanical Properties As a consequence of significant microstructural changes that take place within and around the stirred zone, the mechanical properties of the friction stir welded sheets vary across the weld. The change in hardness across the nugget is different for age-hardenable (precipitation hardened) and for non-hardenable (solid solution hardened) aluminium alloys. The focus here is the age-hardenable aluminium alloys. Figure 2-25 shows the hardness profile of the friction stir welded AA6063-T6 aluminium alloy. This alloy shows significant softening in the nugget, which was also reported for many other age-hardenable aluminium alloys [Bar06, Liu97, Cab07, Che06b, Gen05, Sat99a]. Sato et al. [Sat99b] reported that the hardness profile was strongly affected by the precipitate distribution rather than the grain size in the weld. The reason for a decrease in hardness was traced back to the precipitation type, size and distribution. Slightly higher hardness in the nugget than in the solution treated material was explained by the smaller grain size and higher density of subgrains.


49

Figure 2-25: Hardness profile of friction stir welded AA6063-T5 [Sat99b]. The as-received or original material condition has a significant influence on the resulting mechanical properties across the weld. Chen et al. [Che06b] measured a hardness increase in the nugget of the fully annealed material (AA2219-O) and a hardness decrease in the nugget of the solution heat treated and artificially aged material (AA2219-T6). The stable precipitates in the AA2219-O weld do not dissolve in the matrix inspite of the high temperature, while the AA2219-T6 state contains a large amount of quasi-stable precipitates, which dissolved during FSW and led to an overall decrease in hardness. On the other hand, Genevois et al. [Gen05] investigated a similar aluminium alloy AA2024 in two different initial states namely T351 and T6 and found that the initial state of the base material does not influence the hardness of the zones after welding. It can therefore be concluded that due to a complex strain and temperature distribution during the FSW process, the influence of FSW parameters on general properties of the weld are still difficult to predict. It seems that the process has to be investigated for every system and every material and its conditions independently in order to draw clear conclusions. Nevertheless, most authors seem to agree on one point: the hardness profile of the friction stir welded age-hardenable aluminium alloys depends mainly on the precipitation type, size, and distribution, and only slightly on the grain size and the dislocation structure. 2.4.2 Friction stir welding of ultrafine-grained aluminium sheets Although the friction stir welding process has already been established and used for technical applications as a joining technique for conventionally grained aluminium alloys, copper or steel, the research regarding friction stir welding of UFG materials is still limited. Until now Sato et al. [Sat04, Sat06] are the only known researchers who have conducted FSW on accumulative roll bonded or equal channel angular pressed


50

aluminium. Friction stir welding was found to be a desirable joining technique for UFG materials, since it provides excellent mechanical properties and retains the fine grained microstructure. They reported a strongly elongated ultrafine-grained microstructure, with approximately 260 nm grain size, after the ARB process shown in figure 2-26 a), pancake-shaped grains in the TMEZ and equiaxed, dynamically recrystallised grains in the nugget (Fig. 2-26 b), of the UFG AA1050 and AA1100. a) b)

Figure 2-26: TEM images of the ARB processed AA1100 aluminium a) before and b) after friction stir welding (nugget zone) [Sat04, Sat06]. The hardness of fully annealed and ARB processed commercial purity AA1100 aluminium is compared in figure 2-27. Even though there is some softening of the ARB material after FSW, the hardness of the nugget of the starting and ARB material remains the same. This indicates that the stored energy in the base material hardly affects the evolution of the grain structure in the stir zone and that the mechanical properties of the FSW commercial purity aluminium are predominantly affected by grain size. Until now, there are no known publications regarding the friction stir welded accumulative roll bonded hardenable aluminium alloys.

Figure 2-27: Vickers hardness profiles of friction stir welded AA1100 [Sat04].


51

Furthermore it must be mentioned that the conventionally grained material AA1100 shows higher hardness in the nugget than in the base material. This observation is contradictory to that observed for the conventionally grained hardenable aluminium alloy AA6063-T5 (Fig. 2-25). It seems that the mechanical properties of non-hardenable aluminium alloys such as AA1100 are predominantly affected by the grain size, whereas hardenable aluminium alloys seem to be predominantly affected by the evolution of precipitates during FSW and the precipitation morphology.


52

2.5 Corrosion of aluminium and aluminium alloys 2.5.1 Corrosion of conventional aluminium and aluminium alloys Aluminium and aluminium alloys are very stable in most environments due to the rapid formation of the aluminium oxide layer Al2O3 at the surface that inhibits the corrosion of the underlying bulk material (Fig. 2-28). The amorphous aluminium oxide layer is thin and dense, and is composed of a base layer with a thickness of 1 - 2 nm and a top layer of 5 - 10 nm. The base layer has an extremely small electron and ion conductivity and functions as an insulator. If the surface of aluminium is scratched sufficiently enough to remove the protective oxide layer, a new layer will form spontaneously. The top layer may contain different oxides of alloying elements like Mg or Zn, as well as intermetallic phases, which may affect the overall properties of the oxide layer [ Kam02].

Figure 2-28: Composition of the aluminium oxide layer [Kam02]. From a Pourbaix diagram (Fig. 2-29), the protective layer is stable in aqueous solutions of a pH range between 4.5 - 8.5, while it is soluble in strong acids and alkalis [Pol98].

Figure 2-29: Simplified sketch of a Pourbaix diagram for aluminium [Rev00].


53

It is important to note that the alloying elements may be present in a solid solution of aluminium or as micro-constituents or precipitates such as Al2CuMg, Mg2Si or FeAl3. In most cases, a solid solution is the most resistant form of an alloy whereas micro-constituents can cause localised corrosion attack or pitting and therefore be detrimental for corrosion resistance. Iron and silicon for example often occur as impurities and form compounds most of which are cathodic with respect to aluminium. Excess of magnesium in a solid solution of binary aluminium alloys can lead to intercrystalline attack due to formation of the strongly anodic Mg5Al8 phase. However, magnesium in combination with silicon e.g. Mg2Si has a similar electrode potential to aluminium and no localised attack occurs. Presence of copper in micro-constituents reduces the corrosion resistance more than any other alloying element, although when added in small amounts (0.05 - 0.2 %), the localised attack is reduced [Pol98]. 2.5.2 Corrosion of ultrafine-grained aluminium and aluminium alloys The amount of literature on corrosion properties of ultrafine-grained materials is limited and even more so in case of UFG aluminium alloys. It is generally believed that the UFG materials possess considerably lower corrosion resistance due to high number of defects (dislocation substructures or impurities such as Fe particles which may become embedded in the soft aluminium surface during wire brushing in case of ARB processing) introduced into the materials during ARB or ECAP processes. However, this belief has proven to be controversial, since both corrosion improvement as well as corrosion deterioration have been reported. Nanocrystalline Fe - 8 % Al tested in 0.1 M Na2SO4-solution showed corrosion pitting in a strongly acidic medium pH:1 and improved corrosion resistance in a weakly acidic medium pH:6 in comparison to the conventionally grained material that showed exactly the opposite reaction [Zei95]. Therefore, it seems reasonable to say that the corrosion response is strongly influenced not only by the type of material system and material production history, but also by the environment. a) b)

Figure 2-30: Polarisation curves showing a) CG and UFG Al-Mn specimens in 3.5% NaCl solution [Wei07] and b) the influence of various number of ECAP passes of AA1050 aluminium in 0.1 M Na2SO4 + 100 ppm Cl⎯⎯ solution [Chu04].


54

From the polarisation curves of CG and ARB produced UFG Al-Mn alloy submerged in 3.5 % NaCl solution, it is evident that the UFG material shows higher corrosion resistance (Fig. 2-30 a), which was proven to be coupled with a decrease in size of the second phase particles MnAl6 [Wei07]. Authors Chung et al. [Chu04] investigated the effect of the number of ECAP cycles on the electrochemical properties of the commercial purity aluminium AA1050. Potentiodynamic polarisation curves performed in 0.1 M Na2SO4 + 100 ppm Cl⎯⎯ on an exposed area of 0.25 cm2 using a saturated calomel electrode (SCE) as a reference electrode and a graphite electrode as a counter electrode, showed that a dominating form of corrosion was pitting corrosion and that the pitting resistance increased with increasing number of ECAP passes (Fig. 2-30 b). Similarly to [Wei07], it was concluded that the size of the cathodic Si-containing impurities, prone to creating a galvanic cell with an anodic aluminium matrix, was reduced with increasing number of ECAP cycles and therefore improved the corrosion pitting resistance. 2.5.3 Corrosion of friction stir welded conventional aluminium and aluminium

alloys A number of reports have been published regarding friction stir welding of aluminium and aluminium alloys, although only limited amount addresses the problem of corrosion of such materials. Most publications focus predominantly on the 2xxx and 7xxx aluminium alloy series due to a generally higher corrosion susceptibility. Wrought aluminium alloys of 2xxx series contain copper as a major alloying element, which generally decreases the corrosion resistance due to a high electrode potential difference with respect to pure aluminium or copper-rich second phase particles which may create a galvanic cell. The 7xxx series aluminium alloys are more anodic to pure aluminium due to the zinc content and are among the aluminium alloys most susceptible to stress corrosion cracking. Until now, there are no known publications regarding the corrosion behaviour of friction stir welded ultrafine-grained aluminium sheets.

CHAPTER 3 Experimental Procedure

55

3 EXPERIMENTAL PROCEDURE

In this chapter the main experimental techniques will be highlighted and summarised. The chapter provides technical insight into ultrafine-grained material processing, microstructural and mechanical characterisations as well as the characterisation regarding sheet metal formability, welding and corrosion. 3.1 Materials Investigated The main focus of research was placed on commercial purity aluminium AA1050 and aluminium alloy AA6016 due to the high technical potential as light weight construction components in the automotive industry. Additionally, emphasis was placed on transferring the ARB process on other technically relevant materials. For this reason, materials such as magnesium, titanium, copper and niobium were also investigated. The materials investigated are listed in table 3-1. Table 3-1: List of materials investigated

Material Designation As-received state

1. Commercial purity aluminium AA1050 (Al 99.5 %) EN-Norm: H14 (cold rolled) DIN-Norm: F11 (cold rolled)

2. Aluminium alloy AA6016 T4

3. High purity oxygen free copper C10100 (Cu 99.99 %) “soft state”

4. Titanium Grade 2 Ti G2 (Ti 99.2 %) rolled

5. Niobium Nb (Nb 99.8 %) rolled

6. Magnesium Mg AZ31 (Al 3 %, Zn 1 %) rolled

Initially, the ARB process was conducted on aluminium materials. Aluminium and aluminium alloys are characterised by low density, high strength and good formability. Commercial purity aluminium AA1050 was used as a model material and the first ARB results concerning processing and mechanical properties were then compared to technically relevant aluminium alloy AA6016. In the automotive industry there is a special interest regarding this alloy, due to its strength and good formability. Aluminium alloy AA6016 derives its strength primarily from the precipitation strengthening mechanism of the second phase particles Mg2Si. Another advantage of this alloy is that it does not build any strain marks upon sheet metal forming and it is for this reason usually used for the outer body automobile panels. The chemical composition of two aluminium alloys can be seen in table 3-2.


56

Table 3-2: Chemical composition of aluminium AA1050 (highest allowable values) and AA6016 [Kam02]

wt.% Si Cu Fe Mn Mg Cr Zn Ti other Al AA1050 0.25 0.05 0.4 0.05 0.05 - 0.07 0.05 0.03 balance AA6016 1.0-1.5 0.52 0.5 0.2 0.25-0.6 0.1 0.2 0.15 0.15 balance

Magnesium and magnesium alloys are also interesting for light weight applications due to their high specific strength. Therefore, it was also attempted to produce the UFG magnesium AZ31 sheets by the ARB process. Unfortunately, the ARB process was not successful, due to the hexagonal crystal structure, which is hard to deform at low process temperatures. Details will be discussed in the following chapter. The chemical composition of the magnesium alloy is shown in table3-3. Table 3-3: Chemical composition of magnesium alloy AZ31 (highest allowable values) [ Ave99]

wt.% Al Zn Mn Cu Si Ni Fe other Mg AZ31 2.50-3.00 0.60-1.40 0.20-1.00 0.04 0.10 0.005 0.005 0.30 balance

Materials such as titanium, copper and niobium (see Tables 3-4, 3-5 and 3-6 for chemical composition) were all investigated for different reasons. In the framework of this work, these materials were predominantly used as model materials and the results which were obtained were used as reference tests for parameter studies. In contrast to aluminium and magnesium alloys, these materials have significantly higher melting points (see Appendix). Therefore, the homologous temperature of these materials is significantly lower than that of aluminium even if they are rolled at 300 °C. Table 3-4: Chemical composition of titanium grade 2 (highest allowable values) [Boy94]

wt.% C Fe H N O Ti Ti G2 0.1 0.3 0.015 0.03 0.25 (99.2 ) balance

The material crystal structure plays an important role during processing i.e. rolling. Copper has a face centred cubic structure and can be compared to the aluminium alloys. Niobium on the other hand has a body centred cubic structure and it can therefore be compared to the interstitial free (IF) steel. Interstitial free steel was originally planed to be investigated because of the excellent mechanical properties and the already known potential in the automotive industry for the autobody panels. However, until now roll bonding of IF steel was not possible due to the limited permissible process temperature of the rolls. Therefore, the work on niobium was the preliminary work, which will be used in the future as the basis for processing the ultrafine-grained IF steel.


57

Table 3-5: Chemical composition of oxygen free copper, C10100 (highest allowable values) [ Wie08]

wt.% O P S Fe Ni Cu OFE Cu 5 ppm 3 ppm 15 ppm 10 ppm 10 ppm (99.99 ) balance

Magnesium and titanium alloys both have a hexagonal crystal structure and it was expected that these materials show similar behaviour during ARB processing. For reasons mentioned in chapter 4, it was not possible to roll bond magnesium. However, the ARB process was successfully conducted on high purity titanium and titanium-aluminium laminates, which were technologically interesting due to their high specific strength. The rolling of laminates will be mentioned shortly in the following section. Table 3-6: Chemical composition of niobium (highest allowable values) [ Pla08]

wt.% O N H C W Ta Mo Nb

Nb 150 ppm 100 ppm 15 ppm 50 ppm 100 ppm 3000 ppm 100 ppm (99.8) balance

In order to be able to compare these materials more easily, a table (Table 8-2) of physical properties was compiled and can be used as a quick reference.


58

3.2 Accumulative Roll Bonding (ARB) Process The accumulative roll bonding (ARB) process belongs to a group of severe plastic deformation (SPD) techniques used to develop ultrafine-grained sheets by applying repeated rolling, which leads to high levels of shear strain throughout the sheet thickness. The surfaces of one millimetre thick and 50 - 100 mm wide metal strips were preliminary wire brushed in order to remove the oxide layer for better interlamellar bonding of sheets, and subsequently folded or stacked on top of each other and rolled together without lubrication using a two-high rolling mill (Carl Wezel, Mühlacker; roll diameter: Ø = 130 mm) or a four-high rolling mill (Carl Wezel, Mühlacker; roll diameter: Ø = 32 mm). The process was then repeated a number of times. The four-high rolling mill, which was predominantly used for processing can be seen in figure 3-1. The roll diameter and the peripheral roll speed of the four-high rolling mill averaged 32 mm and 80 revs/min, respectively.

Figure 3-1: Carl Wezel rolling mill BW 200/130 (Mühlacker) used for ultrafine-grained metal sheet manufacturing. The most important ARB process parameters include the initial state of the material, process temperature, thickness reduction and the number of rolling cycles N.


59

Commercial purity aluminium AA1050 and aluminium alloy AA6016 were solutionised before the ARB process in order to obtain a defined reference material. Depending on the type of material being processed, the process temperature varied between 20 - 350 °C and the minimum amount of thickness reduction averaged 50 %. Rolling at elevated temperatures involved prewarming of metal sheets in a furnace for a couple of minutes before rolling. All samples were air cooled before repeating the process. The maximum number of ARB cycles of each material is generally dependent on the initial strength of the material, the development of the UFG microstructure and the achievable saturation in strength. The process was usually completed after the UFG microstructure has been developed and the saturation in strength was reached. The initial state of the material and the ARB parameters are listed in table 3-7. Table 3-7: Initial states of materials prior to rolling and the corresponding ARB process parameters

Preheating in furnace Material State before

rolling

Initial sheet

thickness Temp. Time

Percentage thickness reduction

Max. number of cycles N

mm °C min. %

AA1050

solutionised (520°C/1h)

and H2O quenched

1

RT

3-5

50

8 or 10

AA6016

solutionised (520°C/1h)

and H2O quenched

1

180, 230, 250

3-5

50

8 or 10

Cu 99.99 as-received 1 RT 3-5 50 12

Ti G2 as-received 0.8 250 3-5 > 50 6

Nb 99.98 as-received 1 250 3-5 > 50 3

Mg AZ31 as-received 1 20-350 3-5 50 /

At the beginning the rolling mill used for the ARB process was a two-high rolling mill (Carl Wezel, Mühlacker; roll diameter: Ø = 130 mm). The maximum width of metal strips produced without severe cracking at the edge was approximately 5 cm wide. In order to improve and optimise the process as well manufacture wider metal sheets, a four-high rolling mill was acquired. Some technical details of the two rolling mills can be seen in table 8-1 (Appendix), while the differences in ARB process, process improvement as well as a detailed comparison of the mechanical and surface properties of the sheets will be discussed in the results chapter. The most important technical characteristics of the two rolling mills, which influence the quality of the surface and the mechanical properties of the metal sheets, are the roll diameter and hence the projected length of arc of contact, i.e. the deformation zone, as well as the roll speed and the surface of the rolls.


60

3.3 Microstructural Investigation The microstructural evolution, precipitation morphology in aluminium alloy AA6016, interlamellar bonding and surface fracture after uniaxial tensile tests of the conventionally grained (CG) and the ultrafine-grained (UFG) sheets were investigated using different microscopy techniques. The directions and planes of samples were defined with respect to the original rolling direction (Fig. 3-2). The type of microscopy used depended primarily on the grain size of the rolled samples. Different microscopy techniques are listed together with the materials and the preparation procedures in table 3-8. Figure 3-2: Definition of planes and directions of 1 mm thick metal sheets and samples with respect to the original rolling direction. A light microscope Leica DMRM was predominantly used for the analysis of the as-received materials with a typical grain size of 10 - 50 µm. On the other hand, grain size analysis of ultrafine-grained materials with a typical grain size of 100 nm was carried out using a transmission electron microscope (TEM Philips CM200) with considerably higher magnification and resolution capabilities. Compositional analysis of the precipitation present in aluminium alloy AA6016 were carried out using an EDS technique (Energy Dispersive Spectrum) by means of STEM (Scanning Transmission Electron Microscope) incorporated in the TEM. The quality of interlamellar bonding between the individual aluminium sheets as well as the fracture surfaces of samples after tensile testing, have both been analysed using a scanning electron microscope SEM JSM 6400, company JEOL. Due to excellent lateral and vertical resolution, initial microstructural investigations of UFG aluminium alloys were performed using an atomic force microscope (AFM Dimension 3300, contact mode). During the course of research it was recognised that this technique could only be used for qualitative evaluation of ultrafine-grained materials. Quantitative analysis of grain size in ultrafine-grained materials or precipitation was not possible due to the strongly deformed microstructure and aggressive corrosion attack. Corrosion pitting was especially evident around the iron based intermetallic particles, which usually occur in aluminium alloys during processing.

Experimental Procedure

1 A1 Standard Struers Electrolyte: 90 ml distilled water 730 ml ethanol 100 ml butylcellosolve 78 ml perchloric acid

2 D2 Standard Struers Electrolyte: 500 ml distilled water 250 ml phosphoric acid 250 ml ethanol 2 ml Vogel’s Sparbeize and 50 ml propanol

61

Table 3-8: List of microscopy techniques, materials, types of investigations conducted and sample preparation procedures

Type of microscopy

Material investigated

Type of investigation Sample preparation procedure

1. Cutting to size: 1 mm thickness X 1 cm length (RP), 1 cm width X 1 cm length (NP) 2. Grinding 3. Polishing: OPU suspension (0.04 µm) Light

microscope (LM)

Al, Cu, Ti, Nb

(CG

and / or UFG)

Grain size, Interlamellar

bonding

4. Al Etching: Electrolyte A21 Voltage = 20-25 V Time = 20-30 s

4. Cu Etching: Electrolyte D22 Voltage = 1-2 V Time = 2 s

4. Ti Etching: 600 ml Methanol 340 ml Butanol 60 ml Perchloric acid Voltage = 20-60 V Time = 20-30 s

4. Nb Etching: 70 ml distilled H2O 20 ml Hydrogenperoxid (30%) 10 ml Amoniac solution (32%) Voltage = 10-30 V Time = 60-120 s

1. Cutting to size: 1 mm thickness X 1 cm length (RP), 1 cm width X 1 cm length (NP) 2. Grinding 3. Polishing: OPU suspension (0.04 µm)

Atomic force microscope (AFM)

Al

(CG and / or UFG)


bonding 4. Etching: Electrolyte A21; Voltage = 20-25 V; Time = 20-30 s

Scanning electron microscope (SEM)

Al, Ti

(CG and / or UFG)


bonding, Fracture

Please see light microscopy details!

1. Cutting to size with diamond saw: 1 mm thickness X 1 cm length (RP)

2. Grinding: to 200 µm 3. Al Electrolytic polishing:

Perchloric acid + vinegar (Ratio = 1 : 10) Voltage = 45-75 V Time = 20-60 s

3. Cu Electrolytic polishing: Electrolyte D22

Voltage = 7-10 V Time = 20-60 s

Transmission electron microscope (TEM)

Al, Cu, Ti, Nb

(UFG)

Grain size, Precipitation

Or 3. Ion beam milling (conditions similar for all materials; Al, Ti and Nb): Max. Voltage = 7,0 kV; Max. Current = 2,0 mA; Angle of incidence = 8°; Time = 1-12 h (depending on material and thickness)

CHAPTER 3


62

It should be noted that the sample preparation procedure especially in case of electropolishing was strongly dependent on the material type and condition (CG or UFG) as well as sample size i.e. surface area. The time taken for electropolishing and the voltage applied should therefore be regarded as the recommended rather than the exact values. Generally, electropolishing of samples in the rolling plane (RP) for the TEM preparation was especially difficult due to the very narrow sample geometry, which was restricted by the original sheet thickness of maximum 1 mm. During electropolishing the chemical attack and thinning would take place only along the sides rather than the central sample area. Some materials like titanium and niobium were also difficult to electropolish due to strong and uncontrollable etching. For this reason, some TEM samples were thinned using Ion Beam Milling (Balzer, Rapid Exchange System). This technique is more time consuming and depending on the material and sample thickness, 1 - 12 h may be necessary for the sample preparation to be completed. Nevertheless, the samples thinned using the ion beam milling technique showed good quality and reproducibility. 3.4 Surface Roughness In the Diploma Thesis of C. Klösters [Klö06] the influence of surface roughness on the quality of the bond between the metal sheets was investigated. In this work, the surface roughness measurements were extended to further analyse the difference in surface quality between metal sheets rolled using a two high rolling mill and a four high rolling mill. The measurements were carried out using a roughness measuring machine Perthometer S6P from Perthen. A diamond tip scans horizontally over a certain distance along the sample surface, while at the same time registering and digitally processing the vertical displacement of the tip using a displacement transducer. The output data include the most common surface roughness parameters such as the average roughness parameter Ra and the mean peak-to-valley height Rz. The average roughness parameter Ra is the arithmetic mean of absolute values of the surface displacements from the mean plane within the sampling area. The mean peak-to-valley height Rz is the average difference between five highest peaks and five deepest valleys within the sampling area. The measurements were taken in the normal top and bottom plane of the sheets, and along the rolling and transverse direction of the as-received and ARB samples.


63

3.5 Heat Treatment This section addresses the influence of annealing temperature and time on the microstructural evolution, precipitation kinetics, mechanical properties and thermal stability, and is divided in three major areas:

• the influence of precipitates in aluminium alloy AA6016 • thermal stability measurements of CG and UFG materials and • the influence of post-roll annealing on the mechanical properties of UFG

materials. 3.5.1 Precipitation Development in AA6016 The precipitation evolution and the influence of precipitates on the mechanical properties, in this case hardness, were investigated by means of annealing. The coarse grained and ultrafine-grained samples were annealed either at a constant (furnace annealing) or increasing temperature (Differential Scanning Calorimetry, DSC). The matrix of investigations is summarised in table 3-9. Table 3-9: Matrix of various annealing procedures

Investigated Method Conditions Measured 1.

Influence of natural aging on the mechanical properties of CG AA6016. AA6016 investigated after solutionising, 1 day and 1 week of natural aging.

Furnace annealing

180°C, varying time

Vickers hardness after annealing

2.

Influence of natural aging on the mechanical properties of CG AA6016. AA6016 investigated after solutionising, 1 day and 1 week of natural aging.

DSC

Temperature increase

10 K/min. up to 400 °C

Heat flow in mW/mg

3.

Influence of artificial aging on mechanical properties of CG and UFG AA6016

Furnace annealing

180°C, varying time

Vickers hardness after annealing

4.

Influence of artificial aging on mechanical properties of CG and UFG AA6016

DSC

Temperature increase

10 K/min. up to 400 °C

Heat flow in mW/mg

Differential scanning calorimeter, DSC 204 F1 Phoenix from Netzsch (Fig. 3-3), measures the difference in heat flow between the test sample and the reference sample while both are simultaneously heated in a chamber according to a given temperature programme, where the relevant parameters such as the rate of temperature increase and the maximum allowable temperature can be specified. From the DSC curves, it is possible to recognise if processes such as transformation changes, precipitation development or recovery and recrystallisation are taking place. Generally, any internal microstructural change, which occurs during heating will cause a peak in the DSC curves, indicating that the sample requires more (or less) heat than the reference


64

sample in order to maintain the constant temperature. The samples were weighed and heated in the chamber at the rate of 10 K/min. The chosen temperature profile was repeated twice. During the first run, it was possible to observe the internal processes taking place within the sample and the second run was used as a reference.

Figure 3-3: Schematic drawing of a differential scanning calorimeter, see [Klö06]. 3.5.2 Thermal Stability Thermal stability of CG and UFG samples was investigated by heat treating the samples in a furnace for 1 h at different temperatures, usually 100 °C, 200 °C, 300 °C, 400 °C, 500 °C, and subsequently quenching them in water. Finally, the Vickers hardness of the samples was measured using a Vickers Hardness Tester V-100 from Leco in the normal plane. The thermal stability of the CG samples was measured and used for comparison purposes. 3.5.3 Post-roll Annealing Post-roll annealing was performed on the tensile test samples of UFG aluminium alloy AA6016. The post-roll annealing was performed at 100°C, 200°C and 300°C. The total time of annealing averaged one hour. Tensile testing was then conducted at room temperature and a strain rate of 10-4 s-1.


65

3.6 Mechanical Characterisation The main purpose of mechanical testing was to characterise the novel ultrafine-grained materials, compare them to their coarse grain counterparts and explore their technological potential. The investigations included hardness testing, uniaxial tensile testing, hydraulic biaxial bulge testing and deep drawing. The mechanical characterisation was also performed on friction stir welded aluminium sheets. 3.6.1 Hardness Testing Depending on the area of interest, three different hardness measuring techniques were conducted on the CG and UFG metal sheets. Macroscopic Vickers hardness measurements proceeded on rolled metal sheets in the normal plane and transverse direction after each or every second ARB cycle using Vickers Hardness Tester V-100 from Leco. Most measurements were taken with HV5 or HV10 for a period of 15 s. The measurements were carried out on all materials investigated. The microscopic hardness measurements of all welded sheets were conducted using a Leco M-400-6 hardness measuring machine. The as-received and the ultrafine-grained commercial purity AA1050 and aluminium alloy AA6016 were friction stir welded and the hardness measurements were carried out in two rows in the transverse plane (ND-TD plane) across the weld zone with HV0.05 (Fig. 3-4). The samples were preliminary cut to size (1 mm × 15 mm), embedded in resin and mechanically ground and finally polished with 1 µm diamond suspension.

Figure 3-4: Schematic drawing of a typical friction stir welded sheet (hardness measurements were taken in the transverse plane, as indicated by the indents). Finally, Nanoindenter XP, MTS Systems, (Fig. 3-5), was used to investigate the accumulative shear strain or hardness distribution across the thickness of the rolled ARB sheets. Nanoindentation provided a good overview of the mechanical properties on the nano or submicron scale and was advantageous due to the relatively small indentation a small sample thickness of only 1 mm and small interlamellar spacing. The technique applied to measure the hardness distribution was a continuous stiffness module (CSM), which records stiffness data together with the load and displacement, allowing hardness as well as Young’s Modulus to be calculated for every data point. A


66

Berkovich diamond indentation tip, with the maximum displacement into surface of 3000 nm and an indentation strain rate of 0.05 s-1 was used for these investigation purposes.

Figure 3-5: Sketch of a Nanoindenter XP [MTS, Nano Instruments INC, Product Specifications Nanoindenter XP, Oak Ridge, TN, USA, 2003]. 3.6.2 Tensile Testing Tensile testing incorporated the comparison of mechanical properties between CG and UFG materials with different number of ARB cycles N, the investigation of strain rate sensitivity using three different strain rates 10-3, 10-4 and 10-5 s-1, tensile testing of friction stir welded samples and tensile testing of CG and UFG samples after rolling and additional heat treatment. The use of two different tensile sample geometries is shortly explained below. The CG and UFG materials tested and the overview of the experimental matrix can be seen in table 3-10. All tensile tests were carried out using a tensile testing machine Instron 4505 with a controller unit and software from Hegewald & Peschke, Meß- und Prüftechnik GmbH.


67

Tensile Testing of Samples with Old Geometry As a result of high shear stresses and strong cracking of metal sheets perpendicular to the rolling direction during the ARB process, preliminary tensile samples were limited in size. The original tensile sample geometry can be seen in figure 3-6. Fifty millimetres long and fifteen millimetres wide samples were cut out of the ARB sheets in the rolling direction and used for tensile testing. The materials investigated using this sample geometry included aluminium alloys AA1050 and AA6016.

Figure 3-6: Original tensile sample geometry with a gauge length of 12.5 mm.

Experimental Procedure

68

Table 3-10: Experimental matrix for tensile testing

Tensile test geometry of samples produced by two-high

rolling mill (CG and ARB)

New tensile test geometry of samples produced by four-high rolling mill

(CG and ARB)

Friction stir welded samples

(CG and ARB)

Additional heat treatment

(CG and ARB)

Tensile sample geometry (mm3)

15x50x1 12x115x1 12x115x1 12x115x1

Gauge length (mm)

12.5 33.5 33.5 33.5

DIN number / 50125 50125 50125 Material AA1050 AA6016 AA1050 AA6016 Ti G2 Cu 99.99 AA1050 AA6016 AA6016 Strain rate (s-1) 10-3, 10-4, 10-5 10-3, 10-4, 10-5 10-4 10-4

Direction of testing

0° (RD) 0° (RD) 45° 90° (TD)

0° (RD) 45° 90° (TD)

0° (RD) 0° (RD) 90° (TD) 0° (RD)

Testing temperature

RT RT RT RT

Post annealing temperature and time

/ / / 100°C / 1h 200°C / 1h 300°C / 1h

CHAPTER 3


69

Tensile Testing of Samples with New Geometry Once the four-high rolling mill was acquired and the ARB process was optimised, the quality of metal sheets improved and their width increased. The tensile sample geometry was not only standardised (Fig. 3-7), but it was also possible to cut it out from the ARB sheets in different direction, i.e. in the rolling direction RD (0°), at 45° to rolling direction (45°) and perpendicularly to the rolling direction TD (90°). Therefore, the differences of mechanical properties arising from the anisotropy could also be taken into consideration.

Figure 3-7: New and standardised tensile sample geometry according DIN 50125 with a gauge length of 33.5 mm. Tensile Testing of Friction Stir Welded Samples The commercial purity aluminium AA1050 and the aluminium alloy AA6016 in the CG and UFG state were friction stir welded and the tensile test samples were cut in the transverse direction as shown in figure 3-8. The tensile sample geometry complies with DIN 50125. All samples were tested at room temperature and a strain rate of 10-4 s-1.

Figure 3-8: New and standardised tensile sample geometry according DIN 50125 with a gauge. Tensile Testing after Post-Roll Annealing Aiming to improve the ductility of the accumulative roll bonded AA6016 sheets, post-roll annealing was performed on samples at 100 °C, 200 °C and 300 °C for 1 h and quenched in water. The tensile test geometry used conforms to the standards DIN 50125.


70

3.7 Sheet Metal Forming The formability potential of the ultrafine-grained aluminium sheets was investigated using the conventional sheet metal forming techniques, namely hydraulic bulge testing and cup drawing at room and at elevated temperatures, and compared to their conventionally grained counterparts. 3.7.1 Hydraulic Bulge Testing Hydraulic bulge tests were used to investigate the sheet forming potential of commercial purity aluminium AA1050 and aluminium alloy AA6016 before and after accumulative roll bonding. The investigations also included bulge testing on friction stir welded AA1050 and AA6016 sheets in the as-received and accumulative roll bonded state. The experimental matrix for bulge testing can be seen in table 3-11. Table 3-11: Experimental matrix for bulge testing

Temperature (°C)Thickness (mm) 20

0.5 ARB AA1050, ARB AA6016 0.8 ARB AA1050, ARB AA6016 1.0 ARB+FSW AA1050, ARB+FSW AA6016

Although tensile testing is widely adopted in industry and science as a standard method of determining some of the most important mechanical properties, its limitation is that it only predicts the material behaviour under the uniaxial stress state. However, the stress state in many industrial sheet metal forming processes like stamping, bending or deep drawing is not uniaxial. Therefore, bulge tests are necessary for the investigation of the material response to deformation under biaxial stress state, which is closer to a real industrial process and which at the same time complements the uniaxial tensile tests. The results, which can be obtained from bulge tests include: burst pressure, von Mises strain, thickness reduction, pole height, etc.


71

Principal of operation Hydraulic bulge tests were conducted in cooperation with the Chair of Manufacturing Technology at the Friedrich-Alexander University Erlangen-Nürnberg (Lehrstuhl für Fertigungstechnologie) using a hydraulic press Lasco Rig TSP 100S0 with a maximal force of 100 tons (Fig. 3-9 a). a) b)

Figure 3-9: a) Hydraulic press Lasco Rig TSP 100S0 and b) experimental set up. The metal sheet was clamped between the drawing ring with 58 mm diameter and a lower die with a small oil orifice (Fig. 3-9 b). The force exerted from above by the hydraulic press (approximately 500 kN for AA1050 and 900 kN for AA6016) must be high enough to prevent the sliding of the metal sheet between the drawing ring and the lower die. In this manner, the sheet is only stretched and no draw-in occurs. When the fluid, here oil (Marlotherm®), is pressurised from the lower chamber, the sheet deforms or bulges into a dome shape. The pressure increases continuously and fracture occurs once the metal sheet deformation exceeds its formability limit. The advantage of this test method is that the frictional effects are completely eliminated and therefore lower pressures are required for the deformation of the sheets. The relevant testing parameters are listed in table 3-12. Table 3-12: Relevant parameters for hydraulic bulge testing

Size of aluminium blanks

ARB AA6016: 0.5 mm×120 mm×100 mm ARB AA1050, ARB AA6016: 0.8 mm×120 mm×100 mm ARB+FSW AA1050, ARB+FSW AA6016: 1.0 mm×120 mm×100 mm

Die diameter 58 mm Fillet radius of die 5 mm

Clamping force 500 kN (ARB AA1050) 900 kN (ARB AA6016) 750 kN (all friction stir welded samples)

Forming medium Oil: Marlotherm®

Temperature 20 °C


72

During deformation the strain distribution is registered optically using two CCD cameras situated directly above the metal sheet. Individual images of the sample surface are recorded at different load stages and ARAMIS software, GOM Optical Measuring Techniques GmbH, based on the principle of photogrammetry is used for the evaluation. For the purpose of measuring the strain distribution, a stochastic pattern was sprayed onto the sample’s surface. The x and y coordinates of a point within a facet (small quadratic areas) and a grey scale distribution of the facet in a non-deformed state, are defined before the test start (Fig. 3-10). The grey scale gradient describes a local contrast and can be found within every single facet. Since the grey scale gradient of a non-deformed sample equals that of the deformed sample, while the original facet changes its coordinates during deformation, it is always possible to assign the amount of deformation of a facet at any point in time and thus measure the strain.

Figure 3-10: Optical measurement of strain distribution across the sample surface. Left and right sketches represent the surfaces of non-deformed and deformed samples, respectively [Ara00]. Sample preparation for hydraulic bulge tests In order to obtain wider blanks necessary for bulge testing, the original 1 mm thick sheets were rolled perpendicularly to the original rolling direction one to two times after the ARB process. Cross-rolling was performed at room temperature using a two-high rolling mill (Carl Wezel, Mühlacker). All sheets were cut to 100 mm length and 100 mm width. Initially, the thickness of the ARB sheets was reduced to 0.8 mm (AA1050 and AA6016). Later, it was recognised that the aluminium alloy AA6016 severely cracked at the radius of the die, so the sheet thickness was further reduced to 0.5 mm. Friction stir welded samples maintained the thickness of 1 mm.


73

Practical challenges During uniaxial tensile tests, the strain rate is usually predefined before the test and kept constant during testing. During hydraulic bulge tests, it is only possible to measure the strains and/or the strain rate after the test has been conducted. In order to compare the strain rate sensitive UFG materials, it was attempted to keep the strain rate constant by predefining the voltage input of the pressure fluid valve. However, due to the relatively small sample size in comparison to the lower die holding the sample, the voltage input had to be adjusted at times in order to avoid leakage during testing. The voltage input varied by approximately 20 % between both materials. Another way of performing a leak proof test involved positioning of extra material of the same thickness around the test sample. Even though the reproducibility in quality and mechanical properties of the sheets was greatly improved after acquiring the four-high rolling mill, factors such as the sheet thickness variation, waviness and tolerance must still be optimised. The variation of these parameters may lead to statistically broad spectrum of results and therefore defeats the purpose of the forming limit diagrams.


74

3.7.2 Cup Drawing Tests Cup drawing tests were used to evaluate drawability of the as-received, accumulative roll bonded, as-received and friction stir welded, as well as the accumulative roll bonded and friction stir welded samples. The tests were restricted to commercial purity aluminium AA1050 and aluminium alloy AA6016. The experimental matrices can be seen in table 3-13 for the non-welded samples and table 3-14 for the friction stir welded samples. The first set of experiments was conducted on as-received and accumulative roll bonded samples. The tests were performed at room temperature and at 180°C. All samples were cross rolled down to approximately 0.8 mm thickness in order to obtain larger sample diameters. Cross-rolling was performed at room temperature using a two-high rolling mill (Carl Wezel, Mühlacker). All samples were laser cut to different diameter and the maximum diameter which could be obtained from the rolled sheets was 110 mm. A larger sample size was at this stage not possible to produce, because of the limited width of the rolls. Table 3-13: Experimental matrix for deep drawing of as-received and accumulative roll bonded samples, at room temperature and at 180°C

Material Blank diameters used for room temperature experiments (mm)

Blank diameters used for 180°C experiments (mm)

AA1050 N0, N2, N4, N6, N8 Ø 80 Ø 90 Ø 105 Ø 80 Ø 90 Ø 105 Ø 110 AA6016 N0, N2, N4, N6, N8 Ø 80 Ø 90 Ø 105 Ø 80 Ø 90 Ø 105 Ø 110

Three to four blanks with different diameters were cut for each material condition. In order to determine the limit forming ratio using the Schmidt-Kapfenberg method, at least two blank diameters must be successfully drawn and one blank diameter must fracture. If all blanks fracture, it can generally be said that the formability of the material is poor. If all blanks can be successfully deep drawn, the material drawability is satisfactory under these experimental conditions, but larger blank diameter is necessary in order to determine the limiting drawing ratio. Table 3-14: Experimental matrix for deep drawing of as-received friction stir welded and accumulative roll bonded and friction stir welded samples at room temperature

Blank diameters used for room temperature experiments (mm) AA1050N0, N2, N4, N6, N8 Ø 80 Ø 85 Ø 90 Ø 95 AA6016N0, N2, N4, N6, N8 Ø 80 Ø 85 Ø 90 Ø 95

In case of the friction stir welded samples, large blank diameters of 105 mm or 110 mm were not able to be cut out of the sheets due to limited amount of material as well as limited length of the weld. The biggest blank diameter was restricted to 95 mm.


75

Principal of operation The tests were conducted in cooperation with the Chair of Manufacturing Technology at the Friedrich-Alexander University Erlangen-Nürnberg (Lehrstuhl für Fertigungstechnologie) on a hydraulic press Lasco Rig TSP 100S0 with a maximal force of 100 tons. As the punch moves down, a round metal sheet (blank) with diameter d0 is bent over the die radius, and it is drawn into a cylindrical cup shaped component with a specific diameter. The punch or the drawing force and the punch distance are registered during testing. The most important experimental parameters are listed in table 3-15. Table 3-15: Parameters used for deep drawing experiments

Sample thickness (mm) ARB AA1050, ARB AA6016: 0.8 ARB+FSW AA1050, ARB+FSW AA6016: 1.0

Punch diameter (mm) 50 Punch transition radius (mm) 6 Punch velocity (mm/sec) 20 Blank holder load (kN) 20 Blank holder transition radius (mm) 5 Draw ring transition radius (mm) 6 Oil used for room temperature experiments Ometa IHK36

Oil used for 180°C experiments C+H2O Sample heating time at 180°C (min.) 3.5 Temperature of the draw ring (°C) 285 Temperature of the blank holder (C°) 360

It is desirable to obtain good drawability using the biggest possible ratio between the sample diameter (d0) and the punch diameter (d1), the so-called limiting drawing ratio (LDR), without fracturing the bottom of the cup. Therefore, the most important parameters for this test are the maximal punch drawing load, the load required to fracture the base of the cup and the limiting drawing ratio. During deep drawing the blank holder prevents wrinkling of the flange and the centring ring is used to position the blank at the centre of the die and insure symmetrical drawing of the cup. Sample preparation for cup drawing tests As previously mentioned, the as-received and accumulative roll bonded samples were cross rolled down to 0.8 mm thickness using a two-high rolling mill (Carl Wezel, Mühlacker). All friction stir welded samples were 1 mm thick. All samples were laser cut and cleaned in acetone before performing the experiment. The samples tested at room temperature and 180°C were lubricated using chloride free drawing oil (Möllenberg & Sonntag) and “Beruforge 393 G” (Bechem), respectively.


76

3.8 Friction Stir Welding (FSW) The ultrafine-grained materials can be seen as potential structural components in the automotive industry where joinability, besides the mechanical properties and sheet metal forming, plays a major role. Therefore, a special focus was placed on the joinability of the UFG commercial purity aluminium AA1050 and the UFG aluminium alloy AA6016. Friction stir welding was a preferred joining process due to its successful history and the fact that no melting occurs during the process. Since FSW is a solid state joining process, the influence of temperature on the UFG microstructure formed during the ARB process was minimal. The process was conducted at EADS (European Aeronautic Defence and Space Company) in München using a friction stir welding machine FSW5C SuperStir™ from ESAB, Sweden (Fig. 3-11).

Figure 3-11: Friction stir welding machine FSW5C SuperStir™ at EADS. A number of different parameter are important for the process and are summarised in table 3-16. The material used for the pin was a high speed steel HSS E and the approximate length of the pin varied according to the thickness of the sheet between 0.7 mm and 1 mm. The tilt angle of the tool or the nuting angle was 2°. During welding the position was regulated with a closed loop control. Table 3-16: Technical parameters for friction stir welding

Material Sheet thickness

Sheet width

Rotational velocity

of the pin

Vertical force

Transverse velocity of

the pin

Control

(mm) (mm) (rpm) (N) (mm/min.) AA1050 ∼ 1.00 150 800 1900-2400 20 Position AA6016 (Test 1) ∼ 1.00 80 800 2400-3200 150 Position

AA6016 (Test 2) ∼ 1.00 150 800 2400 20 Position


77

3.9 Corrosion Measurements It is well known that welding changes the microstructure of the base material and thus its mechanical properties. In order to understand the corrosion behaviour and predict the corrosion resistance or susceptibility, corrosion measurements were performed on friction stir welded aluminium sheets of the base material and across the weld zone. A relatively new method, based on the electrolytic effect, the so-called EC-Pen method (Electrochemical pen) was used for these investigation purposes. It was especially developed for in-situ quality control of welds. The principal of operation is sketched in figure 3-12. By simply positioning the EC-Pen on the sample surface electrolytic contact is established and electrochemical characterisation is possible. The pen, a polymer body, scans the sample surface and permits the electrolyte (NaCl) to continue to flow out of a container, enabling surface wetting. Its capillary effect prevents the electrolyte from flowing out of the open cell. The electrochemical potential of the surface is influenced by the electric current flow between the surface and the counter electrode. The reference electrode is typically coated with silver choride and it is incorporated within the pen. This method includes a number of advantages: it is quick and easy to use, it requires minimal sample preparation, which corresponts to that of lightmicroscopy, and it can be performed on variaty of geometries and surfaces [Büc02a]. Typical results that can be obtained from such measurements are the polarisation curves and the current density distribution curves across the weld. Figure 3-12: Principal of operation of the electrochemical cell pen, EC-Pen [Büc02b]. This technique allows equilibrium potential measurements, polarisation curve measurements, current density distribution, potentiostatic and galvanostatic measurements and etc. The contact area depends on the size of the pen and is approximately 1.3 mm2. The reference electrode is made up of silver wire coated with a layer of silver chloride (for further details see [Wlo07].


78

The friction stir welded samples in the as-received and the ARB states were prepared similarly as the samples used for the light microscopy. The analysis was conducted across the weld and included the corrosion behaviour of the base material as well as the weld zone (nugget). Additional electrochemical measurements were conducted using the so-called microcapillary technique. Since the usual large-scale electrochemical techniques provide only average data over a large surface area (mm2 to cm2), they are inadequate to investigate local corrosion processes. The microcapillary technique is especially useful, since it allows the study of local processes on metal surfaces. The entire setup is mounted on a microscope allowing for precise positioning of the microcapillary, which is filled with electrolyte. In this case, the diameter of the capillary varied between approximately 50 µm to 500 μm. A layer of silicone rubber is used as the sealant between the front end of the microelectrode and the surface of interest. The microcell is fixed at the revolving nosepiece, replacing an objective, and the specimen is mounted on the microscope stage. In this way, simple, precise, and fast positioning of the microcell is possible. A reference and counter-electrode is connected to the capillary to allow the electrochemical control of the investigated surface [Sut01, Wlo07]. In this work, this technique was used in order to obtain polarisation curves of the local area of interest.

CHAPTER 4 Results

79

4 RESULTS

4.1 ARB Process Optimisation The first accumulative roll bonding tests were carried out on commercial purity aluminium AA1050 using a two-high rolling mill, which strongly restricted the final width of the sheets due to the crack development at the edges and crack propagation perpendicular to the rolling direction (Fig. 4-1). Prior to roll bonding, the edges of the sheets used to be heat treated using a welding torch in order to relieve the stresses and avoid crack propagation further across the sheet. The front and end edges of the sheets were also mechanically joined together using small studs in order to avoid shearing of the sheets during rolling. Furthermore, the two-high rolling mill rolls used for the first experiments had relatively worn surfaces and a thick oxide layer, which contributed to the overall poor surface quality of the rolled sheets. The two top examples in figure 4-1 show AA1050 and AA6016 aluminium sheets with a lot of cracking at the edges and a dull and rough surface finish.

Figure 4-1: Surface finish and aluminium sheet width improvement of commercial purity aluminium AA1050 and aluminium alloy AA6016 by roll bonding using a four-high rolling mill. With an intention to improve the process, a four-high rolling mill was acquired. This contributed not only to the improvement of the final width of the sheets up to 100 mm (Fig. 4-1), but also to the quality of the surface. The variation in thickness along the length of the sheet was reduced (Fig. 4-2) and the surface of the sheets became

CHAPTER 4 Results

80

smoother (Fig. 4-3). The overall deformation became more homogeneous, the crack development was retarded due to smaller deflection of the rolls during the process and the annealing of edges between each ARB cycle was no longer necessary.

Figure 4-2: Variation in sheet thickness of aluminium alloys AA1050 and AA6016 along the sheet width and length obtained during rolling using a two-high and a four-high rolling mill. The improvement of bonding between individual aluminium sheets was also investigated as part of the ARB process improvement. After different surface treatments, including sand blasting, grinding and wire brushing, it was found that the best bonding can be achieved with the roughest surfaces [Klö06]. In addition to that, the quality of the ARB surfaces was analysed by measuring the surface roughness of the sheets produced by a two-high and a four-high rolling mill. From figure 4-3, it can be seen that the surface roughness of the sheets rolled using the four-high rolling mill is significantly lower. Thus, smaller roll diameter or higher rolling pressure leads to an overall better surface quality. Another parameter that plays an important role in the improvement of the ARB process is the process temperature. Aluminium alloy AA6016 was always roll bonded at elevated temperatures due to its higher hardness. Roll bonding at various process temperatures was performed and the difference in hardness evolution can be seen in figure 4-4 for three temperatures: 180 °C, 230 °C and 250 °C. The highest hardness values were achieved by roll bonding at 180 °C, due to slower dynamic recovery. Higher temperatures decrease the potential for a rapid grain refinement and lead to material softening as a result of dynamic recovery and/or partial recrystallisation during pre-heating and rolling.

CHAPTER 4 Results

81

Figure 4-3: Quality improvement of the metal sheet surfaces by means of a four-high rolling mill.

Figure 4-4: Hardness evolution at various process temperatures of the aluminium alloy AA6016 with an increasing number of ARB cycles. However, roll bonding of individual sheets at 180 °C was not satisfactory. The SEM micrograph in figure 4-5 a) shows aluminium alloy AA6016 after 5 ARB cycles and confirmes poor bonding between the sheets at low process temperatures ranging between 180 - 200 °C. On the other hand, roll bonding at higher temperatures of 250 °C results in premature thermal instability, but a better bonding between sheets (Fig. 4-5 b). As a result, the ARB process was generally carried out at an intermediate temperature of 230 °C in order to obtain a compromise between good thermal stability and good bonding between sheets.

CHAPTER 4 Results

82

a) b)

Figure 4-5: SEM micrographs showing interlamellar bonding of AA6016 after 5 ARB cycles at a) 200 °C and b) 250 °C process temperatures [Geb04]. It must be emphasised that there are a number of factors contributing to slightly different mechanical properties of aluminium sheets after rolling using a two or a four-high rolling mill (Fig. 4-6). The major ones are the roll diameter (two-high rolling mill = 130 mm diameter; four high rolling mill = 32 mm diameter) and thus the angular velocity as well as the deformation zone of the rolls and surface roughness. In addition, rolling with a two-high rolling mill required annealing of the edges of the sheets, which was used to suppress the development of cracks. While higher rolling speed of the four-high mill leads to higher deformation rates and a smaller deformation zone to higher pressures, the in-between annealing of the edges, which was necessary in the two-high rolling mill, allowed for prior precipitation evolution and therefore strengthening.

Figure 4-6: Comparison of the hardness evolution between the two-high and the four-high rolling mill (AA6016 rolled at 230 °C). High surface roughness of the sheets and the rapid build-up of the oxide layer during rolling also affect the hardness values and evolution. May [May04] showed that

CHAPTER 4 Results

83

hardness of the ARB sheets increased with an increasing thickness of the oxide layer on the rolls. The development of the oxide layer on the rolls of a two-high rolling mill was more rapid than on the rolls of a four-high rolling mill, because of relatively old and worn out surfaces, higher surface roughness as well as a bigger roll diameter. Thus, the mechanical properties of the sheets produced by a two-high and a four-high rolling mill cannot be directly compared. It is also worth pointing out that the hardness evolution of samples rolled using a two-high rolling mill preceded more quickly within the first 6 ARB cycles, but then subsided more quickly compared to a four-high rolling mill. 4.2 Transfer of the ARB process on other technologically relevant

materials The first ARB experiments were performed on commercial purity aluminium AA1050 and the results and the know-how were then transferred on other naturally hard aluminium alloys of the 5xxx series as well as on the age hardenable aluminium alloys of the 6xxx series. Besides aluminium alloys, materials like oxygen free copper, pure titanium, niobium and magnesium were also investigated. The materials are listed in table 3-1 together with their technical designations. The ARB process of naturally hardenable aluminium alloys of the 5xxx series was investigated as part of a diploma thesis [Geb04] and it was performed using a two-high rolling mill. The ARB process of the aluminium alloys AA5754 and AA5182 was difficult and the ARB samples were usually destroyed after the second ARB cycle. Increasing the process temperature to up to 300 °C did not contribute to any significant improvement of the process. Due to the strong crack development and crack propagation at the edges of the sheets, it was only possible to investigate the alloy in its as-received condition and after the first ARB cycle. Both alloys reached a constant hardness after the first ARB cycle conducted at 300 °C. The hardness values after the first ARB cycle reached approximately 132 HV1 and 144 HV1 for AA5754 and AA5182 aluminium alloys, respectively. The ARB process on these aluminium alloys was not transferred onto a four-high rolling mill, since the main focus of the work was placed on the aluminium alloy AA6016. However, all the other materials, which were investigated, were rolled using a four-high rolling mill. It was also attempted to produce the UFG magnesium sheets from the commercial magnesium alloy AZ31 by employing the ARB process. Magnesium has a hexagonal crystal structure and therefore only a limited number of slip systems, which are active at room temperature. This made the production of the UFG magnesium sheets and the overall ARB process much more difficult in comparison to aluminium alloys. In order to improve the deformation capability of magnesium, the process was conducted at elevated temperatures up to 350 °C. Further increase of the process temperature was not possible due to the maximum temperature specification of the rolls. The actual limit of the process temperature was 250 °C. The magnesium sheets broke into thin strips after leaving the rolls at room temperature as well as at elevated temperatures (Figure 4-7 a-d). However, some improvement of the process can be seen when roll bonding at 350 °C, although this was not sufficient to be able to perform the complete

CHAPTER 4 Results

84

ARB process (Figure 4-7 e). In conclusion, it can be said that in order to perform the ARB process of high strength alloys or alloys with a hexagonal crystal structure, it is necessary to use a high temperature rolling mill.

Figure 4-7: Accumulative roll bonding of magnesium alloy AZ31 at different process temperatures ranging from 20 °C to 350 °C. Other materials like copper, niobium and titanium were investigated for different reasons. These materials were used as model materials and the results obtained from roll bonding were used as reference tests for parameter studies (see chapter: experimental procedure). Oxygen free copper (Cu 99.99) was roll bonded at room temperature up to 8 ARB cycles using a four-high rolling mill with a thickness reduction slightly greater than 50 %. In figure 4-8, it can be clearly seen that the sheets have a good surface quality and that there is only a limited amount of crack development at the edges of the sheets even at higher ARB cycles.

Figure 4-8: Accumulative roll bonded oxygen free copper sheets, rolled at room temperature using a four-high rolling mill and approximately 50 % thickness reduction.

CHAPTER 4 Results

85

Pure titanium grade 2, Ti 99.2, had to be roll bonded at a temperature of 250 °C, because of the relatively high initial hardness of 125 HV5. Due to its high initial hardness and high strain hardening, it was also necessary to perform the ARB process at somewhat higher thickness reductions of more than 50 % per cycle. Generally, the ARB process on titanium grade 1 and 2 was successful and it was able to be performed up to 8 ARB cycles. Titanium and aluminium were roll bonded together in order to assess the material properties of Ti and Al laminate composites produced by the ARB process. The idea for this work came from the abalone shell which shows an extraordinary strength and toughness attributed to its fine laminate structure. The aim was to produce the laminates by the ARB process using Ti-Al as well as two cladded raw materials of Al-Ti and Al-Ti-Al, and investigate the mechanical properties. There were two main reasons for the poor success of roll bonding of the cladded materials. The Ti-Al cladding sheets of 0.5 mm thickness deformed during preheating in the furnace at 300 °C, making the ARB process impossible to conduct. The second reason was the poor bonding between the commercial purity aluminium AA1050 and titanium side of the cladding sheet. The thin titanium layer from the cladding used to stick to the rolls, which resulted in poor bonding between the two sheets, i.e. between 1 mm thick AA1050 sheet and the cladded 0.5 mm thick sheet. High purity niobium, Nb 99.8 was rolled at 250 °C and 50 % thickness reduction using a four-high rolling mill. The ARB process was only able to be conducted up to 3 ARB cycle due to the high strain hardening of the material. Further temperature increase was not possible due to the reasons already mentioned.

CHAPTER 4 Results

86

4.3 Mechanical Characterisation Mechanical characterisation in terms of hardness and tensile tests was performed on the as-received and accumulative roll bonded samples. 4.3.1 Hardness Tests Hardness of ARB processed Aluminium Alloys Hardness tests were conducted across the width of the sheet in the RD-TD plane using a Vickers hardness tester (Dia Testor 2N, Wolpert). Figure 4-9 shows that the hardness of both materials after approximately 4 ARB cycles is doubled compared to the solution treated state. The accumulative roll bonding of the aluminium AA1050 was conducted at room temperature, while the aluminium alloy AA6016 was processed at 230 °C for better interlamellar bonding. The saturation in hardness of AA6016 is reached after approximately 5 ARB cycles and is delayed in comparison to AA1050, where the saturation is reached after 2 ARB cycles. This effect is clearly due to the influence of the process temperature. Furthermore, earlier work has shown that a compromise between good thermal stability and interlamellar bonding can be achieved by choosing the appropriate process temperature [Top07]. Process temperature of 230°C was found to be the most suitable process temperature for aluminium alloy AA6016.

Figure 4-9: Evolution of hardness vs. number of ARB cycles for commercial purity aluminium AA1050 processed at room temperature and aluminium alloy AA6016 processed at 230 °C. The error bars represent standard deviation from the average of six measurements. Both materials were processed using a four-high rolling mill.

CHAPTER 4 Results

87

Hardness Distribution in Cross Section It is well know that the microstructural evolution during the ARB process varies through the thickness of the sheet due to the inhomogeneous shear strain distribution, already described in the literature review. During the first ARB cycle, the surfaces in contact with the rolls and at the interface of two sheets are exposed to high frictional forces. In addition, the sheet is exposed to high compressive, tensile and shear stresses. The microstructural inhomogeneity and shear strain distribution is especially dominant after the first ARB cycle. This effect can be clearly seen in figure 4-10 in terms on hardness measured through the thickness of the sheets. There is an overall increase in hardness after the first ARB cycle in comparison to the as-received state AA6016-N0. However, the hardness after the first cycle is higher at the surfaces than in the middle of the sheet. After subsequent rolling, the surfaces with the highest shear strain migrate to the middle during the second cycle. This leads to a complex combination of plane strain and shear deformation, depending on the location and the number of ARB cycles [Lee02a]. However, after a sufficient number of ARB cycles, the shear strain distribution as well as the grain size become more homogeneous. Similar observations were made in this work in terms of hardness, where the hardness inhomogeneity seems to decrease with an increase in number of ARB cycles (see AA6016-N4 and AA6016-N8).

Figure 4-10: Hardness distribution through the sheet thickness of aluminium alloy AA6016 of the as-received condition (N0) and after 1, 2, 4, and 8 ARB cycles.

CHAPTER 4 Results

88

Hardness of other ARB processed materials The materials, which were successfully accumulative roll bonded were commercial purity aluminium AA1050, aluminium alloy AA6016, oxygen free copper, pure titanium and niobium. The development of hardness with the number of ARB cycles is shown in figure 4-11 for each material. Irrespective of the ARB process temperature and process parameters, all materials show rapid grain refinement and the hardness increases with the growing number of ARB cycles (Fig. 4-11). Hardness of the ARB processed commercial purity aluminium AA1050 and aluminium alloy AA6016 increases up to 2.5 times in comparison to the as-received solutionised state. A similar behaviour was observed for copper. Oxygen free copper (Cu 99.99) was roll bonded at room temperature up to 8 ARB cycles and 50 % thickness reduction. Figure 4-11 shows that the hardness increases up to 2.5 times already after 2 ARB cycles in comparison to the as-received state. A saturation level is reached after 2 ARB cycles and stays constant up to 8 ARB cycles.

Figure 4-11: Development of hardness of different ARB processed materials with increasing number of ARB cycles. Note that the materials were processed at different process temperatures. Process temperature and homologous temperatures are indicated in the diagram. Pure titanium (Ti 99.2) had to be roll bonded at an elevated temperature of 250 °C due to its high initial hardness of 125 HV5. A significant hardness increase from 125 HV5

CHAPTER 4 Results

89

to 275 HV5 was observed after 4 ARB cycles. Despite the elevated process temperature, the hardness values after 4 ARB cycles are 2.2 times higher than those of the as-received state. Niobium was also roll bonded at 250 °C. During roll bonding, the material showed much strain hardening and the ARB process was only able to be repeated 3 times. Further increase of the process temperature was not possible due to the temperature restriction imposed on the rolls. The hardness values continually increase from 50 HV5 to 150 HV5 after 3 ARB cycles.

CHAPTER 4 Results

90

4.3.2 Uniaxial Tensile Tests Tensile testing was conducted on commercial purity alumnium AA1050, aluminium alloy AA6016, oxygen free copper Cu 99.99 and pure titanium Ti Grade 2. Aluminium alloys were tested in the rolling direction and transverse direction, as well as at 45° to the rolling direction. Copper and titanium were only tested in the rolling direction. All tensile tests were performed at room temperature using three different strain rates of 10-5 s-1, 10-4 s-1 and 10-3 s-1. Characteristic Stress-Strain Curves of ARB processed Aluminium Alloys Representative tensile stress-strain curves of the samples tested in the rolling direction can be seen in figures 4-12 a) and b). A general trend is that yield and tensile strength, as well as elongation to failure progressively increase with increasing number of ARB cycles. After only two ARB cycles, the yield strength can be increased by more than 100 %, but elongation to failure is compromised by up to 90 % in comparison to the as-received state. However, further ARB rolling can considerably improve elongation to failure of both materials. The difference in elongation to failure between the as-received and the ARB processed samples seems to be much bigger in the AA6016 than in the AA1050 material. In order to understand this behaviour, different initial states of AA1050 and AA6016 have to be considered. As already mentioned, the as-received state of commercial purity aluminium AA1050 is a cold rolled state. The initial, as-received state of aluminium alloy AA6016 is a solutionised state. Therefore, considerable decrease of elongation to failure of AA6016 is correlated with high plastic deformation of the material after a single ARB cycle. A more appropriate comparison of the elongation to failure of AA1050 and AA6016 can be made by comparing the N2 state of both materials. After approximately 4 ARB cycles, ultrafine-grained aluminium alloy AA6016 shows higher yield and tensile strength than the T8X state (paint-bake state), which is usually applied to the freshly solutionised and stretched sheets to maximise the strength by age hardening. From figure 4-12 b) it is evident that the yield strength of AA6016 increases by up to three times compared to the as-receied T4 state with increasing number of ARB cycles. The difference between yield and tensile strength values is only pronounced in the as-received or the paint baked T8X state, but diminishes upon further ARB rolling. This clearly indicates that the strain hardening potential of aluminium alloy AA6016 decreases with increasing number of rolling cycles. In both materials, limited strain hardening can be attributed to a small grain size. By switching to the nano scale, the classical deformation behaviour of crystalline materials may be changed as other deformation processes, like grain boundary sliding, dislocation annihilation or twinning become dominating. The ARB structure has a high dislocation density which affects the deformation behaviour as moving dislocations are normally not stored in the grain interior, but rather at the grain boundaries. Consequently dynamic recovery takes place mainly along grain boundaries. Hence, the strain hardening capability is reduced as dislocation annihilation and recovery can easily take

CHAPTER 4 Results

91

place at grain boundaries. Furthermore, recovery processes also have to be considered since solid solution hardening and precipitation morphology can change at elevated processing temperatures in hardenable aluminium alloys such as AA6016.

a)

b)

Figure 4-12: Representative engineering stress/strain curves in the rolling direction of a) commercial purity aluminium AA1050 processed at room temperature and b) aluminium alloy AA6016 processed at 230°C. Both materials were processed using a four-high rolling mill. The tensile tests were performed at room temperature and a strain rate of 10-4 s-1.

CHAPTER 4 Results

92

Characteristic Stress-Strain Curves of ARB processed Copper and Titanium Figures 4-13 a) and b) show the characteristic stress/strain curves of the accumulative roll bonded copper and titanium compared to the as-received states. Tensile curves of ARB processed copper indicate a tremendous increase in yield and tensile strength after only two cycles (N2). The yield strength and the ultimate tensile strength increase in comparison to the as-received state by a factor of five and by a factor of two, respectively. However, the elongation to failure of the same sample is significantly sacrifised from 34.5 % in the as-received state (“soft state”) to only 2 % after two ARB cycles. Note that the initial state of OFE copper (“soft state”) differs from that of aluminium AA1050 (cold rolled). Further roll bonding of copper can lead to an improvement of elongation to failure, as it is the case for N4 and N8. However, OFE copper shows slightly different characterstics in comparison to other ARB processed materials, which usually show an increase in elongation to failure and an increase in strength with an increase in number of ARB cycles. Accumulative roll bonded copper on the other hand, shows a definate decrease of strength with the growing number of ARB cycles. It seems that the material experiences cycles of high strength and low elongation to failure (eg. N2) followed by slightly lower strength and higher elongation to failure after the following two ARB cycles (eg. N4). The same trend is evident for samples N6 and N8. It appears that the higher number of ARB cycles (eg. N12) do not contribute to a significant improvement of strength or ductility. In fact, the tensile test curve of the sample processed to 12 ARB cycles corresponds to the tensile test curve of 4 ARB cycles, as it can be seen in figure 4-13 a. A possible reason for the somewhat different mechanical characteristics of OFE copper 99.99 may be due to the high purity of the material. During the first few ARB cycles, the material experiences a large amount of plastic deformation, which is stored within the material in form of dislocations and dislocation cells. Upon further roll bonding, the high level of stored energy increases the driving force towards a more stable material state, which can be achieved by the recovery process. Since there are no alloying elements or precipitates within the high purity copper, the material is able to recover more quickly. The process temperature which develops as a result of friction also contributes to easier recovery. However, until now this was not verified. It is also important to mention that the tensile tests were repeated more than once on samples obtained from different copper sheets after the ARB process and the tensile curves were reproducible. The tensile curves of ARB processed titanium show a similar behaviour as many other ARB processed materials including aluminium alloy AA6016 (Fig. 4-13 b). The strength is significantly increased with increasing number of ARB cycles, while the elongation to failure decreases. The yield strength and the ultimate tensile strength increase in comparison to the as-received state by a factor of 3.2 and 2, respectively. It is important to note that even though the elongation to failure decreases, it remains by approximately 8-10 % after N2 and N4 cycles. Therefore, a decrease of the elongation to failure is not as significant as in accumulative roll bonded high purity copper.

CHAPTER 4 Results

93

a)

b)

Figure 4-13: Representative stress/strain curves in the rolling direction of a) OFE copper (Cu 99.99) processed at room temperature and b) titanium grade 2 (Ti 99.2) processed at 250°C. Both materials were processed using a four-high rolling mill. The tensile tests were performed at room temperature and a strain rate of 10-4 s-1.

CHAPTER 4 Results

94

Strain rate sensitivity In contrast to the conventionally grain-sized materials (CG), UFG materials including commercial purity aluminium AA1050 and aluminium alloys are strain rate sensitive [Höp04, May05, Dal04]. An increase in strain rate causes an increase in strength and a decrease in fracture strain. This effect is also important for sheet metal forming tests like the bulge or deep drawing test where the strain rate sensitivity influences the deformation capability and higher elongation to failure could be obtained at lower strain rates or elevated temperatures. a)

b)

Figure 4-14: Strain rate sensitivity of as-received and ARB processed a) commercial purity aluminium AA1050 and b) aluminium alloy AA6016. The AA1050 and AA6016 samples were produced by a four-high and a two-high rolling mill, respectively. All tensile tests were performed at room temperature and strain rates of 10-3 s-1, 10-4 s-1 and 10-5s-1.

CHAPTER 4 Results

95

Figures 4-14 a) and b) show tensile test curves of commercial purity aluminium AA1050 and aluminium alloy AA6016 at three different strain rates. Both figures show that there is no significant difference of the as-received samples tested at different strain rates. It can therefore be said that the conventionally grained CG materials are strain rate insensitive. On the other hand, the ultrafine-grained samples show a clear increase of strength and reduced ductility at higher strain rates (eg. 10-3 s-1) and considerably lower strangth and higher ductility at lower strain rates (eg. 10-5 s-1). Therefore, these UFG materials are highly strain rate sensitive even at low homologous temperatures. A comparison of strain rate sensitivity exponents between CG and UFG materials is given in table 4-1. Strain rate sensitivity exponents of the accumulative roll bonded AA1050 and AA6016 aluminium sheets were calculated from tensile tests performed at room temperature and strain rates of 10-3 s-1, 10-4 s-1 and 10-5 s-1. The results were subsequently compared to those found in the literature. As already mentioned in the literature review, strain rate sensitivity is strongly dependent on the grain size, the material purity and the testing temperature or strain rate. From table 4-1, it can be seen that the strain rate sensitivity exponent is generally higher for UFG materials than for CG materials. It also increases with increasing temperature or decreasing strain rate. Table 4-1: Comparison of strain rate sensitivity exponent m between CG and UFG materials at different temperatures

ARB processed materials ECAP processed materials AA1050 AA6016 AA8011 AA1050 AA1050 AA6061 Tensile

tests Tensile

tests Tensile

tests Compression

tests Compression

tests Compression

tests ε& = 10-5 -

10-3 (s-1) ε& = 10-5 - 10-3 (s-1)

ε& = 10-4 - 10-3 (s-1)

ε& = 10-5 - 10-3 (s-1)

ε& = 10-5 - 10-3 (s-1)

ε& = 10-5 - 10-3 (s-1)

m @ RT m @ RT m @ RT m @ RT m @ 100°C m @ 100°C

CG 0.034 0.005 0.01-0.015 [Kim05]

0.004 [Höp05]

0.005 [Höp05]

0.005 [Vev08]

UFG 0.040 0.0125 0.047 [Kim05]

0.014 [Höp05]

0.06-0.09 [Höp05]

0.021 [Vev08]

Due to the high dislocation density and dislocation storage at grain boundaries in UFG materials, statistically it becomes highly likely for the dislocation annihilation to take place in UFG materials even at low homologous temperatures. Thus, strain rate sensitivity in UFG materials is enhanced due to higher fraction of grain boundaries, which represent the likely locations for the dislocation activity, storage as well as annihilation. Additionally, it is believed that the strain rate sensitivity increases due to a decrease in activation volume in the nanocrystalline regime. Conventional FCC metals have a large activation volume (V α 102-103·b3), which is associated with dislocations cutting through forest dislocations. On the other hand, the activation volume for grain-boundary diffusion processes is much lower (V α 1-10·b3) [Mey06]. Because of the small size of the crystals the interfaces form an extremely dense network of paths for fast diffusion through the nanocrystalline material [Hor87].

CHAPTER 4 Results

96

Higher strain rate sensitivity of UFG AA1050 in comparison to UFG AA6016 is reflected in the higher ductility of UFG AA1050 previously observed from the tensile tests curves (Fig. 4-12 a and b). At this point, lower m values of UFG AA1050 in comparison to UFG AA6061 must further be clarified. The results may differ due to the fact that the materials were processed by ECAP and the fact that the m values were determined from the compression tests. In case of the ARB processed materials, the exponent m is significantly higher for AA1050 than for AA6016, in the CG and UFG conditions. This effect is believed to be contributed by the material purity and the ease of material recovery. In pure materials the dislocation annihilation would occur more easily, because it is not obstructed by alloying elements and second phase particles, as the case may be for AA6016. Dynamic recovery can therefore occur more easily in single phase materials such as AA1050 and AA8011 [Kim05], because of the enhanced cross-slip of screw dislocations and higher stacking fault energy. The effect of strain rate sensitivity becomes even more dominant for UFG materials tested at elevated temperatures. The grain boundary diffusion becomes more enhanced because of the extremely dense network of paths for fast diffusion through the nanocrystalline material and the effect of alloying elements and second phase particles is reduced. The dislocations are now able to overcome the obstacles more easily, while the precipitates may loose their pinning effect due to overaging. In this case, the process becomes dependent on the thermally activated dislocation annihilation controlled by the dislocation climb at grain boundaries. This was reported by Höppel et al. [Höp05] for AA1050 commercial purity aluminium and Vevecka-Priftaj et al. [Vev08] for hardenable aluminium alloy AA6061, both tested up to 250°C. Höppel et al. observed an increase of the SRS exponent in AA1050 from room temperature to 250°C by a factor of 6.25 and 17.8 for CG and UFG states, respectively. From table 4-1 it can be seen that there is no pronounced difference in SRS exponents of AA1050 and AA6061 at room temperature and up to 100°C. The m values obtained for AA6061 at 100°C are also comparable with those of AA6016 at room temperature. However, one should bear in mind that these two alloys belong to the same series of hardenable aluminium alloys, but they have different compositions. Additionally, their SRS exponents were measured using different techniques. It was found that strain rate sensitivity exponents of materials with similar purity, UFG AA1050 (99.5% purity) and UFG AA8011 (99.0% purity) are comparable and average 0.04 and 0.047, respectively. However, m values of CG AA1050 (m = 0.034) are considerably higher than those of CG AA8011 (m = 0.040). This is most probably due to different material conditions. Commercial purity aluminium AA1050 was cold rolled and AA8011 was solution treated. Nevertheless, these values are one order of magnitude higher than the values calculated for AA1050 from the compression tests [Höp05]. Thus, it appears that the method for determining the SRS exponent also plays an important role. In case of the ARB materials, m values were determined from three samples, each tested at different strain rates. On the other hand, the compression test is a more advantageous method for measuring the SRS effect, because the m values can

CHAPTER 4 Results

97

be determined from the strain-rate jump tests on a single sample and hence minimise the effects of microstructural inhomogeneity. a)

b)

Figure 4-15: Strain rate sensitivity of as-received and ARB processed a) OFE copper (Cu 99.99) and b) titanium grade 2. All tensile tests were performed at room temperature and strain rates of 10-3 s-1, 10-4 s-1 and 10-5s-1. Like most conventionally grained materials, CG copper does not appear to be strongly strain rate sensitive, although some differences in elongation to failure between the three strain rates were evident. However, the UFG state after 12 ARB cycles indicates some strain rate sensitivity (Fig. 4-15 a), although the behaviour differs from that of

CHAPTER 4 Results

98

commercial purity aluminium AA1050 or aluminium alloy AA6016. In this case, higher strain rates lead to higher strength as well as higher ductility, whereas lower strain rates show lower strength and lower ductility, for room temperature testing. This behaviour deviates from most literature results regarding strain rate sensitivity and the exact reasons for this are not yet clear. It is however possible that the material purity may have considerable effects. Conventionally grained titanium as well as the ultrafine-grained titanium were both found to be strain rate insensitive at room temerature tensile testing (Fig. 4-15 b). Since titanium’s homologous temperature is relatively low, it is believed that the strain rate sensitivity may become more pronounced at higher testing temperatures. Influence of Anisotropy on Mechanical Properties of AA1050 and AA6016 The evolution of mechanical properties in the rolling direction (RD) and the transverse direction (TD) with an increasing number of ARB cycles can be seen in figures 4-16 a) and b). As shown earlier, yield and tensile strength, as well as elongation to failure progressively increase with the growing number of ARB cycles. A growing divergence of yield and tensile strength curves after the first ARB cycle of AA1050 suggests a growing strain hardening potential. The saturation level is reached after 2 ARB cycles and is attributed to a relatively fast dynamic recovery process during rolling. Similar trends and values of yield and tensile strength as well as elongation to failure in the rolling and transverse direction of AA1050 were also reported by Kim et al. [Kim06]. However, it should be noted that the authors investigated the improvement of ductility by cross rolling performed at elevated temperatures and conducted the tensile tests at a strain rate of 10-3 s-1. From figure 4-16 a) and b) it is evident that samples in the transverse direction (TD) show no significant deterioration of strength or elongation to failure in comparison to the samples in the rolling direction (RD), even though a strongly elongated microstructure was observed for both materials.

CHAPTER 4 Results

99

a)

b)

Figure 4-16: Yield strength, ultimate tensile strength and elongation to failure, in the rolling (full symbols) and transverse direction (half full symbols) vs. the number of ARB cycles N for a) commercial purity aluminium AA1050 processed at room temperature and for b) aluminium alloy AA6016 processed at 230 °C. Both materials were processed using a four-high rolling mill. The tensile tests were performed at room temperature and a strain rate of 10-4 s-1. The error bars represent the standard deviation of an average of four to six measurements.

CHAPTER 4 Results

100

However, the tensile tests of samples taken at 45° to the rolling direction showed surprising results. From figures 4-17 a) and b), it can clearly be seen that there is no considerable difference in the yield and tensile strength between the samples taken in the rolling direction and at 45° to the rolling direction. However, the samples taken at 45° to the rolling direction (dashed lines) show much higher elongation to failure, of more than 50 %, in comparison to the samples taken in the rolling and transverse direction. In addition, the elongation to failure increases continuously with an increase in number of ARB cycles and it does not reach a typical saturation level after approximately 4 to 6 cycles, as is usually the case for the samples oriented in the rolling direction. a)

b)

Figure 4-17: Yield strength, ultimate tensile strength and elongation to failure, in the rolling direction (full symbols) and at 45° to the rolling direction (half full symbols) vs. the number of ARB cycles N for a) commercial purity aluminium AA1050 processed at room temperature and for b) aluminium alloy AA6016 processed at 230 °C. Both materials were processed using a four-high rolling mill. The tensile tests were performed at room temperature and a strain rate of 10-4 s-1. The error bars represent the standard deviation of an average of four to six measurements.

CHAPTER 4 Results

101

One possible reason for the higher elongation to failure of samples oriented at 45° to the rolling direction might be related to the crystallographic texture. With an increase in number of ARB cycles, the texture develops into a typical rolling texture, which usually occurs in face centred cubic (fcc) metals with high stacking fault energy. The AA6016 samples show a characteristic ß-fibre texture with a Cu component after 8 ARB cycles, see [Skr07] for details. It was shown that the anisotropy values of the as-received state in the rolling direction (rRD) and transverse direction (rTD) are higher than those at 45° to the rolling direction. However, after the ARB process, this relationship was reversed. The anisotropy values (in terms of the Lankford parameter r) of samples taken at 45° to the rolling direction increased with increasing number of ARB cycles and reached a maximum at 8 ARB cycles (Fig. 4-18). This indicates a possible change in the strain state during tensile testing and therefore different values of the elongation to failure. Another possible explanation is that the earlier mentioned process of thermally activated recovery of dislocations at grain boundaries [Höp04] is triggered by the ß-fibre texture.

Figure 4-18: Lankford parameter r calculated for tensile deformation of AA6016 in different directions as a function of the number of cycles [Skr07]. Authors also investigated the normal anisotropy <r> and the variation of anisotropy Δr for the as-received and ARB processed AA6016 aluminium alloy. The deep drawing conditions of the sheets can be estimated from the magnitudes of <r> and Δr. Larger <r> values result in reduced thinning during deep drawing, while smaller Δr values result in reduced earing. As can be seen in figure 4-19 a), <r> increases steadily with increasing number of ARB cycles, while Δr changes its sign. Therefore, the initial and high-cycle ARB state are best suited in order to avoid sheet thinning. In contrast, the low-cycle ARB state should not result in any earing. From figures 4-19 a) and b), it can clearly be seen that the best compromise between these two parameters can be achieved for samples rolled up to 4 ARB cycles.

CHAPTER 4 Results

102

a) b)

Figure 4-19: (a) Coefficient of normal anisotropy <r> and (b) its variation Δr of AA6016 as a function of the number of ARB cycles calculated using two different models [Skr07]. The Importance of Rolling Mill/Tensile Test Geometry The first accumulative roll bonding tests were carried out using a two-high rolling mill, which strongly restricted the final width of the sheets due to crack development at the edges and crack propagation perpendicular to the rolling direction. In this case, the tensile sample geometry was restricted to a gauge length of 12.5 mm. With the acquirement of a four-high rolling mill, the width of the sheets was considerably improved and standardised tensile test sample geometry (DIN 501253) with a gauge length of 33.5 mm was used for all further investigation purposes. Figure 4-20 shows some differences of the tensile test curves of the samples produced by a two-high and a four-high rolling mill. The samples in the as-received state do not show a significant difference in the stress level, because they have not been roll bonded and the effect of the roll geometry and speed can be neglected. However, these samples do indicate a variation in elongation to failure, which is believed to predominantly be due to the tensile sample geometry. Other roll bonded samples like N2 and N4 show a clear difference in the achievable stress level as well as the elongation to failure. The difference in sample strength between the two-high and four-high rolling mill may be due to different surface friction of the rolls as well as the oxide layer on the rolls. In the diploma thesis of May [May04], it was shown that the oxide layer which gradually develops on the surfaces of the rolls also affects the mechanical properties of the ARB samples. With an increase of oxide layer on the rolls, the hardness of the rolled samples also increases. At this stage it is also important to mention the role of tensile sample geometry. Many researchers are constrained to use small tensile samples with a gauge length of not more than 10 mm, due to the originally limited size of the ARB sheets. In short samples, necking take place over the entire gauge length, and in long samples it is

CHAPTER 4 Results

103

restricted to a small fraction of the gauge length. Since the measured elongation to failure represents an average strain or strain distribution over the entire gauge length of the sample, the elongation to failure of short samples will always be larger than in the long ones [Iss95]. Similarly, Zhao et al. [Zha06] advise to characterise the ductility in terms of uniform elongation (strain before necking) rather than elongation to failure, because the elongation to failure gives a false impression of good ductility especially for samples with a small gauge length of less than 5 mm.

Figure 4-20: Differences in mechanical properties between the two-high rolling mill and the four-high rolling mill of the as-received and ARB processed aluminium alloy AA6016. Besides the obvious geometrical differences regarding the sample geometry and more specifically regarding the gauge length, it must be emphasised that there are a number of factors contributing to slightly different mechanical properties of aluminium sheets after rolling using the two-high rolling mill or the four-high rolling mill. The major ones are the roll diameters (two-high rolling mill = 130 mm diameter; four high rolling mill = 32 mm diameter) and thus the angular velocity as well as the deformation zone, surface roughness and the annealing of sheet edges used to suppress the development of cracks. Higher rolling speed of the four-high mill rolls led to higher deformation rates, while the small deformation zone of the rolls increased the roll pressure. The annealing of sample edges, which was necessary when rolling using the two-high rolling mill caused premature precipitation evolution and strengthening. Surface roughness of the rolls and metal sheets would also affect the hardness values and evolution. Therefore, the mechanical properties of the sheets produced by the two-high and the four-high rolling mill can not be directly compared. It is, however, worth

CHAPTER 4 Results

104

pointing out that the hardness evolution of samples rolled using the two-high rolling mill preceded more quickly within the first 6 ARB cycles, but then subsided more quickly than in the case of the four-high rolling mill. At this stage, it is not possible to separate the influence of the sample geometry and the rolling mill on the final mechanical properties of the ARB samples. However, for future references it is recommended to investigate the influence of the geometry and use the minimum required sample size in order to eliminate the geometrical and especially the gauge length influence of the elongation to failure. Tensile tests of UFG AA6016 samples annealed after roll bonding In order to improve the elongation to failure of UFG aluminium alloy AA6016, post-roll annealing was performed at 100°C, 200°C and 300°C. The total time of annealing averaged one hour. Most UFG samples showed no considerable improvement of the elongation to failure between 100°C and 200°C. The maximum elongation to failure reached approximately 5 %. Material softening, associated with enhanced ductility was evident after annealing at 300°C, is believed to be due to the recrystallised microstructure reported in figure 4-32. The UFG AA6016 samples reach a yield and tensile strength of approximately 50 MPa and 100 MPa, respectively and an elongation to failure of 25 %. It can therefore be concluded that it is not possible to increase the elongation to failure of the ARB processed AA6016 samples by a simple post-roll annealing treatment.

CHAPTER 4 Results

105

4.4 Microstructural Characterisation The microstructural characterisation was performed on the as-received and accumulative roll bonded commercial purity aluminium AA1050, age hardenable aluminium alloy AA6016, oxygen free copper, pure titanium and niobium. The microstructural evolution during the ARB process is discussed and the similarities and differences between the materials are pointed out shorty. Finally, the characteristic fracture surfaces of accumulative roll bonded samples after tensile testing are presented and discussed. 4.4.1 As-received and ARB processed materials Typical microstructures of the as-received and the accumulative roll bonded samples can be seen in figure 4-21 for all materials investigated. On the right hand side, the graphs show the cumulative frequency of the grain size distribution of CG and UFG samples. The grain size was determined using the line intercept method and the median grain size of the UFG samples was measured perpendicular to the rolling direction. The term grain size is used disregarding any differences in the misorientation between adjacent grains. The grain size of the as-received samples is mostly homogeneous and ranges between 10 µm to 50 µm (Fig. 4-21 a, d, g, j, m). During ARB the grains become strongly elongated in the rolling direction with a relatively big aspect ratio (grain length to grain width) and dislocation substructures inside the grains. A typical elongated, ribbon like structure develops already during the first few cycles, while the ultrafine-grained microstructure emerges after 3 to 4 ARB cycles. This was confirmed for all materials investigated in this work and can be seen in figures 4-21 b), e), h), k), and n), showing example microstructures of all five materials. Strong microstructural refinement occurs as a result of the repeated high plastic shear strain during the ARB process. After 6 to 8 ARB cycles a constant grain size is normally established and no significant grain size reduction occurs upon further roll bonding. Figures 4-21 b) and e) show the TEM micrographs of the commercial purity aluminium AA1050 and aluminium alloy AA6016 after 8 ARB cycles processed at room temperature and at 230 °C, respectively. The median grain size, measured perpendicular to the rolling direction reaches 324 nm (Fig. 4-21 b) and 218 nm (Fig. 4-21 e) for the AA1050 and AA6016, respectively. Even though the AA6016 was roll bonded at a higher process temperature, the final grain size remained smaller than in the commercial purity aluminium AA1050 processed at room temperature, also reported by Höppel et al [Höp04]. This may be due to slower dynamic recovery, retarded by the alloying elements and the Mg2Si precipitates. Similar observations were reported by Pérez-Prado et al. [Per04] for accumulative roll bonded Mg–Al–Zn alloy, where the smallest grain size was attributed to the high aluminium content and the precipitation pinning of grain boundaries. Even though the solutionised state of the AA6016 alloy is used for the ARB process, the development of the Mg2Si precipitates during the process takes place due to elevated process temperatures. The AA6016 alloy ages at room temperature and ARB processing at higher temperatures like 230

CHAPTER 4 Results

106

°C increases the driving force for precipitation and reduces the time needed for overageing. However, higher processing temperatures are necessary in order to achieve good bonding between the individual layers of metal sheets. The precipitates may also have an effect on the grain aspect ratio in AA6016. The differences in microstructures between AA1050 and AA6016 can clearly be seen in figure 4-21 b-e.

Commercial purity aluminium AA1050 a) b) c)

Aluminium alloy AA6016 d) e) f)

Oxygen free copper Cu 99.99 g) h) i)

CHAPTER 4 Results

107

Titanium Ti G2 (Ti 99.2) j) k) l)

Niobium Nb 99.8 m) n) o)

Figure 4-21: Left: Light microscopy pictures of as-received (CG) samples (a-e); middle: TEM micrographs of UFG samples (f-j); middle top right hand corner: TEM diffraction patterns of UFG samples (k-o); right: cumulative frequency of the grain size distribution of CG and UFG samples as seen in pictures on the left and middle (a-j). Note: figure a) is the only SEM micrograph. After 8 ARB cycles the UFG oxygen free copper (Fig. 4-21 h) shows a similar microstructure as the UFG aluminium alloys AA1050 and AA6016 (Fig. 4-21 b and e). The grains are largely elongated in the rolling direction and have high aspect ratios, while the grain size differs. Commercial purity aluminium AA1050 shows the biggest grain size of > 300 nm (Fig. 4-21 c) after 8 ARB cycles in comparison to the aluminium alloy AA6016 (Fig. 4-21 f) and oxygen free copper (Fig. 4-21 i), that reach 218 nm and 160 nm, respectively. The microstructural similarities are largely due to the same, face centred cubic crystal structure. The differences in grain size arise as a result of alloying elements, second phase particles or stacking fault energy, which all influence the dynamic recovery process of the material. Higher alloying element content and second phase particles slow down the recovery process, while higher stacking fault energy leads to easier recovery. Pure materials without second phase particles are therefore expected to have the biggest grain size, because the grain boundaries can not be stabilised by the second phase particles. This can be confirmed from the grain size measurements on AA1050 and AA6016. Even though OFE copper

CHAPTER 4 Results

108

has the highest purity of all three materials in question, it nevertheless shows the smallest grain size. This is believed to be the effect of the lower stacking fault energy than that of aluminium (AlSFE: 0.2 J/m2 and CuSFE: 0.07 J/m²). The influence of alloying elements, second phase particles and stacking fault energy on the microstructural evolution during ARB is acknowledged, although the individual contributions are not always easy to separate. The results regarding the microstructural refinement of copper are conflicting with the results reported by Li et al [Li06b]. Oxygen free copper showed a coarser grain structure of approximately 350 nm after 6 ARB cycles in comparison to the grain size measured in this work. The reason for this may be related to different diameter of the rolls, different force exerted on the sheets during rolling as well as different processing speed. From TEM micrographs (Fig. 4-21 k and n) and cumulative frequency figures of the grain size distribution (Fig. 4-21 l-o), it can be seen that smaller grain size was reached for pure titanium and niobium, rolled up to 6 and 3 ARB cycles, respectively. Both titanium and niobium have high melting temperatures and high initial hardness values, which makes them more difficult to process than aluminium and copper alloys. In addition, they have different crystal structures with limited number of slip systems available for deformation at low processing temperatures. Nevertheless, processing of titanium showed that a very fine and homogenous grain size can be reached after 6 ARB cycles (Fig. 4-21 k and l), which was also reported by Terada et al [Ter07]. Similar observations were found for niobium, although niobium was not able to be processed more than 3 ARB cycles due to the extremely high hardness and the lack of power of the rolling mill. The TEM micrograph (Fig. 4-21 n and o) shows a less homogenous microstructure and slightly bigger grain size than that of titanium. Figure 4-22 shows the influence of the number of ARB cycles on the average grain size, for all materials investigated. Generally, all materials show rapid grain refinement, irrespective of their homologous temperatures, and reach a grain size of less than 1 µm already after a few ARB cycles. Titanium and niobium develop an UFG microstructure after only one ARB cycle, although the microstructural homogeneity, especially for niobium rolled only up to 3 ARB cycles, still remains questionable. Commercial purity aluminium AA1050 and aluminium alloy AA6016 show somewhat slower grain refinement due to easy recovery and the highest homologous temperatures in comparison to other materials investigated. Generally, it seems that the finest microstructure and the fastest grain refinement during the ARB process can be achieved for high temperature melting point alloys with bcc or hcp crystal structure and low stacking fault energy.

CHAPTER 4 Results

109

Figure 4-22: Grain size versus the number of ARB cycles, as measured for different materials processed at various temperatures.

CHAPTER 4 Results

110

4.4.2 Grain Size Refinement of ARB processed Aluminium Sheets More detailed investigations regarding the microstructural evolution during accumulative roll bonding was carried out on the commercial purity aluminium AA1050 and aluminium alloy AA6016. Figures 4-23 a) and b) show two light micrographs of the AA6016 alloy rolled at 230 °C after 1 and 6 ARB cycles, respectively. Already after the first ARB cycle, the microstructure becomes strongly elongated. The grains close to the interlamellar boundary and the metal sheet surfaces in contact with the rolls appear smaller than the grains in the middle of the metal sheet. This occurs as a result of non-uniform shear strain distribution across the thickness of the sheets, which leads to an inhomogeneous microstructure. The last interlamellar boundary is usually the weakest link in the ARB material, since it experiences only one ARB cycle. Thus, it can usually be seen in LM or SEM micrographs before as well as after etching. After a few ARB cycles, the ultrafine grained microstructure develops. The grain size, which now ranges between 100 nm to 300 nm, is not visible using a light microscope and further microstructural investigations usually have to be conducted using transmission electron microscopy. Figure 4-23 b) reveals some evidence of the previous interlamellar boundaries as indicated by the red arrow, but most of the boundaries are well integrated into the material. a) b)

first/last interlamellar boundary

Figure 4-23: Light micrographs of AA6016 processed at 230 °C after a) 1 ARB and b) 6 ARB cycles. The gradual microstructural development after two, four, six and eight ARB cycles is shown in figure 4-24 for aluminium alloy AA6016. From figure 4-22 and the micrographs and diffraction patterns all taken using the same aperture, it can be observed that the grain size significantly decreases from N0 to N6 (Fig. 4-24 a, c and e) and that the diffraction patterns develop closed rings which become finer with an increase in number of ARB cycles. However, it seems that further roll bonding (after e.g. 8 ARB cycles) does not contribute to a significant grain size reduction. On the contrary, it appears that the grain size increases slightly. This could be the effect of the Mg2Si second phase particles, which grow and age during the ARB process and finally loose some of the grain boundary pinning potential due to numerous pre-heating in the furnace and roll bonding.

CHAPTER 4 Results

111

a) b)

c) d)

Figure 4-24: TEM micrographs after a) two, b) four, c) six and d) eight ARB cycles, and the corresponding diffraction patterns of aluminium alloy AA6016. In order to understand the influence of the Mg2Si particles on the microstructure and mechanical properties of the ARB processed AA6016 sheets, it is important to mention the precipitation sequence which occurs in the conventional material after solutionising. From the literature [ Per99, Sch04] it is well known that the precipitation sequence, which occurs in the solution treated material (typically of the 6xxx series) during age hardening precedes as follows: formation of clusters ⇒ GP zones ⇒ β’’ and β’ phases and finally the formation of stable Mg2Si second phase particles. In this work, the precipitational evolution was not investigated in detail, although some transmission electron microscopy analyses was conducted in order to identify the particles and show similarities with literature. Figure 4-25 a) shows the precipitational distribution next to a triple point grain boundary in the as-received aluminium alloy AA6016-T4. The red files indicate the more advanced stage of the precipitational evolution, presumably the β’ phase or the stable Mg2Si particles. The blue files indicate what seem to be the etched away needle shaped precipitates or the β’’ phase, usually found in the AA6016 alloy after short aging (10 min.) at approximately 225 °C [Sch04]. During aging, the transition from semicoherent β’’ phase to incoherent β’ phase results in a maximum hardness peak. Thereafter, the precipitates grow and become incoherent, thereby loosing some of the grain boundary pinning effect.

CHAPTER 4 Results

112

Figure 4-25 b shows a close up view of the final precipitational stage of the rod or disk-like Mg2Si particles, being approximately 200 - 500 nm in size. Aluminium alloy AA6016 is a naturally age hardenable alloy. Even though the material is usually supplied in the solutionised condition the Mg2Si particles form due to a room temperature precipitation process. Since the time between the metal sheet manufacturing and supply may vary, the exact state of the precipitates within the alloy is usually unknown and difficult to estimate. Thus, before conducting the ARB process, the AA6016 sheets are solutionised in order to achieve a defined material state. The ARB process is performed at elevated temperatures and the precipitates tend to develop from the solution and grow more rapidly than at room temperature. After 8 ARB cycles, the material is subjected to 8 × 3.5 min. (a total of 28 min.) of heating in the furnace at 230 °C, which may be sufficient for the precipitates to grow and/or overage. It is therefore possible that the strengthening effect is slightly lost after 8 to 10 ARB cycles. Unfortunately, this effect is not easily seen from the TEM micrographs, because the Mg2Si particles are not clearly visible in the deformed and dislocation rich microstructure. a) b)

Figure 4-25: TEM micrographs showing a) an overview of precipitational distribution next to a triple point grain boundary and b) a close up view of different precipitation morphology in the as-received T4 state of the aluminium alloy AA6016. At this stage the influence of oxide particles must also be mentioned. During ARB, the oxide layer is removed mechanically using a rough wire brush, leaving behind only a very thin oxide layer. The rest of the oxide layer which remains on the metal sheet surface can break during the ARB process and potentially contribute to dispersion strengthening, as already suggested by others (see literature review). The oxygen as well as the nitrogen content was also measured in this study. The investigation was carried out by Professor Martin Heilmaier at the Otto-von-Guericke University in Magdeburg using spectroscopy analysis. Each sample was analysed twice. The results show that the oxygen and nitrogen contents are very low for both as-received and ARB processed samples (Table 4-2). Therefore, it can be assumed that the oxide particles do not contribute significantly to dispersion strengthening of the material.

CHAPTER 4 Results

113

Table 4-2: Oxygen and nitrogen content in the as-received and ARB processed AA6016. Sample Oxygen Nitrogen

AA6016-N0 8.0 ppm 9.8 ppm 0.3 ppm 1.7 ppm AA6016-N4 4.6 ppm 15.5 ppm 4.0 ppm 1.5 ppm

For the purpose of investigating the influence of high number of ARB cycles on the microstructure, the AA1050 sheets were accumulative roll bonded up to 20 ARB cycles. Surprisingly, it can be seen that there is almost no evidence that the material was rolled i.e. the grain structure is not elongated, but still ultrafine (Fig. 4-26 a, b and d). The grains are equiaxed and homogeneous, but some microstructural inhomogeneity was found in sample shown in figure 4-26 a. The area number 1 shows very fine grains, while the area number 2 reveals much larger grains, which was also confirmed from the diffraction patterns in figures 4-26 c) and e). It is possible that the finer grains developed as a result of a more effective shear process, which usually takes place at interlamellar boundaries, while the grains in other areas remain larger. Theoretically, the number of interlamellar boundaries and the interlamellar spacing after 20 ARB cycles would reach approximately one million boundaries and one nanometre spacing, respectively. Practically, this is not possible since the microstructure dynamically recovers during the process and the interlamellar boundaries become integrated in the microstructure. Generally, the ultrafine-grained microstructure is retained, but the typical elongated and laminar microstructure is lost after a high number of ARB cycles. In addition, the equiaxed grains indicate that the microstructural anisotropy can be reduced by performing the ARB process more than 8 times. Until now, no similar microstructural observations were reported in the literature. a) b) c)

1

2

Figure 4-26: TEM micrographs showing a) interface between areas of smaller (1) and bigger grains (2), b) close up view of larger grains with the corresponding diffraction pattern and c) close up view of finer grains with the corresponding diffraction pattern of commercial purity aluminium AA1050 rolled at room temperature up to 20 ARB cycles.

CHAPTER 4 Results

114

4.4.3 Fracture and fracture surfaces of ARB samples after Tensile Testing The fracture behaviour of the accumulative roll bonded samples has not been widely investigated. However, simple SEM investigations of fracture surfaces can possibly reveal important material and fracture characteristics, such as the crack initiation and propagation. Figure 4-27 shows the accumulative roll bonded samples after tensile testing. As it can be seen in figures 4-27 a) and c), the specimens fracture typically at 45 ° to the tensile direction. The samples show very little or no necking. This implies that the sample deformation can not be stabilised or prolonged by a localised deformation, but occurs rather rapidly. However, it is also possible that the limited amount of necking is also affected by the sample geometry. The deformation of flat samples may be limited by the thickness of the sheet, because of the geometrical restriction i.e. 1 mm thickness. In contrast, the deformation of samples with circular cross sections may proceed more homogeneously throughout the thickness. a)

b)

c)

d)

Figure 4-27: Accumulative roll bonded aluminium alloy AA6016 showing a) front (NP: normal plane) and b) side view (RP: rolling plane) of the sample of 2 ARB cycles and c) front (NP: normal plane) and d) side view (RP: rolling plane) of the sample of 4 ARB cycles after a tensile test.

CHAPTER 4 Results

115

From figures 4-27 b) and d), it appears that the ARB samples usually fracture at the last interlamellar boundary. Since the last interlamellar boundary or the last bonded layer is rolled only once, it is also the weakest link in the chain. Once the crack initiates, it propagates along the interface, which provides the least resistance and costs the least amount of energy. The ARB samples usually undergo some delamination along the length of the last bonded layer, which is similar to the so-called cleavage fracture. This effect can be clearly seen in figures 4-27 b) and d). a) b)

c) d)

e) f)

Figure 4-28: Fracture surfaces after tensile testing of accumulative roll bonded aluminium alloy AA6016 after a) two, c) four and e) six ARB cycles. The pictures b), d) and f) show the close up areas (red squares) of the corresponding ARB cycles. All samples were investigated with an SEM. Systematic investigations of the ARB sample fracture surfaces were conducted using a scanning electron microscope, and can be seen in figure 4-28 for two, four and six

CHAPTER 4 Results

116

ARB cycles. The fracture surface of the tensile sample after 2 ARB cycles reveals the last and the second last interface. This indicates that the initial pores probably build up at the interfaces, they grow and coagulate with each other and finally lead to a crack initiation. A close up view even shows some evidence of slight thickness reduction of the individual layers (Fig. 4-28 a). This implies that the sample deformation is not homogeneous throughout the thickness of the sheet and that necking takes place within each individual layer rather than the whole sample. As a consequence, the elongation to failure or even the strength would be lower than that of the bulk material. The material behaves as a laminate and the externally applied force during tensile testing acts over a number of thinner layers (with a smaller cross sectional area) rather than a whole sample. Further ARB cycles (Fig. 4-28 c and e) also show previous interfaces, but no necking of the individual layers, as the case was for AA6016-N2 sample. Generally, the fracture surfaces show many voids and dimples (Fig. 4-28 b, d and f), which are characteristic features of a ductile fracture.

CHAPTER 4 Results

117

4.5 Microstructural Stability during Annealing of CG and UFG Materials

Thermal properties of conventionally grained (CG) and ultrafine-grained (UFG) materials were investigated with regard to the annealing response in terms of temperature and its influence on the mechanical properties, differential scanning calorimeter (DSC) measurements and thermal stability. 4.5.1 Annealing Response of CG Aluminium Alloy AA6016 Conventionally grained aluminium alloy AA6016 in the T4 state was of particular interest due to the relatively rapid precipitation of Mg2Si particles at room or elevated temperatures. The development of Mg2Si preciptates directly affects the mechanical properties. Figure 4-29 shows the variation of hardness of different initial states of AA6016 with respect to annealing time. Three initial material states were compared: as-received T4 state, solutionised state, and solutionised and aged state (20h/180°C). It can clearly be seen that the hardness of the as-received T4 state and solutionised state differ by approximately 30 HV10 without any artificial aging i.e. at room temperature (arrows 1 and 2). During artificial aging at 180°C, the hardness of these two states (as-received and solutionised) increases due to preciptation hardening. However, there is still a slight difference in hardness of the as-received and solutionised state during annealing. It appears that a higher hardness level can be achieved by artificial aging of the initially solutionised state. Similar results were also reported by Birol [Bir05] who investigated the pre-aging behaviour of AA6016 in order to improve the bake hardening response.

Figure 4-29: Hardness of aluminium alloy AA6016 after annealing at 180°C in three different conditions: as-received T4 state, solutionised and solutionised and artificially aged for 20 hours at 180°C.

CHAPTER 4 Results

118

In order to estimate the time necessary for the precipitation completion and overaging, the three different material states were aged up to 85 hours. The initial hardness of the sample artificially aged for 20 hours at 180°C is 100 HV10 (arrow 3). After further annealing the hardness continues to increase and reaches a maximum after additional 10 hours. After altogether 30 hours of annealing, the hardness reaches its maximum and decreases due to overaging of the Mg2Si precipitates, which increase in size and change their coherency. The precipitates can not effectively pin the dislocations or act as barriers to the dislocation motion, leading to an overall decrease in hardness. A hardness decrease after 30 hours can be observed for all three material states. 4.5.2 Annealing Response of ARB processed Aluminium Alloy AA6016 In order to determine the thermal stability of the ARB processed materials, all samples were furnace annealed at 180°C up to 65 hours (Fig. 4-30). During the ARB process, the AA6016-N2 and AA6016-N4 samples experienced a total annealing time of 10.5 min. and 17.5 min., respectively. After annealing for approximately 2-3 hours, they show a slight increase in hardness. This can be contributed to the second phase particles, which may still grow and change their morphology. The samples rolled up to 6 ARB cycles show no significant difference in hardness during 2-3 hours of annealing at 180°C. On the other hand, the samples roll bonded up to 8 times show only softening during annealing. This indicates that most second phase particles have already precipitated out of the solution and reached the maximum hardening effect during rolling. Thus, the peaks in the hardness vs. annealing time curves shift to the left i.e. towards shorter annealing times with increasing number of ARB cycles. It seems that there is a transition stage after which no further material strengthening effect can be observed during annealing. The transition stage lies presumably somewhere between the fourth and the sixth ARB cycle. All ARB processed samples soften rapidly after 10-20 min. of annealing at 180°C and reach a plateau between 45 and 65 hours of annealing. At this stage the ARB samples have most likely recrystallised. From figures 4-29 and 4-30, it can be observed that the ARB processed samples precipitate faster compared to the solutionised CG sample. The solutionised CG material reaches saturation in hardness after approximately 30 hours, while the ARB processed samples require only ~60 min. (e.g. AA6016-N4) for the same annealing temperature.

CHAPTER 4 Results

119

Figure 4-30: Annealing response of ARB processed aluminium alloy AA6016. 4.5.3 DSC Measurements of CG and UFG Aluminium Alloy AA6016 The DSC measurements were performed on the CG and UFG AA6016 samples in order to investigate the precipitation evolution and annealing response during a continuous temperature increase. Figure 4-31 a) shows three different initial conditions of aluminium alloy AA6016, namely: the solutionised, solutionised and 1 day aged, and solutionised and 7 days aged conditions. The three DSC curves show the same trend, but slightly different precipitation evolution. The solutionised state shows two dominant peaks at approximately 260 °C and 300 °C. Similar observations were also reported in literature for the same alloy and suggest the formation of the β″-phase and β′-phase, respectively [Bir05]. One day and seven days room temperature aging (natural aging) results in the formation of β″-phase and β′-phase at higher temperatures. Small Mg, Si and Mg-Si cluster form during early stages of heating at approximately 100 °C and influence the formation of the β″-phase and the β′-phase. It seems that it is necessary for the dissolution of clusters to occur prior to the formation of the two phases, because they act as nuclei for further formation of the β″-phase and β′-phase, and finally the stable Mg2Si precipitates [Edw98, Bir05]. It is therefore required that the clusters in the naturally aged samples are firstly dissolved in the matrix, before the formation of β″ and β′ phases can begin. The precipitation sequence of the naturally aged samples is retarded and the DSC curves are therefore shifted to higher temperatures (Fig. 4-31 a). The early clustering at room temperature is generally detrimental for the mechanical properties and usually impractical. During the paint bake cycle used in the automotive industry for the autobody panels, annealing must be performed at higher temperatures or longer annealing times in order to achieve a maximum strength. This makes the paint bake cycle of the naturally aged AA6016 sheets less economical.

CHAPTER 4 Results

120

a)

b)

Figure 4-31: DSC curves of aluminium alloy AA6016 of a) CG and b) UFG samples processed by accumulative roll bonding. All samples were heated to a maximum temperature of 400°C at a rate of 10 K/min. Figure 4-31 b) shows the DSC curves of the ARB processed AA6016 samples. The high amount of stored energy in terms of plastic deformation in the rolled samples results in a high driving force for recovery and recrystallisation. Therefore, there are all together three possible exothermal processes which may occur during heating of UFG samples. The samples may precipitate, recover during early stages of heating and later recrystallise. All three processes are superimposed to each other and it is not clear whether the exothermic peaks are due to precipitation or recovery. Samples AA6016-N2 and AA6016-N4 show small exothermic peaks at 320 °C and 340 °C, respectively.

CHAPTER 4 Results

121

Therefore, it is possible that there is still some precipitation activity such as precipitation growth and changes in the precipitation morphology which can result in small peaks detected in the DSC curves (Fig. 4-31 b) as well as in an increase in hardness after annealing (Fig. 4-30 b). The AA6016-N6 sample does not show any distinct peaks in the DSC curves, although it can be argued that a slight exothermal reaction is present at approximately 310°C. On the other hand, the AA6016-N8 sample does not show any exothermal peaks. These results correspond well with the results observed in figure 4-30, where the AA6016-N8 sample continually softens during post-roll annealing i.e. there was no further increase in hardness. Both results obtained from the DSC curves as well as the hardness values after annealing of the AA6016-N8 sample indicate that the precipitation evolution and growth may have already completed during the ARB process. 4.5.4 Thermal Stability of ARB processed Materials It is expected that the severely deformed materials have low thermal stability. In order to estimate the temperatures which can be used for the subsequent metal forming processes of UFG materials, it is important to determine the appropriate process temperatures where the UFG microstructure is still stable.

Figure 4-32: Thermal stability curves of ARB processed AA6016 aluminium alloy roll bonded to 2, 4, 6, 8 and 10 ARB cycles. All samples were annealed for one hour at 100 °C, 200 °C, 300 °C and 400 °C. Figure 4-32 shows thermal stability curves of the ARB processed AA6016. The hardness values can be compared to the as-received, T4 state and the solutionised state. It can be seen that hardness increases from 40 HV10 in the solutionised state to 95 HV10 after approximately 6 ARB cycles. The ARB samples start to loose their thermal stability once they are exposed to elevated temperatures. The ARB samples are stable to 200 °C and start to soften rapidly between 200 - 300 °C. After one hour annealing at 300 °C, the ARB samples reach approximately the same hardness as the

CHAPTER 4 Results

122

solutionised state. This indicates that the UFG microstructure is completely lost and the material is completely recrystallised. Similar observations were also made by Park et al. [Par01] and Slamova et al [Sla07]. It can be assumed that thermal stability of the ultrafine-grained AA6016 is comparable to that of the UFG commercial purity aluminium AA1050 [May04, Bar01]. a)

b)

Figure 4-33: Thermal stability curves of ARB processed a) oxygen free copper and b) titanum grade 2 roll bonded to 2, 4, 6, 8 and 10 ARB cycles. All samples were annealed for one hour from 100 °C to 600 °C. Thermal stability curves of oxygen free copper (Cu 99.99 %) and pure titanium grade 2 (Ti 99.2 %) are shown in figure 4-33. It is important to mention that the ARB

CHAPTER 4 Results

123

process of copper was conducted at room temperature and that of titanium at 250 °C. Initial hardness of the as-received copper averaged 50 HV5. During the ARB process, the hardness increased by approximately 2.6 times and reached approximately 130 HV5 after 4 ARB cycles. During annealing the material softens rapidly between 100 - 200 °C and reaches the same hardness of the as-received state after annealing at 300 °C for 1 hour. The UFG copper samples seem to be completely recrystallised at 300 °C (Fig. 4-33 a). These results are comparable to those of aluminium alloy AA6016. Moldova et al. [Mol07] who investigated ECAP processed copper of slightly lower purity (99.95% Cu) found a shift of thermal stability to slightly higher temperatures. It is therefore believed that material purity as well as second phase particles have a significant influence on thermal stability of UFG materials. Pure titanium grade 2 has a higher initial hardness and melting point in comparison to aluminium and copper. In addition, titanium has a hexagonal crystal structure, which makes the ARB process at low temperatures much more difficult due to the limited number of slip systems. Therefore, the ARB process of titanium was performed at 250 °C. During ARB the hardness increases from 125 HV5 to 275 HV5 after 4 ARB cycles i.e. a factor of 2.2. The UFG microstructure stays stable up to 400 °C, after which the material starts to soften. The decrease in hardness takes place gradually between 400-600 °C. Therefore, it can be assumed that the material completely recrystallises at 700 °C (Fig. 4-33 b).

CHAPTER 4 Results

124

4.6 Formability of CG and UFG Aluminium Sheets 4.6.1 Hydraulic Bulge Tests Hydraulic bulge tests were used in order to evaluate the sheet forming capability of the ARB processed aluminium alloys. It is an important testing method from a process design point of view since the loads acting on the metal sheet are similar to many industrial processes such as bending, deep drawing or stretching. Until now there is very little information available on the deformation behaviour or formability of UFG sheets under multiaxial stress state conditions [Lap07a]. Another advantage of the bulge test is that localised necking does not take place (or is postponed) under biaxial stress state conditions and that the samples generally show higher uniform strain than under the uniaxial stress state conditions [Han08]. Therefore, higher strains can be expected from the bulge tests than from the tensile tests for the as-received as well as the ARB samples. Commercial purity aluminium AA1050 and aluminium alloy AA6016, in the as-received states and after 2, 4, 6 and 8 ARB cycles, have been investigated using a hydraulic bulge test. All tests were performed at room temperature, using oil as a forming medium. The initial bulge tests were carried out on 0.8 mm thick samples. Figure 4-34 shows the influence of the number of rolling cycles on the achievable burst pressure and pole height (out of plane sheet deformation). Remarkably enough, roll bonding of commercial purity aluminium AA1050 shows higher burst pressure and at the same time higher deformation capability in terms of pole height compared to the as-received cold rolled state (Fig. 4-34 a). Burst pressure and pole height increased by approximately 30 %, from 40 bar to 55 bar and from approximately 12.5 mm to 17.5 mm, respectively. After the first few cycles of roll bonding, burst pressure remains constant, while pole height drops at 8 ARB cycles. Due to high scatter of values, it can not be clarified whether this effect arises from surface defects and microstructural instability, or simply requires a higher number of test values. The bulge test results of aluminium alloy AA6016 were evaluated up to 4 ARB cycles and generally show high statistical scattering (Fig. 4-34 b). Samples of higher ARB cycles generally failed at the rim of the die and the registered burst pressure acting at the pole of the blank was therefore inaccurate. The achievable burst pressure drops after 2 ARB cycles, but increases upon further rolling. Until now it could not be confirmed whether higher burst pressure or pole height can be reached at higher rolling cycles due to failure of the blanks at the rim of the die, which possibly occurred as a result of higher bending stiffness or notch sensitivity of the UFG sheets and/or unfavourable geometry of the die such as the small transition radius. Low ductility and premature fracture can generally be caused by a number of factors including high internal stresses as well as processing parameters which lead to different inhomogeneities such as metal sheet waviness or thickness fluctuations. Nevertheless, the achievable pole height of the roll bonded samples can still be compared to the as-received state and a tendency towards higher burst pressures increases with the development of UFG microstructure.

CHAPTER 4 Results

125

a)

b)

Figure 4-34: Burst pressure and pole height (out of plane sheet deformation) vs. the number of ARB cycles for a) commercial purity aluminium AA1050 and for b) aluminium alloy AA6016. The sheet thickness averaged approximately 0.8 mm. Note that standard deviation was omitted for AA6016 of 4 ARB cycles, because only one sample was successfully tested and failed at the pole of the blank. Standard deviation was plotted for all other data points, although some may not be individually visible because they are too small.

CHAPTER 4 Results

126

Similar results were observed from the von Mises equivalent strain vs. the number of ARB cycles for both materials. Von Mises equivalent strain increases from 0.40 in the as-received state to 0.80 at N4 of the accumulative roll bonded commercial purity aluminium AA1050, as it can be seen in figure 4-35 a. Aluminium alloy AA6016 shows a slight decrease of the von Mises equivalent strain from the as-received state to four ARB cycles, but the values do not differ significantly (Fig. 4-35 b). a)

b)

Figure 4-35: Burst pressure and von Mises equivalent strain distribution vs. the number of ARB cycles of a) commercial purity aluminium AA1050 and b) aluminium alloy AA6016. The sheet thickness averaged approximately 0.8 mm. Note that standard deviation was omitted for AA6016 of 4 ARB cycles, because only one sample was successfully tested and failed at the pole of the blank. Standard deviation was plotted for all other data points, although some may not be individually visible because they are too small.

CHAPTER 4 Results

127

Figures 4-36 a)-c) show the von Mises equivalent strain distribution across AA1050 blanks in the as-received state, after 4 and 8 ARB cycles, shortly before the final failure as well as after the fracture has taken place (Fig. 4-36 d-f). Higher formability in terms of von Mises equivalent strain was achieved at higher burst pressures in comparison to the as-received state with an increase in the number of ARB cycles. This relates well with the tensile test results obtained for the AA1050, where an increase in the number of ARB cycles leads to higher achievable strength and at the same time higher strain to failure. After 8 ARB cycles, a slight decrease in the von Mises equivalent strain was observed. This is most likely due to a highly deformed, dislocation rich microstructural or surface instability. a) AA1050-N0 b) AA1050-N4 c) AA1050-N8

d) AA1050-N0 e) AA1050-N0 f) AA1050-N0

Figure 4-36: Von Mises equivalent strain distribution of aluminium AA1050 in the a) as-received state, b) after 4 and c) after 8 ARB cycles, and the respective sample fracture positions indicated by red arrows in d), e) and f). The sheet thickness averaged approximately 0.8 mm. Other than by AA1050, the increase in the number of ARB cycles of AA6016 showed a slight increase of the von Mises equivalent strain, from the as-received T4 state to 4 ARB cycles (Fig. 4-37 a and b). Both samples failed at the pole of the blank as indicated in figures 4-37 d) and e). These results correspond once again with the tensile tests where the strength increases with the number of ARB cycles, while somewhat sacrificing the ductility. Most ultrafine-grained samples, with more than 4 ARB cycles, fractured at the rim of the blank i.e. at the radius of the die (Fig. 4-37 c and f). It is believed that this behaviour is due to a supposedly higher bending stiffness and notch sensitivity of the UFG AA6016. Highly plastically strained aluminium alloy AA6016 was therefore found to have a lower deformation capacity than the AA1050, but has nevertheless shown some promising results. Furthermore, it can be assumed that the ultrafine-grained, strain rate sensitive materials would show better deformation behaviour by testing at a lower strain rate or at higher temperatures and that an

CHAPTER 4 Results

128

improved sheet thickness to die diameter ratio would eliminate the problem of sample fracture at the transition radius of the die. a) AA6016-N0, T4 state b) AA6016-N4 c) AA6016-N8

d) AA6016-N0, T4 state e) AA6016-N4 f) AA6016-N8

Figure 4-37: Von Mises equivalent strain distribution of aluminium alloy AA6016 in the a) as-received T4 state, b) after 4 and c) after 8 ARB cycles and the respective sample fracture positions indicated by red arrows in d), e) and f). The sheet thickness averaged approximately 0.8 mm. With the intention to improve the deformation behaviour of the UFG aluminium, as well as avoid premature fracture at the transition radius of the die and investigate the possible role of sample/die geometry on the deformation behaviour during bulge testing, the AA6016 ARB sheets were rolled down to 0.5 mm thickness. The ratio between the sample thickness and the die diameter were adapted for every material in order to insure that the deformation takes place through the thickness of the sheet and not at the transition radius of the die where bending stresses prevail. Figure 4-38 shows the maximum burst pressure and the maximum von Mises equivalent strain versus the number of ARB cycles for the aluminium alloy AA6016. It can clearly be seen that the maximum burst pressure increases by a factor of 3 in comparison to the as-received state, which is also reflected in the significant increase of the ultimate tensile strength observed during tensile testing. The maximum burst pressure reaches a saturation level after 4 ARB cycles. However, the maximum von Mises equivalent strain obtained from the bulge tests decreases from approximately 20 % in the as-received state to 10 % after 8 ARB cycles. These values are qualitatively comparable to those obtained from the tensile tests. However, it is important to note that the two methods of testing can not be

CHAPTER 4 Results

129

quantitatively compared with one another due to different stress states as well as the different sample thicknesses used in the experiments.

Figure 4-38: Burst pressure and von Mises equivalent strain distribution vs. the number of ARB cycles of aluminium alloy AA6016. The aluminium sheet thickness averaged approximately 0.5 mm. Standard deviation was plotted for all other data points, although some may not be individually visible because they are too small. The formability of the ARB sheets can be seen from figures 4-39 a) and b), which show the maximum pole height reached during the bulge test. The as-received and the UFG sample after 4 ARB cycles show comparable pole height and a good forming potential. Similar results were observed for the AA1050 with regard to the maximum burst pressure. The burst pressure increased by 30 % and reached a saturation level after 2-4 ARB cycles. On the other hand, the maximum von Mises equivalent strain increased by a factor of 2 after 4 ARB cycles, where it averaged approximately 75 % (increase of pole height from 12. mm to 17.5 mm, from the as-received to four ARB cycles, respectively). In addition, it is possible to estimate the formability potential of metal sheets for e.g. a deep drawing process from the coefficient of normal anisotropy <r> and the anisotropy variation Δr. Higher <r> values generally reduce the tendency towards sheet thinning and smaller Δr values reduce earing (see chapter 2.2.2 for details). Skrotzki et al. measured the anisotropy values of the ARB AA6016 and concluded that a good compromise between strength, ductility, small reductions in sheet thickness during deformation and reduced earing, can be reached from samples rolled up to 4 ARB cycles [Skr07]. The results obtained from bulge tests support the conclusions made by Skrotzki et al.

CHAPTER 4 Results

130

The influence of the rolling direction of the ARB sheets on the direction of crack propagation was also investigated. No clear conclusion could be made regarding the influence of rolling direction, since the failure of most samples occurred arbitrarily even though the microstructure was strongly elongated and anisotropic. The difference in anisotropy could not be detected during bulge testing mostly due to the biaxial stress state conditions. a) AA6016-T4 b) AA6016-N4

Figure 4-39: Maximal pole height registered for aluminium alloys AA6016 of a) as-received T4 state and b) after 4 ARB cycles. The sheet thickness was approximately 0.5 mm. At this stage it is important to mention that the pole height and the von Mises equivalent strain distribution figures (Fig. 4-36, Fig. 4-37 and Fig. 4-39) can only be used for a qualitative comparison of the results, since the figures represent an overall distribution of pole height or von Mises strain values and the exact values at a maximum burst pressure can not be estimated directly from the figures. It is also worth pointing out that a direct comparison between bulge and tensile tests is difficult due to different stress states. Tensile test samples experience a uniaxial stress state and the measured strains corresponding to the change in length over the original gauge length, while the bulge test samples experience a biaxial stress state and the measured strain is the von Mises equivalent strain corresponding to the natural logarithm of the minimum thickness over the original thickness [Gut04]. The von Mises values were mostly considered qualitatively and used for comparison purposes rather than for quantitative analysis, while the tensile tests can be used as a relatively good indicator of the material behaviour under a uniaxial stress state. Nevertheless, the results show that there is a good agreement between the tensile and the bulge tests. Both methods have verified that the ultrafine-grained aluminium sheets show great potential and promising deformation behaviour.

CHAPTER 4 Results

131

5.6.2 Cup Drawing Tests Cup drawing tests were conducted on the as-received and accumulative roll bonded sheets in order to investigate the sheet metal formability under tension-compression stress state conditions. The tests were conducted at room temperature and at 180°C. The focus of the investigation was placed on commercial purity aluminium AA1050 and aluminium alloy AA6016. In order to evaluate deep drawability of aluminium sheets, the limiting drawing ratio (LDR, also known as βmax.) was determined using the Schmidt-Kapfenberg plot from a series of blanks with varying diameters. The limiting drawing ratio is a measure of drawability and defines the ratio of the largest blank to punch diameter, which can be drawn in a single step without fracturing. It is reached when the maximum forming force (or drawing force) exceeds the fracture force of the material. The sheet or cup fracture is manifested in the well known bottom fracture. Deep drawing of CG and UFG Al sheets at room temperature Blank diameters varied between Ø 80 mm, Ø 85 mm, Ø 90 mm and Ø 105 mm for room temperature experiments. All commercial purity aluminium blanks with Ø 80 mm and Ø 90 mm were successfully deep drawn at room temperature, while all Ø 105 mm blanks fractured. Figure 4-40 shows some examples of the successfully drawn and fractured cups. Deep drawn cups in the CG state (N0) show no significant difference compared to deep drawn cups in the UFG state (N6). The largest sample size, which could be drawn from the CG and UFG aluminium sheets without fracture, reached Ø 90 mm. Ultrafine-grained samples generally showed no earing. There was also no delamination of the individual bonded layers during room temperature drawing. The characteristic drawing force vs. displacement curves can be seen in figure 8-1 a (Appendix). Higher drawing force is required for the UFG samples (N6) than for the CG samples (N0), for the same blank diameter. This is also reflected in the Schmidt-Kapfenberg plot. The experimentally obtained fracture forces of N0 and N6 are shown in figure 4-41 as the horizontal straight lines. The fracture force of CG sample and UFG sample average approximately 14000 N and 17000 N, respectively. The ratio between the two fracture forces is a factor of 1.2, while the ratio between the ultimate tensile strength of both samples averages 1.5. The difference between the two ratios is believed to be due to different stress states. The straight line plots through the open symbols were constructed for the successfully deep drawn CG and UFG samples. The limiting drawing ratio (βmax.) for both samples was obtained from the line intersections of the successfully drawn and fractured cups. For room temperature experiments, the limiting drawing ratio of N0 and N6 lie at approximately 2.27 and 2.25, respectively. It can therefore be concluded that the AA1050 UFG samples show higher fracture forces than those obtained for the CG samples and at the same time high and equal drawability.

CHAPTER 4 Results

132

a) AA1050 N0, Ø 80 mm b) AA1050 N0, Ø 90 mm c) AA1050 N0, Ø 105 mm

d) AA1058 N8, Ø 80 mm e) AA1050 N8, Ø 90 mm f) AA1050 N8, Ø 105 mm

Figure 4-40: Deep drawn cups of commercial purity aluminium AA1050 of as-received and UFG state after 6 ARB cycles, tested at room temperature. The blanks were 0.8 mm thick. The maximum blank diameters of AA1050 and AA6016 CG samples, which could be drawn, were similar. However, the maximum blank diameters of AA1050 and AA6016 UFG samples varied and were strongly dependent on the number of ARB cycles. In case of AA6016, all CG samples with Ø 80 mm, Ø 90 mm and Ø 110 mm diameters were successfully drawn. The UFG samples (N4) were successfully drawn using Ø 80 mm and Ø 85 mm blank diameters, while the Ø 90 mm blank diameter fractured. All other UFG samples (N6 and N8) could only be drawn using the Ø 80 mm diameter blank without fracture. The characteristic drawing force vs. displacement curves can be seen in figure 8-1 b (Appendix). For the Schmidt-Kapferberg plot, the theoretical fracture forces for N0 and N4 were calculated, because it was not possible to obtain them experimentally. In case of N0, all samples were successfully drawn i.e. no fracture occurred. In order to obtain the fracture force, a sample of larger blank diameter (> Ø 110 mm) is required. Until now, larger blank size was not possible to obtain due to limited width of the rolling mill. In case of N4, the largest blank diameter (Ø 90 mm) fractured prematurely due to some delimitation of the last bonded layer. Extremely low fracture force of N4-Ø90 mm can clearly be seen in figure 8-1 b (Appendix).

CHAPTER 4 Results

133

Figure 4-41: Schmidt-Kapfenberg plot for commercial purity aluminium AA1050 rolled to N0 and N6 ARB cycles and deep drawn at room temperature.

Figure 4-42: Schmidt-Kapfenberg plot for aluminium alloy AA6016 rolled to N0 and N4 ARB cycles and deep drawn at room temperature. Theoretical fracture forces were calculated from equation 2-31. They are a function of samples thickness, punch diameter and the ultimate tensile strength of the material. The theoretically calculated fracture forces of N0 and N4 average 35000 N and 29000 N, respectively (Fig. 4-42). From the line intercepts, the theoretical limiting drawing ratio was calculated for both samples. The UFG aluminium AA6016 sample

CHAPTER 4 Results

134

after 4 ARB cycles shows significantly lower theoretical limiting drawing ratio of 1.74 compared to the CG sample, which reaches a theoretical limiting drawing ratio of 2.21. The difference of the theoretical limiting drawing ratio between CG and UFG samples may be overestimated, because of the theoretically calculated fracture forces. The theoretically calculated fracture forces are generally higher than the experimental values. Thus, a shift of the measured fracture force to lower values would also result in a smaller difference between the drawing ratios of the samples. Nevertheless, the UFG AA6016 generally shows a deteriorated drawability for room temperature conditions compared to its CG counterparts. However, it must be mentioned that the deformation rate during deep drawing significantly exceeds that of the uniaxial tensile tests, which naturally leads to lower formability of the strain rate sensitive materials. Successfully drawn and fractured cups are documented in the appendix, see figure 8-2. Generally, the CG and UFG samples show no earing during room temperature drawing. Some UFG samples have occasionally delaminated along the last bonded layer. Deep drawing of CG and UFG Al sheets at elevated temperature In order to improve the drawability of UFG aluminium sheets, cup drawing tests were also conducted at elevated temperatures. All aluminium blanks of AA1050 as well as AA6016 were successfully drawn at 180°C. Blank diameters varied between Ø80 mm, Ø90 mm, Ø105 mm and Ø110 mm. Characteristic force vs. displacement curves of AA1050 and AA6016 drawn at 180°C can be seen in figures 8-3 (Appendix). The deep drawing force required for the UFG samples is significantly higher than for the CG samples. This indicates that the temperature is high enough to obtain improved drawability of UFG samples without sacrificing the strength. Figure 8-4 and figure 8-5 (Appendix) show different cup diameters of CG and UFG samples. Larger cup diameters (Ø 110 mm) of CG and UFG samples show some earing. This material behaviour is believed to be linked with the sample anisotropy. Moreover, it is believed that earing takes place at 45° to the original rolling direction, which is in accordance with the tensile tests results as well as the calculated anisotropy. Tensile tests performed on samples oriented at 45° to the rolling direction showed higher elongation to failure than the samples oriented in the rolling direction and in the transverse direction (Fig. 4-16 and 4-17). Additionally, Skrotzki et al. reported higher anisotropy values in terms of r values for samples oriented at 45° to the rolling direction [Skr07]. Although Skrozki et al. measured and calculated the anisotropy parameters for room temperature deformation behaviour, the indication regarding increasing earing tendency with increasing number of ARB cycles seems to correspond quite closely to real deep drawing experiments. In conclusions, all samples were successfully drawn at 180°C and the fracture did not occur even in the samples with the largest diameter of 110 mm. It was therefore not possible to determine the limiting drawing ratio (βmax.), because the Schmidt-Kapfenberg construction requires that at least two samples are successfully drawn and that one sample fractures (see Fig. 2-15). In order to determine the maximum fracture force and therefore the limiting drawing ratio, larger sample diameters (> 110 mm) are required. However, at this stage it is not possible to produce larger ARB sheets or blank diameters due to the limited capacity of the rolling mill.

CHAPTER 4 Results

135

Comparison of room temperature and elevated temperature results As indicated in the previous paragraph, all samples were successfully drawn at 180°C. Therefore, it was not possible to determine the limiting drawing ratio (LDR or βmax.). In order to compare the test data obtained from the experiments conducted at 180°C, the following definition was introduced, namely the minimum drawing ratio (βmin.). In this case, the minimum drawing ratio (βmin.) was not determined from the Schmidt-Kapfenberg plot. It simply represents the ratio between the maximum sample diameter that was drawn in this study and the punch diameter. It is shown in figures 4-43 a) and b) for aluminium alloys AA1050 and AA6016 samples, respectively. Figure 4-43 a) shows the ratio of sample diameter to punch diameter vs. the number of ARB cycles N for commercial purity aluminium AA1050. All AA1050 samples were successfully drawn at room temperature using a Ø90 mm blank diameter (full squares). Thus, the minimum drawing ratio averages 1.8. In addition, all samples that fractured at room temperature have also been included in the diagram for comparison purposes (open squares). All AA1050 samples having a blank diameter of 105 mm fractured. This corresponds to β = 2.1. At this stage it must also be mentioned that only sample diameters of Ø 80 mm, Ø 85 mm, Ø 90 mm and Ø 105 mm were used for testing. Thus, it is not possible to say if sample diameters ranging between 90 mm and 105 mm could be drawn or if they would fracture. Therefore, the process frame for room temperature testing of AA1050 lies somewhere between β = 1.8 and β = 2.1. On the other hand, deep drawing of AA1050 at an elevated temperature shows a considerable improvement of drawability in comparison to room temperature experiments (full triangles). Conventionally grained (N0) and UFG samples (N6 and N8) were successfully drawn using the largest possible blank diameter of Ø110 mm. The minimum drawing ratio increases from 1.8 for room temperature testing to approximately 2.2 when testing at 180°C. The drawability of N2 is slightly smaller and it is most probably due to the highly deformed and not fully developed ultrafine-grained microstructure. The minimum drawing ratio of N4 lies on the same level as the minimum drawing ratio of samples tested at room temperature. This sample seems to be an outlier, because there were no significant differences compared to other samples drawn at the same temperature. The sample shows no asymmetrical deformation during drawing and no delamination. Similar results regarding the minimum drawing ratio (βmin.) were also obtained for aluminium alloy AA6016 (Figure 4-43 b). During room temperature testing, the minimum drawing ratio decreases with increasing number of ARB cycles (full squares). This indicates that the UFG microstructure seems to play unfavourable rather than advantageous role. Compared to the CG sample, which was successfully drawn even using the largest possible blank size of Ø110 mm, the UFG sample after 6 ARB cycles was only possible to draw using a much smaller blank size of Ø80 mm. The open squares represent the samples that fractured. It can be seen that there is not much difference in the drawing ratios between the samples which were successfully drawn and the ones that fractured. Therefore, the process frame in terms of the drawing ratio for room temperature testing of UFG samples such as N4, N6 and N8 lies between 1.6

CHAPTER 4 Results

136

and 1.8. It is possible that the as-received, N0 samples as well as the N2 sample fracture only at somewhat higher drawing ratios. a)

b)

Figure 4-43: Ratio of sample diameter to punch diameter vs. the number of ARB cycles N for a) AA1050 and b) AA6016. The values are presented for room temperature and 180°C experiments. Open symbols represent the samples that fractured at room temperature. Standard deviation was not possible to determine due to the limited number of samples. At 180°C the drawability of AA6016 is improved compared to room temperature experiments. The minimum drawing ratio of the UFG samples increases from approximately 1.6 (at RT) to 2.2 (at 180°C). This corresponds to blank diameters of Ø80 mm and Ø110 mm, respectively. Generally, it can be concluded that the UFG samples show higher fracture forces than those obtained for the CG samples and at the same time comparable drawability at slightly elevated temperatures.

CHAPTER 4 Results

137

4.7 Friction Stir Welding 4.7.1 Microstructure of friction stir welded aluminium sheets Sheet metal joining plays an important role in many industrial processes and especially in the field of light weight construction in the automobile industry. In order to qualify the UFG metal sheets as a new class of materials and bring them into service, it is important to investigate the appropriate joining technique in terms of the microstructural changes and the resulting mechanical properties. Friction stir welding (FSW) was found to be a desirable joining technique for UFG materials, since it provides excellent mechanical properties and retains the fine grained microstructure. a) b)

c) d)

nuggetARB base material

RD

ND

nuggetnugget

as-received base material

HAZ TMAZ

Figure 4-44: Friction stir welded CG AA6016 sheet (T4 state) showing a) an overview of the four main areas, b) the magnified nugget area investigated with an SEM, and friction stir welded UFG AA6016 sheet (after 6 ARB cycles) showing c) a TEM micrograph of the base material and d) the nugget.

CHAPTER 4 Results

138

As-received and ARB processed aluminium AA1050 and AA6016 were successfully friction stir welded. Using the aluminium alloy AA6016 as an example material, the microstructural changes which occur during the FSW process will be highlighted and discussed. Friction stir welding results in frictional heating of the workpiece and plastic strain in the material. Four different areas can usually be differentiated after FSW: the base material, the heat affected zone (HAZ), the thermo-mechanically-affected-zone (TMAZ) and the nugget, affected by the mechanical movement of the pin and the temperature increase. Figure 4-44 shows the microstructure of the CG (as-received, T4 state) and UFG AA6016 samples before and after welding. The four different regions mentioned above can be clearly seen in figure 4-44 a. There is a clear reduction in grain size from the base material (~ 20 µm) to the nugget (1 - 5 µm), which was also confirmed by the SEM micrograph in figure 4-44 b. The grains in the thermo-mechanically-affected-zone appear elongated in the direction of the pin rotation, while the microstructure in the heat affected zone appears more homogeneous. Additionally, the influence of precipitates must be emphasised, although it was not investigated in detail during the course of this work. The complex precipitation distribution and different precipitation types were observed in different areas of the weld by Cabibbo et al. [Cab07] for a similar aluminium alloy, age hardenable AA6056 in the CG state. They confirmed precipitation coarsening and overaging in the TMAZ and some precipitation dissolution in the nugget. Accumulative roll bonding led to a significant grain refinement and produced an elongated grain structure of ~ 200 nm grain size (Fig. 4-44 c). The fine grain structure was confirmed by the diffraction pattern in figure 4-44 c), showing many concentric rings with high intensities. Friction stir welding of ARB samples resulted in bigger, equiaxed and recrystallised grains (~ 1 µm) in the nugget with a low dislocation density and Mg2Si precipitates of approximately 200 nm inside the grains (Fig. 4-44 d). The diffraction pattern in figure 4-44 d) indicates that the grain size is somewhat larger than in the accumulative roll bonded base material. In the case of accumulative roll bonded and subsequently friction stir welded AA6016 alloy, it is difficult to estimate the effect of the Mg2Si precipitates. As mentioned earlier, the precipitates develop and/or may even overage after 8 to 10 ARB cycles. After the ARB material, e.g. AA6016-N8 has been friction stir welded, the precipitates may completely loose their strengthening effect due to overaging, or they may be dissolved into the solution as suggested by Cabibbo et al [Cab07]. Some researchers also suggest that the unstable precipitates which have been dissolved in the matrix during FSW may precipitate once again and increase the material hardness. Even though this effect was not investigated, but it should be mentioned that the influence of precipitates should not be neglected. During FSW, the microstructural development and the precipitational changes in CG materials are complex, and they appear more so in the accumulative roll bonded materials.

CHAPTER 4 Results

139

The microstructural changes which occur in the ARB commercial purity aluminium AA1050 were not investigated. However, Sato et al. [Sat04, Sat06] reported pancake-shaped grains in the TMAZ and equiaxed, dynamically recrystallised grains in the nugget of the UFG AA1050 and UFG AA1100. Similar microstructural changes regarding the grain size were also observed in this work for the aluminium alloy AA6016. 4.7.2 Through Thickness Hardness Distribution of FSW sheets Figures 4-45 a) and b) show hardness profiles from the base material to the nugget of the CG and UFG samples. The friction stir welded CG AA1050 shows no significant change in hardness across the nugget. However, the hardness profile of friction stir welded CG AA6016-T4 reveals softening in the nugget, although the grain size in the nugget decreased, as it can be seen in figure 4-45 b. Thus, in case of the CG AA6016-T4, the previously mentioned precipitational changes rather than the grain size seem to play a more dominating role concerning the reduction in hardness. Coarse and/or overaged precipitates provide little to no material strengthening, leading to low hardness values in the nugget. The reduced hardness in the nugget was also reported for the age hardenable AA6061-T6, AA2024-T4 and AA7075-T6 CG alloys [Bar06, Liu97]. The suggested reasons for softening included the recrystallised microstructure in the nugget and complex changes in the precipitation morphology, distribution and density. A precipitation density drop was also measured in the nugget of AA2024 and AA7075, which confirmed the lower hardness values [Bar06]. The hardness of UFG samples of AA1050 and AA6016 decreases across the nugget with respect to the base material by up to 20 % and 50 %, respectively. This is due to the recrystallisation of the deformed UFG microstructure and the formation of bigger grains in the nugget. Similar hardness profiles of ARB processed material AA1100 were also observed by Sato et al. [Sat04, Sat06] and correspond well with the results in figure 4-45 a. The changes in the mechanical properties of AA6016 are attributed to different grain sizes and supposedly different morphology of the Mg2Si precipitates. The Mg2Si particles precipitate during preheating and rolling, and their influence after 8 ARB cycles may be reduced due to possible overaging. Thus, the recrystallised microstructure and the increase in grain size, rather than the influence of the precipitates seem to play a more dominating effect within the nugget, reducing the hardness to that of the as-received friction stir welded material. Some researchers like Cabibbo et al. [Cab07] observed complete precipitation dissolution and re-precipitation within the nugget of AA6056 aluminium alloy, resulting in a slightly higher hardness in the nugget than in the TMAZ. On the other hand, Chen et al. [Che06b] argue that the process temperature within the nugget is not sufficient enough to dissolve the stable precipitates, although it may be sufficient for dissolution of the metastable ones. This work does not include an in depth analysis of the precipitational changes and the question of complete or partial precipitation dissolution and the formation of new precipitates within the nugget remains to be answered. Nevertheless, it can be stated that the initial grain structure seems to have no significant influence on the mechanical properties of the weld.

CHAPTER 4 Results

140

a)

b)

Figure 4-45: Vickers hardness distribution across the nugget and base material for a) AA1050 and b) AA6016. Samples were processed at room temperature and at 230 °C, respectively. Note that the width of different zones was estimated from Fig. 1 a) and Fig 2. For both materials, the width of the nugget, TMAZ and HAZ is approximately 6-7 mm, 50 µm and 1-3 mm, respectively. Note that the width of different zones may differ slightly with respect to the number of cycles. In case of AA6016-T4, no differences in hardness values were observed and thus no separation of the TMAZ and HAZ was possible. At this point it must also be mentioned that the friction stir welded AA1050 and AA6016 samples were also pulled in tension. In case of aluminium AA1050, yield and

CHAPTER 4 Results

141

tensile strength of friction stir welded CG and UFG samples were almost identical and both averaged 114 MPa. The elongation to failure of UFG samples was higher compared to the CG samples and reached approximately 4 %. The positive effect on the elongation to failure may be contributed to the ultrafine-grained structure of the base material. The base material of the CG samples was in a cold rolled condition. Similarly, the friction stir welded CG and UFG AA6016 samples showed approximately identical tensile strength of 140 MPa. The elongation to failure decreased from 9.6 % in the CG state to 3.8 % in the UFG state. Here, it must be emphasised that the precipitation morphology may have some influence on the elongation to failure of both CG and UFG materials. Generally, the strength of the nugget of the CG and UFG materials remained unchanged, because of the recrystallised microstructure and similar grain size in the nugget zone. 4.7.3 Deformation behaviour of ARB and FSW aluminium alloys In order to determine the formability of welded UFG samples, bulge tests and cup drawing tests were performed on the accumulative roll bonded and friction stir welded sheets. The focus was placed on the commercial purity aluminium AA1050 and aluminium alloy AA6016. Hydraulic Bulge Tests Figure 4-46 shows the evolution of maximum burst pressure and maximum von Mises equivalent strain with increasing number of ARB cycles N of the commercial purity aluminium AA1050 and aluminium alloy AA6016. The results include the as-received and the accumulative roll bonded samples as well as the samples with and without friction stir welding. Friction stir welded commercial purity aluminium AA1050 samples after 4, 6 and 8 ARB cycles show comparable maximum burst pressure to that of the as-received state. Maximum von Mises equivalent strain of AA1050 slightly decreases after 2 ARB cycles. Further accumulative roll bonding leads to some improvement of the maximum von Mises equivalent strain and reaches roughly the same level after 8 ARB cycles as the as-received conventionally grained state, N0. Nevertheless, the von Mises equivalent strain of all friction stir welded AA1050 samples is limited to less than 10 % (Fig. 4-46 a). On the other hand, friction stir welded aluminium alloy AA6016 shows a significant increase in burst pressure with increasing number of ARB cycles and reaches a maximum value of 60 bar after 6 ARB cycles. After 8 ARB cycles, burst pressure drops slightly, but it is nevertheless significantly higher than the burst pressure of the as-received friction stir welded material. The drop in burst pressure after 8 ARB cycles may be affected by the precipitational changes in the material during the ARB process discussed previously. Maximum von Mises equivalent strain of AA6016 increases with an increase in number of ARB cycles and reaches a maximum value of approximately 10 % after 6 ARB cycles. It is important to note that the highest achievable burst pressure and the highest maximum von Mises equivalent strain were

CHAPTER 4 Results

142

obtained for samples rolled up to 6 ARB cycles (Fig. 4-46 b), showing the high potential of UFG materials. a)

b)

Figure 4-46: Maximum burst pressure and maximum von Mises equivalent strain vs. the number of ARB cycles of accumulative roll bonded and friction stir welded a) commercial purity aluminium AA1050 and b) aluminium alloy AA6016. Burst pressure was normalised with respect to sheet thickness. Generally, the FSW AA1050 aluminium sheets in the UFG state show no significant degradation of the mechanical properties in comparison to the as-received state, while the accumulative roll bonded and friction stir welded aluminium alloy AA6016 shows a considerable increase in the maximum burst pressure as well as the maximum von Mises equivalent strain with increasing number of ARB cycles.

CHAPTER 4 Results

143

The achievable burst pressure of the accumulative roll bonded and friction stir welded sheets (dashed lines) is significantly lower than the burst pressure of the accumulative roll bonded sheets (solid lines) for both materials. This indicates that the nugget acts as the weakest link in the friction stir welded material, which will be confirmed later from the von Mises equivalent strain distribution images. a) AA6016-N0 b) AA6016-N4 c) AA6016-N6

d) AA6016-N0 e) AA6016-N4 f) AA6016-N6

Figure 4-47: von Mises equivalent strain distribution and b) the corresponding sample fracture indicated by the arrows of accumulative roll bonded (N0, N4 and N6) and friction stir welded aluminium alloy AA6016. Note that pictures d, e and f were taken from the bottom of the sheets in order to clearly show the nugget zone and the cracks. The nugget runs vertically down the blanks. The circular dashed line indicates the bulge position. The von Mises equivalent strain distribution and the corresponding locations of sample fracture are shown in figure 4-47. The von Mises scale on the right hand side of figures 4-47 a), b) and c) indicates the level of deformation. The highest level of deformation takes place in the nugget (weld zone), irrespective of the material state. Thick bands running vertically across the sample show the position of the weld zone. The investigations have also shown that the fracture of the samples usually takes place on the retreating side (where the rotational direction of the tool opposes the welding direction along the joint line) of the nugget. This is presumably due to the fact that the thickness of the sheets on the retreating side of the nugget is usually slightly smaller than the actual sheet thickness, because of the higher material removal during friction stir welding. The location of cracks for each material state i.e. as-received N0 and accumulative roll bonded N4 and N6 state appears very similar. The cracks initiate at the pole of the blank within the nugget and propagate along the weld zone (Fig. 4-47 d, e and f). In addition, the level of deformation of the nugget after 4 and 6 ARB cycles appears to be higher than that of the as-received state and confirms the measured von Mises equivalent strain values in figure 4-46 b), although the grain size as well as the

CHAPTER 4 Results

144

previously measured hardness distribution of all friction stir welded samples was proved to be similar. Hardness of the friction stir welded CG AA6016-T4 and UFG AA6016 after 8 ARB cycles drops from the base material to the nugget. In both cases, hardness in the nugget reaches ~50 HV0.05. Irrespective of the quality of the weld, the nugget still represents the weakest link in the blank and it is therefore not surprising that the deformation takes place predominantly in the close proximity of the nugget or within the nugget itself. The process of friction stir welding is very complex due to the simultaneous influence of large plastic strain imposed onto the material and the elevated temperature, which results from the friction between the tool and the sheets to be joined. Thus, the microstructure as well as the resulting mechanical properties varies across the weld as well as within the thickness of the welded sheet due to a complex load distribution during welding. Furthermore, it is very likely that the texture of the as-received and accumulative roll bonded sheets changes once the material is friction stir welded. These effects would certainly influence the formability of the friction stir welded sheets, although it may not necessarily be easy to detect their influence by simply measuring the hardness distribution. As a result, in comparison to bulge tests, the hardness measurements may have been too insensitive to detect these changes, which may have developed during friction stir welding. It is also possible that the previous deformation history of the material may play a role regarding the microstructural development during friction stir welding and its deformation behaviour under biaxial stress state conditions. Similar results were also observed for the commercial purity aluminium AA1050. In this context, it is important to note that figures 4-47 a), b) and c) can only be used for a qualitative comparison of the results, since the figures represent an overall distribution of von Mises equivalent strain and the exact values can not be estimated directly from the figures. Cup Drawing Tests Cup drawing tests were conducted on the friction stir welded as-received and accumulative roll bonded sheets in order to investigate the sheet metal formability under tension-compression stress state conditions, which is typical for deep drawing operations. The sheet metal formability of joined components is especially important to investigate, because of the known microstructural changes which occur during welding and the resulting differences in the mechanical properties between the nugget and the base material. Therefore, it is essential to obtain compatible deformation behaviour of the nugget and the base material in order to successfully draw the sheets. The tests were conducted at room temperature. The major difficulty in conducting the cup drawing experiments was the limited number and the limited size of the friction stir welded sheets. Thus, the sample size of the friction stir welded AA1050 and AA6016 aluminium sheets ranged between Ø 80 mm, Ø 85 mm, Ø 90 mm and Ø 95 mm.

CHAPTER 4 Results

145

a) AA1050 N0, Ø 80 mm b) AA1050 N0, Ø 85 mm c) AA1050 N0, Ø 90 mm

d) AA1050 N2, Ø 80 mm e) AA1050 N2, Ø 85 mm f) AA1050 N2, Ø 95 mm

g) AA1050 N4, Ø 80 mm h) AA1050 N4, Ø 85 mm i) AA1050 N4, Ø 90 mm

Figure 4-48: Deep drawn cups of friction stir welded commercial purity aluminium AA1050 of as-received and accumulative roll bonded state after 2 and 4 ARB cycles, drawn at room temperature. The nugget runs horizontally along the blanks. All sheets were 1 mm thick. The arrows indicate the split of the sample at the nugget and the nugget fracture. Figure 4-48 shows a number of cup tests conducted on commercial purity aluminium AA1050. The images include the friction stir welded as-received samples and the friction stir welded samples after 2 and 4 ARB cycles. As-received samples (N0) and samples after 2 ARB cycles were successfully drawn using blank diameters of Ø 80 mm and Ø 85 mm (see Fig. 4-48 a, b, d, e), while all samples with larger diameters fractured. The samples which fractured also included N4 (Ø 80 mm, Ø 85 mm, Ø 90 mm), N6 (Ø 85 mm, Ø 90 mm) and N8 (Ø 90 mm) UFG aluminium sheets (see for e.g. Fig. 4-50 g, h, i). The samples which were successfully drawn showed no earing. There was also no delamination of the bonded layers. However, some samples show that the nugget deforms to a larger extent than the base material

CHAPTER 4 Results

146

(Fig. 4-48 d and e). This observation is not unusual, since the previous hardness measurements across the nugget confirmed material softening (Fig. 4-48 a). However, larger blanks need to sustain higher drawing forces. In case of the Ø90 mm samples, the fracture takes place within the nugget. Similar to the fracture behaviour observed in the hydraulic bulge tests, the samples fracture predominantly along what appears to be the retreating side of the weld. This is presumably due to the fact that the thickness of the sheet on the retreating side of the nugget is usually slightly smaller than the actual sheet thickness, because of the higher material removal during friction stir welding. Thus, the weld is not only softer than the base material, but there is also some thickness variation in the nugget. These inhomogeneities of the nugget appear to be the weakest link of the welded sheet and locally decrease the strength of the material. This was also reflected in the drawing force vs. displacement curves obtained for different samples sizes and different sample states (Appendix, Fig. 8-6 a). In case of the N0 state, the drawing force increases with increasing blank diameter. On the other hand, most ARB samples, including N2, show lower achievable force when drawing larger samples. This is predominantly caused by the lower strength of the nugget and the inability of the nugget to support the increasing loads necessary for the deformation of larger blanks. The limiting drawing ratio (LDR) was determined for N0 and N2 using the Schmidt-Kapfenberg plot (Fig. 4-49), because these were the only two material states which were successfully drawn using two different sample diameters, namely Ø80 mm and Ø85 mm. Due to the fact that the sample N2- Ø95 mm fractured at lower loads than the samples N2- Ø80 mm and N2- Ø85 mm, it was only possible to calculate the theoretical fracture force. For comparison reasons, the theoretical fracture force was also calculated for N0. The theoretical fracture forces based on the ultimate tensile strength of the friction stir welded samples N0 and N2 are similar due to the fact that the materials undergo dynamic recrystallisation during welding. Thus, irrespective of the previous material deformation history, the nugget zones always have similar hardness and strength, as it was already reported in figure 4-45 a. The theoretical limiting drawing ratio (βmax.) was calculated from the line intercepts of the successfully drawn cups and the theoretical fracture force. The theoretical limiting drawing ratio averages approximately 2.22 and 1.80 for AA1050-N0 and AA1050-N2, respectively. The ARB samples show lower drawability than the as-received samples. Similar observations were also obtained for the friction stir welded aluminium alloy AA6016. The as-received samples were successfully drawn into cups using Ø80 mm, Ø85 mm, Ø90 mm blank diameters. However, most of the ARB samples were possible to deep draw only using the smallest sample size or not at all (see appendix Fig. 8-7). Thus, it was not possible to determine the limiting drawing ratio of the FSW AA6016 samples. However, similar to the AA1050 observations, the deformation of the successfully drawn cups predominantly takes place within the nugget, while all fractured cups locally failed on the retreating side of the nugget.

CHAPTER 4 Results

147

Figure 4-49: Schmidt-Kapfenberg plot for friction stir welded commercial purity aluminium AA1050 rolled to N0 and N2 ARB cycles and deep drawn at room temperature. Minimum drawing ratio (βmin.) is presented in figures 4-50 for both friction stir welded materials. This ratio is used to compare the maximum sample diameter which was drawn in this study. In case of FSW AA1050, the minimum drawing ratio varies only slightly with increasing number of ARB cycles. As-received and N8 sample were both successfully drawn using Ø 85 mm samples. The N6 sample was only possible to draw with Ø 80 mm sample diameter. The minimum drawing ratio of FSW AA6016 decreases with increasing number of ARB cycles. Generally, both friction stir welded materials show poor drawability and a relatively low drawing ratio, which averages 1.75 and corresponds to approximately Ø 87.5 mm sample size.

Figure 4-50: Ratio of sample diameter to punch diameter vs. the number of ARB cycles N for FSW AA1050 and FSW AA6016. The values were obtained from the room temperature experiments.

CHAPTER 4 Results

148

4.8 Corrosion Electrochemical behaviour and corrosion resistance of ultrafine-grained materials was rarely investigated and reported in the literature (see literature review: corrosion of UFG aluminium alloys). The corrosion results on UFG materials are controversial and seem to depend strongly on the material system, material production history as well as the corrosive environment. The aim here was not only to investigate the corrosion behaviour of the as-received and ARB processed commercial purity aluminium AA1050 and aluminium alloy AA6016, but most importantly the weld or the nugget zone of the friction stir welded materials. The electrochemical response of the friction stir welded ultrafine-grained materials provides additional information regarding the success of this particular welding process. 4.8.1 Electrochemical (EC) Pen Measurements As-received and ARB processed commercial purity aluminium AA1050 and aluminium alloy AA6016 were successfully friction stir welded. The base material and the nugget of friction stir welded materials were subsequently investigated using an electrochemical pen measuring technique. Figure 4-51 shows current versus distance curves at various potentials ranging between - 400 mV to + 200 mV. An average of ten measurements was taken at each potential. Both figures 4-51 a) and b) reveal higher dissolution of the material at higher potentials (eg. +200 mV). Lower potential (-400 mV) measurements reveal little to no difference of the ten measurements of the current. In this case, the current flow and the surface activity are too low to cause any significant dissolution effects. However, with an increase in potential, the difference between the ten measurements of the current increases and varies between 55 µA and 65 µA. This effect is caused clearly by a significantly higher current flow, higher material dissolution at the surface and higher surface activity. However, the most important observation is that there are no considerable differences between the base material and the nugget, even though it is well known that the microstructural changes which occur after friction stir welding lead a recrystallised grain structure in the nugget. Furthermore, the as-received conventionally grained state shows almost identical corrosion response as the ultrafine-grained material.

CHAPTER 4 Results

149

a) b)

Figure 4-51: Current vs. density measurements across the nugget of friction stir welded commercial purity aluminium AA1050 in a) as-received state and b) after 8 ARB cycles. All measurements were performed in 0.1 mol NaCl solution and at room temperature. Figures 4-52 a)-d) show the current versus density curves for the technically relevant aluminium alloy AA6016 in the as-received and UFG state. Similar to the observation made for AA1050, the current measured for the as-received AA16016 in figure 4-52 a) increases from approximately 10 µA to 62.5 µA with an increase in potential from -300 mV to +200 mV, respectively. However, in this case the difference between the ten measurements taken of the current at lower and higher potentials is reversed. The material seems to be more resistant against the corrosive attack and dissolves more slowly at lower potentials, i.e. -300 mV, while the dissolution of the material at higher potentials i.e. +200 mV seem to occur alredy after the first scan of the surface. On the other hand, the results of ultrafine-grained material N6 (Fig. 4-52 b) resemble those of the as-received and UFG AA1050. The measured current range of the UFG material at +200 mV ranges between 50 µA and 60 µA, indicating slower and more controllable material dissolution at higher potentials (+200 mV). Similar to the results of AA1050, there are no significant differences between the as-received and UFG state.

CHAPTER 4 Results

150

a) b)

c) d)

Figure 4-52: Current vs. density measurements across the nugget of friction stir welded a) as-received AA6016 and b) UFG AA6016 after 6 ARB cycles, and the zoomed in views of the nugget area of c) as-received AA6016 and d) UFG state AA6016 after 6 ARB cycles. All measurements were performed in 0.1 mol NaCl solution and at room temperature. Figure 4-52 c) and d) shows the zoomed in view of the nugget zone of the as-received AA6016 and UFG AA6016, respectively. The as-received condition shows a slight increase of current across the nugget at +200 mV potential (Fig. 4-52 c). This implies that the nugget of the as-received friction stir welded AA6016 is more prone to dissolution than the base material i.e. the nugget area is more anodic in comparison to the rest of the material. The observations of the UFG state revealed exactly the opposite. In this case, the current decreases across the nugget indicating lower dissolution than in the UFG base material and therefore a more cathodic nature of the weld zone. The reason for this material behaviour may be caused by a number of different factors including the grain size, precipitates and oxide particles. Friction stir welding of the as-received CG state leads to a decrease in grain size in the nugget, as it could already be seen from the microstructure in figure 4-44. Therefore, the number of grain boundaries per unit area increases from the base material to the nugget. This means that there are many more interfaces in the nugget prone to a corrosion attack

CHAPTER 4 Results

151

than in the neighbouring base material. Thus, the level of dissolution rises. On the other hand, friction stir welding of the UFG sheets leads to an opposite effect, namely the grain size in the nugget increases compared to the base material. This could easily be explained by the recrystallisation process which occurs as a result of material exposure to elevated temperature during welding. In this case, the number of grain boundaries and hence the number of interfaces per unit area in the nugget decreases compared to the base material. Here, the nugget would be less likely to corrode and the material dissolution would be lower compared to the base material. Furthermore, the influence and the precipitational changes which occur during friction stir welding should not be underestimated. The precipitation size and morphology does not only change after welding the as-received material, but also during the ARB process itself, as well as after welding the ARB sheets. The precipitational changes, which occur during welding of the as-received material are still not completely understood or widely investigated. Therefore, it becomes virtually impossible to predict what occurs after an even more complex sequence of events, such as the ARB process or welding of the ARB sheets. In addition, the oxide particles which are rolled into the material during the ARB process, may also lead to more detrimental effect concerning the material’s corrosion resistance. The oxide particles themselves are not anodic and are therefore not expected to corrode. However, depending on their size, they can form a small galvanic cell around their periphery. This would lead to excess material removal around the oxide particles and possibly cause corrosion pitting. However, due to the high plastic deformation of the material during the ARB process, the oxide particles break down and become very small. It becomes unlikely that they would be a dominating factor influencing the corrosion properties (see Table 4-2). Furthermore, the current increase and decrease across the nugget, as observed in figure 4-52 c) and d) in the as-received and UFG state may be too low for any profound technological implications. 5.8.2 Polarisation Curves As already mentioned in the chapter of experimental procedures, the polarisation curves were taken using the microcapillary technique, which allows for local corrosion measurements over an area of approximately 50 µm to 500 μm. The measurements were only conducted on the aluminium alloy AA6016, since it is known for its second phase particles and iron based inclusions. Figure 4-53 shows the polarisation curves of the as-received and ARB processed, friction stir welded aluminium alloy AA6016. All curves show three different measurements taken at three different areas of the sample. The as-received base material shows that the corrosion potential varies between -700 mV and -600 mV. All three curves show a relatively long stage of passivation followed by a sudden break down of passivity at more positive potentials ranging between -500 mV to -350 mV (Fig. 4-53 a). The polarisation curves of the as-received AA6016 nugget area lie within the same polarisation range. However, the curves show different results depending on the position the measurements were taken. Some areas show some passivation followed by a sudden break down of passivity and others show a very rapid break down of the passive layer (Fig. 4-53 b). Similar results can also be seen for

CHAPTER 4 Results

152

the accumulative roll bonded AA6016 after 8 ARB cycles in figure 4-53 c. The polarisation curves of the nugget zone of the accumulative roll bonded (N8) and friction stir welded AA6016 are shifted to higher potentials and show some level of passivation. This implies that the nugget zone of the accumulative roll bonded materials may have higher corrosion resistance than the accumulative roll bonded base material or the nugget zone of the as-received material. The reasons for different localised corrosion response may be due to a number of factors which are difficult to distinguish from each other. The factors influencing the corrosion behaviour are iron based inclusions, second phase particles as well as the heavily rolled microstructure. a) b)

c) d)

Figure 4-53: Polarisation curves of a) as-received AA6016 base material, b) as-received AA6016 nugget zone, c) UFG AA6016 N8 base material and d) UFG AA6016 N8 nugget zone. All tests were performed in 0.1 M NaCl solution, with polarisation rate of 0.1 mV/sec and at room temperature. The capillary diameter is 220 µm.

CHAPTER 5 Summary and Discussion of Results

153

5 SUMMARY AND DISCUSSION OF RESULTS

5.1 Introduction The importance of ultrafine-grained (UFG) materials, with an average grain size below 1 µm, produced by severe plastic deformation (SPD) techniques, has been recognised by many researchers [Hor05]. Ultrafine-grained materials show especially interesting mechanical properties, namely higher specific strength and/or ductility compared to the conventionally grained (CG) materials with an average grain size between 10 µm and 50 µm. As such, UFG materials have a strong potential for prospective engineering applications for structural, high durability components in the automotive, aerospace and medical industry. The accumulative roll bonding process is an adequate method to produce UFG metal sheets, because it can be easily scaled up and adapted to a conventional rolling process. In order to qualify the accumulative roll bonding process as a new technique for UFG metal sheet production, detailed investigations on microstructural evolution, mechanical properties and sheet metal forming ability using hydraulic bulge tests and cup drawing tests have been carried out and investigated. Furthermore, sheet metal joining is also a technologically important issue and places a further challenge upon the UFG sheet metals. The UFG aluminium metal sheets have especially aroused attention as potential candidates for modern light weight materials in the automotive industry due to their high strength to weight ratio. Thus, this study focuses predominantly on commercial purity aluminium AA1050 and aluminium alloy AA6016, although other materials have also been investigated. The main aims of the study were:

• to produce ultrafine-grained aluminium alloy sheets using the ARB process • to optimise and scale-up the ARB process, as well as increase its robustness • to transfer the ARB process onto other technically relevant materials • to investigate the microstructural development and the mechanical properties • to investigate the potential of sheet metal forming using bulge tests and cup

drawing tests • to adapt the friction stir welding (FSW) process to UFG aluminium sheets and

investigate their formability potential as well as corrosion properties. This chapter summarises and discusses the most important results. It points out the practical challenges of the ARB process, as well as the criteria for the success of the process. The microstructural development and its influence on the mechanical properties of the UFG sheets are discussed, and the sheet metal forming and joining potential, as well as the corrosion properties of the UFG aluminium sheets are highlighted.


154

5.2 ARB Process Optimisation and Transfer of the ARB Process on other Technologically Relevant Materials The ARB process optimisation was predominantly carried out on commercial purity aluminium AA1050. With specifically targeted testing techniques, adapted process parameters and the acquisition of a four-high rolling mill, the ARB process was scaled up, optimised and the quality of the sheets was improved (see Fig. 4-1). The ARB process was significantly shortened and it became more robust. The deformation during rolling became more homogeneous, cracking of the edges was eliminated and crack propagation was suppressed. These factors cumulatively contributed to less material waste during the process. The surface quality was considerably improved and the sheet thickness became more homogeneous. The contribution of a four-high rolling mill was especially manifested in terms of the final width of the metal sheets, which was improved to up to 100 mm. Further increase of the width of the sheets was not possible due to the restricting width of the rolls, averaging 115 mm. Other technologically important materials were also investigated. Besides the commercial purity aluminium AA1050, the ARB process was successfully performed on the following materials:

• aluminium alloy AA6016, • oxygen free copper, • titanium grade two and • niobium.

The only material that could not be processed was magnesium alloy AZ31 (see Fig. 4-7). Accumulative roll bonding of magnesium alloy was not possible due to its high initial hardness and its hcp crystal structure with a limited number of active slip systems at low homologous temperatures. The criteria for successful manufacturing of the UFG sheets using the ARB process can therefore be summarised as follows: - Pure materials and alloy systems Pure materials are generally easier to roll bond because of their initially lower hardness and strength in contrast to alloys. Typical examples are AA1050 and AA6016, which were roll bonded at room temperature and at elevated temperatures, respectively. Additionally, materials with higher stacking fault energy, which are usually pure materials with fcc crystal structure, recover more quickly even during rolling at low homologous temperatures. - Process temperature Higher process temperatures generally lead to better deformation of metal sheets during rolling as well as better interlamellar bonding between the sheets. Especially with regard to alloys, it is important to perform the ARB process at elevated temperatures. However, processing at temperatures above the recrystallisation temperature of the material would destroy the UFG microstructure. Therefore, the


155

process temperature is always a compromise between good metal sheet bonding i.e. interlamellar strength and good thermal stability. The process temperature must be optimised for every material system. Rapid and effective grain refinement was observed for all materials which were rolled by the ARB process. This indicates that the process or homologous temperature (as long as the process temperature does not exceed the recrystallisation temperature) only plays a secondary role concerning the development of the UFG microstructure. - Surface roughness and oxide layer Surface roughness strongly influences the quality and the strength of the bonded layers. Generally, it can be said that a rougher surface finish results in a better bond strength. Furthermore, the removal of the oxide layer is especially important in materials such as aluminium. Aluminium has a thick and a rapidly growing oxide layer, which must be removed mechanically in order to establish a good contact and bond between the two metal surfaces. - Plastic deformation per ARB cycle Most materials require a thickness reduction of at least 50 % per ARB cycle. However, some materials like copper or titanium must be rolled using a slightly higher thickness reduction due to their initially high hardness in order to achieve good bonding. - The c/a ratio of the hexagonal materials Titanium was successfully roll bonded, while magnesium fractured into thin strips during rolling. It appears that the c/a ratio may also have an influence on the processability of hexagonal materials. The c/a ratio of magnesium is approximately 1.62 and that of titanium 1.58. Since the difference between the two c/a ratios is not so high, it is assumed that the difference in processability is due to the fact that titanium was a single phased material, while magnesium was highly alloyed. Generally, pure materials with high stacking fault energy and lower strength are easier to process than alloys. High temperature melting point alloys with bcc or hcp crystal structure and low stacking fault energy are more difficult to process. Irrespective of the process parameters, the UFG microstructure was successfully produced from the conventionally grained commercial materials and there was significant strengthening of the samples during the ARB process. 5.3 Mechanical Properties of ARB Materials All ARB rolled materials showed an increase in hardness, yield strength and tensile strength with increasing number of ARB cycles N (Figs. 4-9, 4-11, 4-12 and 4-13). Depending on the material purity, the strengthening effects may include solid solution, dislocation, grain boundary and precipitation strengthening. Hardness and yield strength of UFG materials are typically 2 to 3 times higher than their conventionally grained counterparts. The increase in tensile strength is somewhat smaller and reaches 1.2 to 1.5 times that of the as-received materials. All materials were found to reach saturation in hardness and strength. The number of ARB cycles needed to reach the


156

saturation level strongly depends on the material and the process temperature. Pure materials, such as commercial purity aluminium AA1050 and oxygen free copper Cu 99.99 tend to reach the saturation level already after the first few cycles. It appears that these materials are able to recover more easily than alloys during the ARB process. Alloying elements and second phase particles in alloys retard recovery by e.g. grain boundary or dislocation pinning. As a consequence, the saturation in hardness is postponed to higher ARB cycles, as it could clearly be seen in case of the aluminium alloy AA6016.

Figure 5-1: Summary of mechanical properties of various materials produced by the ARB and ECAP process. All literature values* for the CG Aluminium can be found in [Val04, Bra92, Dav93, Bei01]. The mechanical properties of the ECAP UFG AA1050 samples were taken out of [May04]. Figure 5-1 shows an overview of the mechanical properties of the ARB materials investigated in this study and compares them to the values of CG materials taken from the literature, as well as to the UFG AA1050 samples processed by ECAP. The CG aluminium samples generally have low strength, but the elongation to failure can vary between 5 % and 55 % depending on the sample condition. The ARB processed materials show very high strength, but generally limited elongation to failure of up to 10 %, which is usually associated with severely deformed materials with UFG structures. The elongation to failure decreases after the first few ARB cycle and is significantly lower in comparison to the as-received state. Once the UFG microstructure starts to develop, the elongation to failure generally tends to increase again with increasing number of ARB cycles. This behaviour was shown by May


157

[May04] for the AA1050 commercial purity aluminium. Strength as well as the elongation to failure progressively increased with increasing number of ARB cycles. However, the elongation to failure which can be achieved after 8 ARB cycles is still lower in comparison to that of the conventionally grained, solutionised materials. The increase in the elongation to failure with increasing number of ARB cycles was reported in this study for AA1050, partly for AA6016 (Fig. 5-1) and partly for OFE copper. OFE copper showed different results to other ARB processed materials. In this case, the strength continuously decreased with increasing number of ARB cycles. Additionally, there was a fluctuation in elongation to failure after every second cycle. In fact, it seemed that the material softened after additional rolling (from N2 to N4) and again hardened (from N6 to N8). Up to now, the reason behind this behaviour was not investigated in detail. Compared to the ARB processed AA1050 samples, the samples processed by ECAP show similar strength, and an improved elongation to failure, which lies between 30 and 40 %. The reasons for limited elongation to failure usually found in the ARB processed materials, is discussed in more detail in the following paragraph. Limited elongation to failure can be explained by the limited amount of work hardening and premature plastic instability. Typical tensile test curves of most UFG materials show very little work hardening and a rapid increase of flow stress, which is usually followed by material softening and limited macroscopic necking (see for example Fig. 4-12). Researchers showed that the work hardening rate (dσ/dε) does not increase with grain refinement [Tsu02b]. As a result, plastic instability or necking in tensile deformation occurs prematurely in UFG materials, which leads to limited uniform and/or total elongation. Even though the microstructure was strongly deformed and the samples showed limited elongation to failure, the fracture surfaces of ARB samples after tensile testing revealed a typical ductile fracture with dimple-like features and some extend of plastic deformation in the form of localised necking of individual bonded layers (Fig. 4-28). However, the biggest disadvantage was reflected in the break down of the last or the second last bonded layer during tensile testing. The last bonded layer forms during the final rolling step and represents the weakest link in the chain, and possibly influences and limits the elongation to failure of the whole sample. Therefore, the difference in elongation to failure between the ARB and ECAP processed samples lies in the fact that the ARB sample are a “composite-like” materials consisting of many bonded layers and the relatively poor bonding particularly of the last layer, which strongly limits the elongation to failure. Other than the ARB samples, the ECAP samples have no bonded layers and thus no potential interfaces within the material where a premature failure may initiate. In addition, the ECAP processed materials are exposed to higher deformation per cycles. They develop a higher volume fraction of high angle grain boundaries, which potentially leads to more pronounced strain rate sensitivity than in the case of the ARB processed samples. The samples with more pronounced strain rate sensitivity and higher strain rate sensitivity exponent would also be more likely to have higher elongation to failure, as the case may be for ECAP processed samples.


158

As indicated earlier, ultrafine-grained materials are usually associated with enhanced strain rate sensitivity (Figs. 4-14 and 4-15). Accumulative roll bonded AA1050 and AA6016 showed pronounced strain rate sensitivity even at room temperature. High strain rates lead to higher strength and lower elongation to failure, whereas lower strain rates lead to reduced strength and higher elongation to failure. This effect is especially important with regard to deformation behaviour of UFG materials. Enhanced strain rate sensitivity increases ductility and stabilises necking. In industrial metal forming processes such as deep drawing, better formability would be achieved at lower strain rates. In order to improve the process efficiency, it is more economical to perform sheet metal forming at higher strain rates. In this case, higher process temperature would lead to similar effects as lower deformation rates and increase the forming potential of UFG materials. With regard to this, UFG materials have a higher sheet metal forming potential than their conventionally grained counterparts, which have proved to be strain rate insensitive. Strain rate sensitivity is more pronounced in the face centred cubic UFG materials due to the higher fraction of grain boundaries, which are the most favoured sites for dislocation activities such as dislocation storage or dislocation annihilation. Because defects like grain boundaries are paths of accelerated diffusion processes (pipe diffusion) it is reasonable to assume that diffusivity in UFG materials may be much higher than in the CG materials, as shown in [Div08]. This would lead to easier dislocation annihilation, resulting in enhanced recovery. Strain rate sensitivity is strongly dependent on the material and the testing temperature. Pure materials generally have a higher strain rate sensitivity exponent due to easier dislocation annihilation at grain boundaries, because they are not obstructed by alloying elements or second phase particles. The effect of strain rate sensitivity becomes even more prominent at higher temperatures. The process becomes dependent on the thermally activated dislocation annihilation controlled by the dislocation climb at grain boundaries. In alloys, solid solution strengthening and precipitation strengthening may become less effective at elevated temperatures. The alloying elements provide less resistance to dislocation motion, because the thermally activated dislocation processes such as dislocation climb become more enhanced. Additionally, the precipitates grow and change their morphology, which can also result in less effective barriers to dislocation motion. In case of the OFE copper, higher strain rates lead to higher strength as well as higher ductility, whereas lower strain rates show lower strength and lower ductility, for room temperature testing. This behaviour deviates from most literature results published regarding strain rate sensitivity and the exact reasons are not yet clear. It is however possible that the testing temperature as well as the high purity of the material may have considerable effects. On the other hand all titanium samples, CG and UFG, have proven to be strain rate insensitive at room temperatures. This is believed to be due to the low homologous testing temperature. The mechanical properties of ARB processed aluminium and aluminium alloys have proven to be anisotropic (Figs. 4-16 and 4-17). Tensile testing performed on AA1050 and AA6016 in three different directions showed pronounced differences between the


159

samples oriented at 0° and 90° to the rolling direction and the ones oriented at 45° to the rolling direction. A considerable increase of strain to failure was observed for samples oriented at 45° to the rolling direction, while the maximum stress was comparable to that of samples at 0° and 90° to the rolling direction. Surprisingly, yield strength, tensile strength and elongation to failure were found to be similar for samples tested in the rolling direction and the transverse direction. One possible reason for the higher elongation to failure of samples oriented at 45° to the rolling direction might be related to the crystallographic texture. 5.4 Microstructural Development during the ARB Process and Thermal

Stability of ARB materials All materials developed the characteristic elongated grain structure with a large aspect ratio and cells or subgrains within the grains as a consequence of accumulative roll bonding. A typical elongated, ribbon like structure (Fig. 4-21) already develops during the first few cycles, while the ultrafine-grained microstructure emerges after 3 to 4 ARB cycles. It was recognised that the initial microstructure was strongly inhomogeneous after the first few ARB cycles and developed as a result of a non-uniform shear strain (Figs. 2-5 and 4-10) distribution through the thickness of the sheets. The microstructural refinement was found to be more pronounced near the interface of the two metal sheets and next to the top and bottom metal sheet surfaces in contact with the rolls (Fig. 4-23). However, the microstructure becomes more homogeneous after a higher number of ARB cycles as a result of repeated rolling. Indications of microstructural homogeneity were confirmed on AA1050 samples rolled up to 20 ARB cycles, which did not show a typical rolled microstructure (Fig. 4-26). Furthermore, with increasing number of ARB cycles, the texture develops into a typical rolling texture, a characteristic ß-fibre texture with a Cu component [Skr07]. The important factors which collectively affect the microstructural development and hence the mechanical properties, formability and corrosion resistance, include the material crystal structure and its stacking fault energy, alloying elements and second phase particles. The microstructural evolution during ARB strongly depends on the type of material being rolled. Generally, it can be said that materials with the same crystal structure tend to respond similarly to ARB processing and as a result similar microstructures can be expected to develop. The finest microstructure and the fastest grain refinement during the ARB process can be achieved for high temperature melting point alloys with bcc or hcp crystal structure and low stacking fault energy. In the case of commercial purity aluminium AA1050, the grain size remains larger than that of aluminium alloy AA6016 (see Fig. 4-21). It was recognised that alloying elements and especially the second phase Mg2Si particles in the aluminium alloy pin down the grain boundaries and retard the dynamic recovery process, which takes place during accumulative roll bonding as a result of a high driving force. However, the exact influence of the precipitates is not easy to estimate, because they continuously grow and age during the ARB process.


160

For this reason, the effect of the second phase magnesium-silicon particles was investigated in the CG and UFG AA6016 samples in more detail. Room temperature aging of the solutionised AA6016 in the CG state leads to a rapid hardness increase from 40HV10 to 70HV10 within 7 days. Artificial aging showed that the material hardness may even increase after 20 hours of annealing at 180°C (Fig. 4-29). Similar observations were also made by the DSC measurements, which show a clear shift of DSC curves of aged specimens to higher temperature (Fig. 4-31 a). Since the evolution of second phase particles contributes to the continuously changing mechanical properties, it is important to define the initial state of the material. For this reason, the conventionally grained AA6016 sheets were always solutionised before conducting the ARB process. The second phase particles precipitate during furnace annealing and during the ARB process. Post roll annealing of UFG sheets revealed that it is still possible to obtain a slight increase in hardness due to precipitation hardening effects. The precipitates can grow and change their morphology and therefore further contribute to an increase in hardness. Similar observations were also confirmed by the DSC measurements. Samples rolled to 2 or 4 ARB cycles show small peaks at 320°C and at 340°C, which may be due to the precipitational effects (Fig. 4-31 b). However, due to superposition of precipitational evolution with recovery and/or recrystallisation, it is not possible to conclude if the peaks arise as a consequence of precipitational effects during annealing. Thermal stability of UFG materials is generally known to be low (see Figs. 4-32 and 4-33). Thermal stability of AA6016, copper and titanium was determined from the post-annealing treatment of the ARB processed sheets. It was found that the ultrafine-grained AA6016 alloy loses its thermal stability at approximately 200°C and recrystallised at 300°C. High purity oxygen free copper (99.99% Cu) loses its thermal stability at 100°C and completely recrystallises at 200°C. As a rule of thumb, it can be said that most ARB processed materials lose their thermal stability at approximately the same temperature at which they were processed or at a temperature 100°C higher than their process temperature. Thermal stability is strongly dependent on the material purity, second phase particles and/or oxide particles which may break up during rolling and lead to some dispersion strengthening. In the case of aluminium alloys, the effect of dispersion strengthening is believed to be minimal because of the previously treated, i.e. wire brushed surfaces. 5.5 Formability of CG and UFG Aluminium Sheets Two types of tests were conducted in order to investigate the formability of the CG and UFG aluminium alloys AA1050 and AA6016, namely hydraulic bulge tests and cup drawing tests. During bulge testing the metal sheets are subjected to biaxial stress conditions. During cup drawing tests, the sheets are subjected to a more complex, tension-compression condition. Both tests were representative of many important industrial processes such as stretching and drawing, and reflect the forming potential of the UFG aluminium sheets. Prior to bulge testing, the geometry of the die and the sample size and thickness must be optimised. The ratio between the sample thickness and the die diameter must be


161

such as to obtain the deformation through the thickness of the sheet and to avoid premature failure of samples. The premature failure may arise as a consequence of higher bending stresses at the transition radius of the die. Successful hydraulic bulge tests were conducted on AA1050 and AA6016 aluminium sheets having 0.8 mm and 0.5 mm thickness, respectively. Burst pressures registered for the AA1050 and AA6016 samples significantly increased with increasing number of ARB cycles N. Burst pressure of the AA1050 samples increased from approximately 52 bar in the as-received state and reached a maximum of 70 bar after only four ARB cycles and subsequently leveled off. In the case of AA6016, burst pressure increased by a factor of two from the as-received CG state to the UFG state after 8 ARB cycles. Metal forming potential is strongly reflected in the maximum obtainable strain, here the von Mises equivalent strain. The accumulative roll bonded samples after 2, 4 and 6 ARB cycles of AA1050 also showed significantly higher von Mises equivalent strain compared to the as-received, CG sample (N0). Unfortunately, this was not the case for aluminium alloy AA6016. Other than by commercial purity aluminium, the maximum von Mises equivalent strain decreased from approximately 20 % in the as-received state (T4 state) to 10 % after 8 ARB cycles (Figs. 4-34, 4-35 and 4-38). Cup drawing tests of commercial purity aluminium AA1050 and aluminium alloy AA6016 were conducted at room temperature and at 180°C. Room temperature experiments of AA1050 showed comparable drawability in terms of the limiting drawing ratio (βmax.) of the CG and UFG samples (Fig. 4-41). However, the UFG samples showed significantly higher fracture forces than the ones obtained for the CG samples (Fig. 8-4 a). On the other hand, UFG samples of aluminium alloy AA6016 generally showed poor drawability compared to the CG counterparts. Most AA6016 UFG samples fractured during drawing (Fig. 8-5) and the limiting drawing ratio could not be calculated (Fig. 4-42). Nevertheless, UFG sheets showed surprisingly good formability during bulge testing and deep drawing, even though the samples were deformed at much higher strain rates than those investigated in uniaxial tensile tests. Because UFG materials are strain rate sensitive, higher ductility and better formability of these materials could be achieved by testing at lower strain rates or at higher temperatures. Cup drawing tests confirmed that considerable improvement of drawability can be achieved at an elevated temperature of 180°C (Fig. 4-43). In this case, all CG and UFG samples were successfully drawn even using the largest possible sample diameter of Ø110 mm. Experiments have also shown that the UFG samples maintain their strength at this working temperature. This characteristic makes the UFG aluminium sheets especially interesting for deformation processes at slightly elevated temperatures. Ultrafine-grained aluminium sheets maintain high strength, show good formability and therefore have a strong potential for further industrial applications and processes.


162

5.6 Friction stir welding of CG and UFG Aluminium Sheets The ultrafine-grained sheets of commercial purity aluminium AA1050 and aluminium alloy AA6016 were processed by accumulative roll bonding and subsequently friction stir welded. The microstructure and the mechanical properties of the welded zone were investigated. It was confirmed that ARB is an effective process for microstructural refinement and that the fine-grained microstructure can be maintained within the nugget after FSW (Fig. 4-44). Microstructural investigations showed that the grain size decreases after friction stir welding of the conventionally grained material AA6016, but increases after welding the ultrafine-grained material due to dynamic recrystallisation. In both cases, the nugget zone appears to be recrystallised, equiaxed and homogeneous with a grain size of 1 µm to 5 µm. Although a drop in hardness is measured in the nugget for both materials, the hardness is comparable to the friction stir welded CG counterparts and is not considered to be a technological limitation (Fig. 4-45). It is also evident that pre-deformation history has no influence on the resulting microstructure and the mechanical properties within the nugget, since both, the CG as well as the UFG friction stir welded samples showed the same mechanical properties in the nugget. Further improvement of FSW is possible by minimising the sheet thickness variation, waviness and increasing the width of the sheets for better heat conductivity and handling, as well as changing the geometry or size of the pin. Altering the process temperature may also be an interesting point for further investigations. Hydraulic bulge tests showed that the achievable burst pressure of the accumulative roll bonded and friction stir welded sheets is significantly lower than that of the accumulative roll bonded sheets for both materials, namely AA1050 and AA6016 (Fig. 4-46). Significantly lower burst pressure of the welded sheets was caused by the localised deformation of the nugget. Generally, the accumulative roll bonded and friction stir welded AA6016 sheets showed more promising formability in terms of burst pressure and von Mises equivalent strain compared to the accumulative roll bonded and friction stir welded AA1050. Poor formability of FSW sheets was also confirmed by the cup drawing tests (Figs. 4-48 and 8-10). The drawability of the friction stir welded AA1050 and AA6016 is generally worse than the drawability of non-welded samples. The drawability of the FSW samples is governed by the deformation and strength of the nugget. The nugget is the weakest link within the welded sheet due to the fact that the strength of the nugget is significantly lower than that of the base material. Thus, the major part of the deformation takes place within the nugget. For small sample sizes, the strength of the nugget is high enough to overcome the drawing force. However, with increasing sample size, the necessary drawing forces also increase. At some point, the sample size becomes too big and the nugget can not transfer the drawing force into the deformation zone, i.e. the flange. Furthermore, the formability of the FSW material may also be affected by the slight variation in sheet thickness and the mechanical properties, as well as the possible changes in texture after FSW. It is believed that a significantly better drawability would be achieved by performing the cup drawing tests at elevated temperatures. In this case one also has to insure that the process temperature does not exceed the recrystallisation temperature which would eliminate the rest of the UFG microstructure of the base material and


163

possibly lead to grain growth within the nugget. Another possibility of improving the general formability of the sheets is to insure a more homogeneous thickness across the nugget. 5.7 Corrosion of CG and UFG Aluminium Sheets Corrosion tests were performed using electrochemical-pen measurements on commercial purity aluminium AA1050, which showed that there was no significant difference in corrosion behaviour between the friction stir welded conventionally grained and the friction stir welded ultrafine-grained materials (Fig. 4-51). Three most important observations included: - no difference in corrosion properties between the base material of the CG state

and the base material of the UFG state - no considerable difference in corrosion properties between the nugget of the CG

material and the nugget of the UFG material - no significant difference between the base material and the nugget zone in case

of the CG and the UFG states. Slightly different results were obtained for the friction stir welded aluminium alloy AA6016 using the same testing technique and the same testing conditions (Fig. 4-52). The friction stir welded CG sample showed very fast material dissolution at higher potentials compared to the UFG state where the current increases gradually during the course of 10 scans. Another important difference between the CG and UFG state was the different corrosion behaviour across the nugget zones. The current across the nugget in the CG sample and the UFG sample increased and decreased, respectively. There is evidence which suggests that the nugget zone of the CG sample behaves more anodically compared to the base material, while the nugget of the UFG material behaves more cathodically compared to the base material. Even though the difference between the current measured in the base material and across the nugget may be very small, the UFG samples definitely show a more advantageous effect. Results suggest that the corrosion would predominantly take place in the base or parent zone and there would be no localised attack of the nugget. The reason for this material behaviour may be caused by a number of different factors including the grain size, precipitates and oxide particles. The grain boundaries can act as potential corrosion sites. After friction stir welding the CG material, the grain size in the nugget decreases and therefore, there are more potential sites for the corrosive attack. On the other hand, after friction stir welding the UFG material, the grain size in the nugget increases and the number of the potential corrosion sites becomes lower. Thus, the transition from the base material to the nugget and the corresponding change of the grain size may have an important effect on the corrosion properties and the amount of material dissolution. Furthermore, the influence of precipitates and the precipitational changes which occur during friction stir welding should not be underestimated. The precipitation size and morphology does not only change after welding the as-received material, but also during the ARB process itself, as well as after welding the ARB sheets. Furthermore, the oxide particles may affect the corrosion properties, but the effect is expected to be


164

minimal, since the oxide particles break down during rolling and become too small to significantly alter the material response to corrosive attack. Generally, both materials, AA1050 and AA6016, showed similar corrosion response with regard to the current measured in the base material and the nugget. However, aluminium alloy AA6016 in the CG state appears more susceptible to corrosion than commercial purity aluminium AA1050 in the CG state. This was confirmed by the very sudden and instantaneous material dissolution of AA6016 at higher potentials, which could be caused by the higher content of alloying elements, as well as second phase particles and impurities. At this stage it is also important to note that the EC-Pen method provided the corrosion measurements and response of the bulk material and a relatively large surface area, while the microcapillary technique provided a more localised feedback. Irrespective of the material state or zone where the polarisation curves were taken, the measurements differed considerably. The reasons for different localised corrosion response may be due to a number of factors including iron based inclusions, second phase particles as well as the heavily rolled microstructure. 5.8 Recommended ARB process parameters for optimal material

properties Most ARB processed materials, which were investigated in this work, reach a constant grain size and show saturation in hardness after 4 ARB cycles irrespective of the process temperature. Further roll bonding does not lead to considerable decrease in grain size. As a compromise between good mechanical properties and process efficiency, it is recommended to process the materials up to at least 4 ARB cycles. However, tensile tests of AA1050 and AA6016 aluminium alloys showed that enhanced elongation to failure can be achieved if the samples are rolled up to 8 ARB cycles. Samples taken at 45° to the rolling direction have shown a significant increase of the elongation to failure from N4 to N8. Table 5-1: Suggested number of ARB cycles for the most favourable compromise between good strength and good elongation to failure or formability under different stress state conditions

Hardness Uniaxial Tensile Test

Biaxial Bulge Tests

Deep Drawing Tests

Anisotropy

RT RT RT RT 180°C RT RD/TD 45° <r> Δr

AA1050 RT roll bonding N4 N4 N8 N4 N6 N6 × ×

AA6016 230°C roll bonding

N4, N6 N4 N8 N4 × N4, N6 N4 N4

Room temperature biaxial bulge tests indicated that the ARB AA1050-N4 and AA6016-N4 samples can reach higher burst pressures in comparison to the as-received samples. However, ultrafine-grained aluminium AA6016-N4 showed particularly


165

limited von Mises strain. In this case, sheet metal forming under biaxial tensile stress conditions of UFG AA6016 samples should be performed at elevated temperatures. Cup drawing tests performed at room temperature showed that all AA1050 samples can be successfully drawn. On the other hand, UFG aluminium AA6016 samples showed limited formability. Room temperature anisotropy investigations on AA6016 suggested that the best compromise between reduced sheet thinning and earing during drawing can be achieved for samples rolled up to 4 ARB cycles (Fig. 4-19). Higher ARB cycles improve resistance to metal sheets thinning, but they increase earing. Promising results under tension-compression type conditions were obtained at an elevated temperature of 180°C for both materials. In order to obtain a good compromise between strength and formability under different stress states, the suggested number of ARB cycles for AA1050 and AA6016 aluminium alloys can be taken from table 5-1. Friction stir welded UFG AA1050 and AA6016 samples showed no particular difference in hardness values compared to each other or to the as-received state, since it was evident that pre-deformation history has no influence on the resulting microstructure. However, room temperature biaxial bulge tests showed an increase in burst pressure from N0 to N4 for both materials. Generally, the accumulative roll bonded and friction stir welded AA6016-N6 sheets showed more promising formability in terms of burst pressure and von Mises equivalent strain compared to the accumulative roll bonded and friction stir welded AA1050. It can therefore be recommended to roll the materials up to at least 6 ARB cycles for promising results under biaxial stress state conditions.


166

CHAPTER 6 Conclusions

167

6 CONCLUSIONS

In this study, a new Accumulative Roll Bonding (ARB) technique for processing of ultrafine-grained (UFG) materials from the conventionally grained (CG), commercial sheet materials, was investigated. The focus of research was placed on the ultrafine-grained aluminium alloys, due to their high specific strength (strength to weight ratio) and strong potential as light weight structural components in the automotive industry. The main objectives of the study were to manufacture the ultrafine-grained sheet materials, to optimise the ARB process and increase its robustness, to transfer the process on other technologically interesting materials and investigate the microstructure, mechanical properties, formability and corrosion properties, as well as adapt a joining process (in this case friction stir welding, FSW) to the UFG aluminium sheets. The most important general conclusions that may be drawn from the results are as follows: ARB Process Optimisation and Transfer of the ARB Process on other Technologically Relevant Materials The ARB process optimisation was predominantly carried out on commercial purity aluminium AA1050. The process was optimised with regard to the overall quality of roll bonded metal sheets, including:

more homogeneous sheet deformation reduced sheet thickness variation along the sheet width and length suppression and elimination of edge cracking improved surface quality improved final width of the sheets up to 100 mm.

Homogeneous metal sheet rolling and at the same time reduced tendency to edge cracking, cumulatively contributed to less material waste. The ARB process was significantly shortened and it became more robust. At this stage, it is important to highlight the fact that this study may be the only published work on the challenges of the ARB process and process optimisation [Top07, Top08b]. Most other researchers only published data for UFG sheets with a maximum width of 30 mm to 50 mm [Sai98, Tsu03b, Sla07]. Scaling up of the process does not only allow for additional testing possibilities of the UFG sheets (such as mechanical anisotropy, sheet metal forming processes and etc.), but it also introduces new challenges which may be present during UFG metal sheet processing on a larger scale.


168

This study also incorporates the investigation of some of the materials that have been rarely researched, such as titanium [Ter07] or the ones that have not been researched at all, such as niobium. Accumulative roll bonding of niobium is presented for the first time in this study. Mechanical Properties of the ARB Process Materials

The UFG materials, produced by the ARB process generally show higher hardness, yield and tensile strength compared to their CG counterparts, but unfortunately only show limited ductility. Hardness and yield strength of UFG materials are typically 2 to 3 times higher than that of the conventionally grained counterparts. The increase in tensile strength is somewhat smaller and reaches 1.2 to 1.5 times that of the as-received materials. The elongation to failure decreases after the first few ARB cycles and is significantly lower in comparison to the as-received state, usually solutionised state. In the case of aluminium alloys, the maximum elongation to failure in the rolling direction reaches up to 10 %. It has to be pointed out that when comparing the elongation to failure of the CG and UFG samples, it is always important to note the condition of the as-received (CG) material. The elongation to failure of UFG sample may appear too low compared to the CG samples, which have been solutionised. However, compared to a cold rolled state, the elongation to failure of UFG samples is similar or even better than for the CG samples.

Limited elongation to failure of the ARB processed materials compared to the

solutionised CG counterparts is attributed to two main factors: limited amount of work hardening and low interlamellar bond strength of the last bonded layer. As a result, limited work hardening, plastic instability or necking in tensile deformation occurs prematurely in UFG materials, leading to limited uniform and/or total elongation and premature sample failure. Low interlamellar bond strength of the last bonded layer additionally contributes to premature failure, since it results in delamination of the sheets.

The UFG materials such as commercial purity aluminium AA1050 and

aluminium alloy AA6016 are strain rate sensitive even at room temperature. High strain rates lead to higher strength and lower elongation to failure, whereas lower strain rates lead to reduced strength and higher elongation to failure. Other materials, such as the UFG titanium were found to be strain rate insensitive at room temperature. All CG samples were more or less strain rate insensitive.

The mechanical properties of ARB samples have proven to be anisotropic. A

considerable increase in elongation to failure was observed for samples oriented at 45° to the rolling direction compared to samples in the rolling direction and those taken at 90° to the rolling direction. Higher elongation to failure of samples oriented at 45° to the rolling direction may be related to the crystallographic texture, which develops into a characteristic ß-fibre texture with a Cu component. The anisotropy values (Lankford parameter r) increased


169

and decreased with increasing number of ARB cycles for samples oriented at 45° to the rolling direction and for samples oriented in the RD and TD, respectively. This indicates a possible change in the strain state during tensile testing and therefore different values of elongation to failure. Another possible explanation is that the thermally activated recovery of dislocations at grain boundaries is triggered by the ß-fibre texture.

Microstructure of the ARB processed materials

The ARB microstructure is initially strongly inhomogeneous through the thickness of the sheets (after one or two ARB cycles). However, with increasing number of ARB cycles and the repeated stacking and folding of the sheets, the microstructure becomes more homogeneous.

All materials develop a typical elongated, ribbon like structure during ARB

processing. The grains are characterised by the large aspect ratio and cells or subgrains within the grains. The ultrafine-grained microstructure generally emerges after 3 to 4 ARB cycles.

Thermal stability of the ARB processed materials was found to be low. As a

rule of thumb, most ARB processed materials loose their thermal stability at approximately the same temperature at which they were processed or at a temperature 100°C higher than their process temperature.


170

Formability of CG and UFG Aluminium Sheets In this study, the formability of UFG aluminium sheets on a larger scale has been investigated for the first time. The hydraulic bulge tests and cup drawing tests were conducted using a 58 mm and 50 mm die diameters, respectively. The size of the ARB sheets was more than 100 mm wide. A previous publication of Lapovok et al. [Lap07a] introduced cup drawing tests on UFG aluminium cups, which were only 1 mm in diameter. Although some material behaviour tendencies may be drawn from these results, the conclusions regarding formability of the UFG sheets tested in the study conducted by Lapovok et al. have to be considered under the very strong constriction of the very small sample size investigated. The following conclusions can be made from the formability tests in this work:

Ultrafine-grained materials generally showed promising sheet metal forming potential compared to their conventionally grained counterparts.

During room temperature hydraulic bulge testing the samples showed a

tendency to higher achievable burst pressures and/or von Mises equivalent strains with increasing number of accumulative roll bonding cycles. Improvement in formability of the UFG aluminium sheets can be achieved at elevated temperatures. This was shown by conducting cup drawing tests at 180°C.

Room temperature deep drawing cup experiments on AA1050 showed

comparable drawability in terms of the limiting drawing ratio (βmax.) of conventionally grained and ultrafine-grained samples, although the ultrafine-grained samples showed significantly higher maximum drawing forces than the ones obtained for the conventionally grained samples. On the other hand, ultrafine-grained AA6016 generally showed poor drawability at room temperature compared to the conventionally grained counterparts. However, drawability of the ultrafine-grained samples could be significantly improved at elevated temperatures, while the maximum drawing force remained higher than that of the conventionally grained samples. This characteristic makes the UFG aluminium sheets especially interesting for deformation processes at elevated temperatures. Ultrafine-grained aluminium sheets maintain the high strength, show good formability and therefore have a strong potential for further industrial applications and processes.

Deep drawing at elevated temperature resulted in earing of some CG and UFG

samples, which is believed to be linked to the microstructural anisotropy and/or texture effects. Room temperature analysis of <r> and Δr have indicated that a good compromise between sheet thinning and earing can be achieved if the samples are rolled up to 4 ARB cycles.


171

Friction stir welding and corrosion properties of CG and UFG Aluminium Sheets Besides a publication by Sato et al. [Sat04], who investigated friction stir welding of the UFG commercial purity aluminium, this is the only known study regarding friction stir welding of an aluminium alloy. Furthermore, this study is the first to report the investigation on the corrosion behaviour of friction stir welded ultrafine-grained aluminium sheets.

Ultrafine-grained AA1050 and AA6016 sheets were successfully joined by friction stir welding and the fine-grained microstructure was maintained after FSW.

Microstructural investigations showed that the grain size in conventionally

grained material decreases after friction stir welding, but increases after welding of the ultrafine-grained material due to dynamic recrystallisation. Although a decrease in hardness was measured across the nugget for both materials (AA1050 and AA6016), the hardness is comparable to the CG counterparts and it is not considered to be a technological limitation.

Both, the hydraulic bulge tests and the cup drawing tests confirmed lower burst

pressure or drawing forces, as well as limited formability in terms of the von Mises strain or the limiting drawing ratio, of the accumulative roll bonded and friction stir welded sheets. In most cases, the formability of the friction stir welded AA1050 and AA6016 samples is generally worse than that of the non-welded samples, because it seems to be governed by the deformation and strength of the nugget.

Corrosion tests performed using the electrochemical-pen measurements on

commercial purity aluminium AA1050 showed that there was no significant difference in corrosion behaviour between the friction stir welded conventionally grained and the friction stir welded ultrafine-grained sheets.

Friction stir welded AA6016 CG sample showed very fast material dissolution

at higher potentials compared to the AA6016 UFG state. The nugget zone of the CG sample behaves more anodically compared to the base material, while the nugget of the UFG material behaves more cathodically compared to the base material. Even though the difference between the current measured in the base material and across the nugget may be very small, the UFG samples definitely show a more advantageous effect. However, the corrosion properties of the materials vary considerably depending on the region (e.g. base material or nugget) of testing. The reasons for the different localised corrosion response may be due to a number of factors including the large iron based inclusions, second phase particles as well as the heavily rolled dislocation rich microstructure.


172

CHAPTER 7 Outlook

173

7 OUTLOOK

7.1 Further improvement of ARB process Although UFG materials have many interesting and enhanced qualities compared to their CG counterparts, there are still a number of parameters, which can be employed in order to improve the quality of the ARB processed sheets which would also lead to better mechanical properties. These include: 1. Upscaling of the ARB process Increasing the size of the ARB processed sheets would broaden the margin of the possible testing techniques which can be used in order to evaluate the mechanical properties and metal sheet formability. Larger samples for biaxial bulge tests and deep drawing tests would generally lead to easier handling, avoidance of undesirable bending stresses during bulge testing, or a possibility to determine the forming limiting ratio of samples deep drawn at elevated temperatures. 2. Sheet waviness and sheet tolerance Small rolls used in the four-high rolling mill increase the pressure during rolling, but they also contributed a more difficult grabbing and pulling of the sheets through the rolls as well as some sheet waviness. The spacing between the waves on the metal sheet surfaces equals the width of the deformation zone and averages 4 mm. A slight increase of the roll diameter would increase the deformation zone and allow for easier roll bonding. 3. Improvement of interlamellar bond strength While some interfaces are rolled more often than others, the last sheet to sheet interface is rolled only once and it is therefore the weakest link in the whole ARB processed metal sheet. Improving the interlamellar bond strength would also result in an improved elongation to failure, which is limited due to premature necking in tension. There are a number of parameters which can be varied in order to improve the interlamellar bond strength of the last layer.

• Higher number of rolling cycles can lead to a more homogeneous microstructure (as it was observed in AA1050), although it must be insured that the material does not soften if it undergoes many pre-heating cycles during ARB.

• A higher process temperature would make the material and the surface asperities softer, resulting in better metal to metal adhesion contact of surfaces and therefore better interlamellar bond strength.

• Higher thickness reduction of the last cycle would cause a more effective break down of the oxide layer and the clean oxide free surfaces would lead to better adhesion of sheets.

CHAPTER 7 Outlook

174

• Higher surface roughness can be obtained by e.g. conventional pre rolling of the sheets with textured roll surfaces. The textured rolls would impregnate the surface features i.e. increase the roughness of the sheets, which can then be used advantageously for further ARB processing. Higher surface roughness would result in better mechanical interlocking of surface asperities.

• Chemical removal of the oxide layer would remove the oxide layer more efficiently and further improve the metal to metal contact. Although this method is difficult to implement under small laboratory conditions, the adhesion bonding of clean surfaces would be improved and the strength of the bonded layer would increase.

• Reducing the rolling speed would lead to less strain hardening of surface asperities and longer contact time, which would insure better interlamellar bonding.

Even though there is still very limited amount of research regarding the formability and direct applications of UFG sheets, their potential should not be underestimated. Increasing the ARB process size to an industrial scale, improving the surface quality and mechanical properties, reducing the sheet thickness variations as well as investigating the formability of UFG sheets at lower strain rates or elevated temperatures, would collectively lead towards an overall improvement of the process efficiency and a greater potential for future industrial applications such as tailored blanking. The production of UFG materials would become commercialised and cost effective, and it would be possible to control the mechanical properties of materials by processing rather than by alloying. In the meantime, the big technical potential of ARB processed materials has also been recognised by the aluminium manufacturers. Future research interests are closely related to superplastic forming of aluminium alloys, accumulative roll bonding of magnesium alloys for lightweight structural components, as well as accumulative roll bonding of IF-steel. Thus, the innovation potential of the UFG materials for advanced applications in engineering is high, and the requirements for producing such materials are becoming more and more economically feasible. 7.2 Potential applications of ARB processed aluminium sheets The properties of UFG materials can be used beneficially in conventional sheet metal forming, superplastic forming, tailored blanks, tailored heat treated blanks as well as laminates which can be produced as a sandwich of different material combinations by ARB. Strain rate sensitivity can be used favourably for sheet metal forming, because it influences the deformation capability and higher elongation to failure could be obtained at lower strain rates. In order to improve the process efficiency, it is more advantageous to perform sheet metal forming at higher strain rates. In this case, performing the sheet metal forming of UFG materials at elevated temperatures may be a more desirable method for achieving better deformation and at the same time better process efficiency. Additionally, higher strain rates are typically associated with crash

CHAPTER 7 Outlook

175

events and consequently higher energy absorption of a given component. From this point of view, strain rate sensitive materials would also be advantageous, because they show an increase in strength with increasing strain rates. At this stage, the ductility of most UFG aluminium alloy sheets is still a limitation, which places another important challenge on ARB processed sheets. On the other hand, there have been some reports on low temperature superplasticity of the UFG metal sheets produced by the ARB process and friction stir welding. The superplastic property of UFG sheets can lead to a high potential for more specialised applications and/or hard to form components with complex geometries, due to excellent ductility which plays a crucial role during the metal forming process. A potential industrial application also concerns structural components and tailored blanks for the automotive industry. The advantage of the UFG sheets is that they would provide the adequate strength in certain parts of a tailored blank, while other parts could be formed at lower strain rates or elevated temperatures in order to achieve better formability and complex geometries. Another possibility is to use the UFG aluminium sheets for tailor heat treated blanks. In this case, the metal sheet would be heat treated locally in order to obtain e.g. better ductility, whereas the other regions would remain unaffected and retain high strength. In this way, one can tailor the mechanical properties of the sheets according to the requirements. The accumulative roll bonding process can also be used for manufacturing laminates or sandwiched structures of different materials. The aluminium-titanium laminates were successfully roll bonded in this work, although such sandwich structures require high capacity rolling mills due to their usually higher initial sheet thickness as well as their high hardness. A combination of different materials e.g. a pure aluminium shell with a hard core of a different aluminium alloy have shown to be processable by ARB in a current study and could provide good ductility, strength and corrosion resistance.

CHAPTER 7 Outlook

176

CHAPTER 8 Appendix

177

8 APPENDIX

8.1 Experimental Procedure Table 8-1: Technical data of two-high and four-high Carl Wezel rolling mills

Two-high rolling mill BW 200/130

Four-high rolling mill BW 200/130

Working roll Diameter ∅ (mm) 130 32 Width (mm) 200 115 Material steel Heat treated steel 32 CrMo Max. allowable temperature (°C) 250 250 Rotational speed (rpm) 20 (slow), 40 (fast) 80 (slow)Max. rolling force (kN) 120 120 Projected length of arc of contact Lp (mm) ~ 8 ~ 4

Supporting roll Diameter ∅ (mm) / 120 Width (mm) / 115 Material / Heat treated CrMo steel Rotational speed (rpm) / 20 (slow)

CHAPTER 8 Appendix

178

8.2 Materials Table 8-2: Physical properties of pure Al, Mg, Ti, Cu and Nb

Units Al Mg Ti Cu Nb Crystal structure fcc hcp hcp fcc bcc

Poisson ratio 0.35 0.35

[Ave99]

0.34 [Rey53] in

[Zwi74] 0.36 0.35

Density g/cm3 2.7 [Ask94]

1.74 [Ask94]

4.51 [Ask94]

8.93 [Ask94]

8.57 [Ask94]

Melting point °C 660.4

[Ask94] 650.0 [Ask94]

1668 [Ask94]

1084.9 [Ask94]

2468 [Ask94]

Thermal conductivity

(at 20°C) W/(m×K) 237 156

[Ave99]

23.9 [Pow61] in

[Zwi74] 401 52

Electrical resistivity

(at 20°C) nΩ×m 26.50 44.5

[Ave99]

420 [Fas39] in [Zwi74]

16.78 152

Young’s Modulus GPa 70 45 [Ave99] 111.8

[Zwi74] 110-128 104

Stacking fault energy J/m2 0.2

[Vol89] 0.125 [Vol89] / 0.07

[Vol89] /

CHAPTER 8 Appendix

179

8.3 Results a) b)

Figure 8-1: Characteristic force vs. displacement curves of a) AA1050 and b) AA6016 drawn at room temperature. a) AA6016N0, Ø80 mm b) AA6016N0, Ø90 mm c) AA6016N0, Ø110 mm

d) AA6016N4, Ø80 mm e) AA6016N4, Ø85 mm f) AA6016N4, Ø90 mm

g) AA6016N8, Ø80 mm h) AA6016N8, Ø90 mm

Figure 8-2: Deep drawn cups of aluminium alloy AA6016 of as-received and accumulative roll bonded state after 6 ARB cycles, drawn at room temperature.

CHAPTER 8 Appendix

180

a) b)

Figure 8-3: Characteristic force vs. displacement curves of a) AA1050 and b) AA6016 drawn at 180°C. a) AA1050N0, Ø80 mm b) AA1050N0, Ø90 mm c) AA1050N0, Ø110 mm


Figure 8-4: Deep drawn cups of commercial purity aluminium AA1050 of as-received and accumulative roll bonded state after 6 ARB cycles, drawn at 180°C.

CHAPTER 8 Appendix

181

a) AA6016N0, Ø80 mm b) AA6016N0, Ø90 mm c) AA6016N0, Ø110 mm


g) AA6016N8, Ø90 mm h) AA6016N8, Ø105 mm

Figure 8-5: Deep drawn cups of commercial purity aluminium AA6016 of as-received and accumulative roll bonded state after 4 and 8 ARB cycles, drawn at 180°C.

CHAPTER 8 Appendix

182

a) b)

Figure 8-6: Characteristic force vs. displacement curves of a) FSW AA1050 and b) FSW AA6016 drawn at 180°C. a) AA6016N0, Ø80 mm b) AA6016N0, Ø85 mm c) AA6016N0, Ø90 mm


g) AA6016N8, Ø80 mm h) AA6016N8, Ø85 mm i) AA6016N8, Ø90 mm

Figure 8-7: Deep drawn cups of FSW aluminium alloy AA6016 of as-received and accumulative roll bonded state after 4 and 8 ARB cycles, drawn at room temperature.

CHAPTER 9 References

183

9 REFERENCES

[Alu08] AluMatter, www.aluminium.matter.org.uk, 2008

[And98] S. J. Andersen, H. W. Zandbergen, J. Jansen, C. Traeholt, U. Tundal and O. Reiso: The crystal structure of the β˝ phase in Al-Mg-Si alloys, Acta Matererialia 46 (1998) 3283-3298

[Ara00] Aramis-Einführung, Gesellschaft für optische Messtechnik, GOM, 2000

[Ask94] D. R. Askeland, The Science and Engineering of Materials, Third edition, PWS Publishing Company, Boston, 1994

[Aud08] Audi AG, www.audi.de, 2008

[Aut08] THE Aluminium Automotive MANUAL (AAM), www.eaa.net/aam, 2008

[Ave99] M. M. Avedesian, H. Baker: ASM Speciality Handbook, Magnesium and Magnesium Alloys, ASM International, 1999

[Bar01] N. Barta-Schreiber: Modellierung und experimentelle Untersuchung des Korngrößeneinflußes auf das Mikrodehnungsverhalten ultrafeinkörnige Metalle, Diplomarbeit, Universität Erlangen-Nürnberg (2001)

[Bar04] C. Y. Barlow, P. Nielsen, N. Hansen: Multilayer roll bonded aluminium foil: processing, microstructure and flow stress, Acta Materialia 52 (2004) 3967-3972

[Bar06] A. Barcellona, G. Buffa, L. Fratini, D. Palmeri: On microstructural phenomena occurring in friction stir welding of aluminium alloys, Journal of Materials Processing Technology 177 (2006) 340

[Bei01] W. Beitz, K.-H. Grote, Dubbel, Taschenbuch für den Maschinenbau, Auflage 20, Springer Verlag 2001

[Ben99] S. Benavides, Y. Li, L. E. Murr, D. Brown, J. C. McClure: Low-temperature friction-stir welding of 2024 aluminum, Scripta Materialia 41 (1999) 809

[Bir05] Y. Birol: Pre-aging to improve bake hardening in a twin-roll cast Al–Mg–Si alloy, Materials Science and Engineering A 391 (2005) 175-180

[Blu04] W. Blum, Y. J. Li: Influence of grain boundaries on steady-state deformation resistance of ultrafine-grained Cu, Physica Status Solidi (a) 201, No. 13 (2004) 2915-2921

[Bow50] F. P. Bowden, D. Tabor: The Friction and Lubrication of Solids, Clarendon Press, Oxford, 1950

[Bow90] A. W. Bowen: Materials Science Technology 6 (1990) 1058

[Boy94] R. Boyer, G. Welsch, E. W. Collings: Titanium Alloys, eds. ASM International, Materials Park, OH, 1994

http://www.aluminium.matter.org.uk/

http://www.audi.de/

http://www.eaa.net/aam


184

[Bra92] E. A. Brandes, G. B. Brook, Smithells Metals Reference Book, 7th ed. (Butterworth-Heinemann, Oxford, United Kingdom, (1992)

[Büc02a] M. Büchler, N. Margadant, S. Siegmann, J. Ilavsky, G. Barbezat, J. Pisacka, R. Enzl: ITSC 2002 International Thermal Spray Conference, Essen (2002) 402-407

[Büc02b] M. Büchler: "Electrochemical Cell, Use of the Electrochemical Cell, and Method for Electrolytically Contacting and Electrochemically Influencing a Surface", Patent WO 02/25249 A2, Switzerland, 2002

[Cao03] W.Q. Cao, A. Godfrey, W. Liu, Q. Liu: EBSP study of the annealing behavior of aluminum deformed by equal channel angular processing, Materials Science and Engineering A 360 (2003) 420-425

[Cab07] M. Cabibbo, H. J. McQueen, E. Evangelista, S. Spigarelli, M. Di Paola, A. Falchero: Microstructure and mechanical property studies of AA6056 friction stir welded plate, Materials Science and Engineering A 460-461 (2007) 86-94

[Che06a] M. C. Chen, H. C. Hsieh, W. Wu: The evolution of microstructures and mechanical properties during accumulative roll bonding of Al/Mg composite, Journal of Alloys and Compounds 416 (2006) 169-172

[Che06b] Y. Chen, H. Liu, J. Feng: Friction stir welding characteristics of different heat-treated-state 2219 aluminum alloy plates, Materials Science and Engineering A 420 (2006) 21-25

[Cho06] S. G. Chowdhury, A. Dutta, B. Ravikumar, A. Kumar: Textural evolution during accumulative roll bonding of an Al–Li alloy, Materials Science and Engineering A 428 (2006) 351-357

[Chu04] M.-K. Chung, Y.-S. Choi, J.-G. Kim, Y.-M. Kim, J.-C. Lee: Effect of the number of ECAP pass time on the electrochemical properties of 1050 Al alloys, Materials Science and Engineering A 366 (2004) 282-291

[Cos05] A. L. M. Costa, A. C. C. Reis, L. Kestens, M. S. Andrade: Ultra grain refinement and hardening of IF-steel during accumulative roll-bonding, Materials Science and Engineering A 406 (2005) 279-285

[Dal04] F. H. Dalla Torre, E. V. Pereloma, C. H. J. Davies: Strain rate sensitivity and apparent activation volume measurements on equal channel angular extruded Cu processed by one to twelve passes, Scripta Matarialia 51 (2004) 367-371

[Dav93] J. R Davis, Aluminum and Aluminum Alloys, ASM International (1993) 76-78 [Daw91] C. J. Dawes: Friction stir butt welding, International Patent No. PCT/GB92/02203, 1991 [Daw96] C. J. Dawes, W. M. Thomas: Friction stir process welds aluminum alloys, Welding

Journal 75 (1996) 41-45

[Die88] G. E. Dieter: Mechanical Metallurgy, McGraw-Hill Book Company, London 1988

[Div08] S. Divinski, G. Wilde: Diffusion in ultrafine grained materials, Materials Science Forum, 584-586 (2008) 1012-1017

[Dut91] I. Dutta, S. M. Allen: A calorimetric study of precipitation in commercial aluminium alloy 6016, Journal of Materials Science Letters 10 (1991) 323-326


185

[Edw98] G. A. Edwards, K. Stiller, G. L. Dunlop, M. J. Couper: The precipitation sequence in Al-Mg-Si alloys, Acta Materialia 46 (1998) 3893-3904

[Fas39] J. D. Fast, Über die Darstellung der reinen Metalle der Titangruppe durch thermische Zersetzung ihrer Jodide, Anorganische Chem. 241 (1939) 42-56

[Fon04] R. W. Fonda, J. F. Bingert, K. J. Colligan: Development of grain structure during friction stir welding, Scripta Materialia 51 (2004) 243

[Fuj06] H. Fujii, R. Ueji, Y. Takada, H. Kitahara, N. Tsuji, K. Nakata, K. Nogi: Friction Stir Welding of Ultrafine Grained Interstitial Free Steels, Materials Transactions 47 No. 1 (2006) 239-242

[Fur01] M. Furukawa, Z. Horita, T. G. Langdon: Developing Ultrafine Grain Sizes Using Severe Plastic Deformation, Advanced Engineering Materials 3, No. 3 (2001) 121-125

[Geb04] S. Gebhard: Erzeugung von ultrafeinkörnigen Mikrostrukturen durch accumulative roll bonding (ARB) an ausgewählten Aluminiumlegierungen, Allgemeine Werkstoffeingenschaften, Friedrich-Alexander Universität Erlangen-Nürnberg, Studienarbeit (2004)

[Gei01] M. Geiger, T. Hennige, Skript zur Vorlesung Umformtechnik 1, Lehrstuhl für Fertigungstechnologie, Friedrich-Alexander Universität Erlangen-Nürnberg, 2001

[Gen05] C. Genevois, A. Deschamps, A. Denquin, B. Doisneau-Cottignies: Quantitative investigation of precipitation and mechanical behaviour for AA2024 friction stir welds, Acta Materialia 53 (2005) 2447-2458

[Gle89] H. Gleiter: Nanocrystalline materials, Progress in Materials Science 33 (1989) 223-315

[Gol75] F. Gologranc: Beitrag zur Ermittlung von Fließkurven im kontinuierlichen hydraulischen Tiefungsversuch (Evaluation of the flow stress curve with the continuous hydraulic bulge test), Dissertation, Institute for Metal Forming Technology, University of Stuttgart, Germany, 1975

[Got98] G. Gottstein: Physikalische Grundlagen der Meterialkunde, Springer Verlag Berlin Heidelberg 1998

[Gut04] G. Gutscher, H.-C. Wu, G. Ngaile, T. Altan: Determination of flow stress for sheet metal forming using the viscous pressure bulge (VPB) test, Journal of Materials Processing Technology 146 (2004) 1-7

[Hal51] E. O. Hall: Proceedings of the Physical Society of London B64 (1951) 747-753

[Han99] N. Hansen, D. J. Jensen: Development of microstructure in FCC metals during cold work, Phil. Trans. R. Soc. Lond. A 357 (1999) 1447-1469

[Han08] A. Hannon, P. Tiernan: A review of planar biaxial tensile test systems for sheet metal, Journal of Materials Processing Technology 198 (2008) 1-13

[Har67] E. W. Hart: Theory of the tensile test, Acta Metallurgica 15 (1967) 351-355

[Hei99] P. Hein, F. Vollertsen: Hydroforming of sheet metal pairs, Journal of Materials Processing Technology 87 (1999) 154-164


186

[Hei02] B. Heinz, B. Skrotzki: Characterization of a friction-stir-welded aluminum alloy 6013, Metallurgical and Materials Transactions B 33 (6) (2002) 489-498

[Hei08] Heise Zeitschriften Verlag, www.heise.de/autos/Leichter-ist-schwer-Leichtbau-als-Weg-zur-CO2-Reduzierung, 2008

[Hof97] A. Hofmann, F. Vollertsen: Deep drawing of locally heat treated blanks, Laser-assisted net shape Engineering 2, Conference Proceedings, LANE97, Meisenbach, Bamberg, (1997) 719-732

[Hom06] P. Homola, M. Slámová, M. Karlík, J.Cizek, I. Procházka: Effect of the Temperature of Accumulative Roll Bonding on the Microstructure and Properties of Twin-Roll Cast AA8006 Alloy, Materials Science Forum Vols. 503-504 (2006) 281-286

[Hor87] J. Horvath, R. Birringer, H. Gleiter: Diffusion in nanocrystalline material, Solid State Communications, 62 (1987) 319-322

[Hor96] Z. Horita, D. J. Smith, M. Furukawa, M. Nemoto, R. Z. Valiev, T. G. Langdon: Evolution of grain boundary structure in submicrometer-grained Al-Mg alloy, Materials Research Society 11 (1996) 1880-1890

[Hor05] Z. Horita: Proc. of the 3rd Int. Conf. on Nanomaterials by Severe Plastic Deformation, Trans Tech Publications Ltd, 2005

[Höp04] H. W. Höppel, J. May, M. Göken: Enhanced strength and ductility in UFG aluminium produced by ARB, Advanced Engineering Materials 6 (2004) 781-784

[Höp05] H. W. Höppel, J. May, P. Eisenlohr, M. Göken: Strain rate sensitivity of ultrafine-grained materials, Zeitschrift für Metallkunde 96 (2005) 6

[Hua01] J. Y. Huang, Y. T. Zhu, H. Jiang, T. C. Lowe: Microstructure and dislocation configurations in nanostructured Cu processed by repetitive corrugation and straightening, Acta Materialia 49 (2001) 1497-1505

[Hua03] X. Huang, N. Tsuji, N. Hansen, Y. Minamino: Microstructural evolution during ARB of commercial purity Al, Materials Science and Engineering A 340 (2003) 265-271

[Hug97] D. A. Hughes, N. Hansen: High angle boundaries formed by grain subdivision mechanism, Acta Materialia 45 (1997) 3871-3886

[Hug03] D. A. Hughes, N. Hansen, D. J. Bammann: Geometrically necessary boundaries, incidental dislocation boundaries and geometrically necessary dislocations, Acta Materialia 48 (2003) 147-153

[Hum95] F. J. Humphreys, M. Hatherly: Recrystallization and related annealing aluminium phenomena, Oxford: Pergamon Press, 1995

[Iss95] L. Issler, H. Ruoß, P. Häfele: Festigkeitslehre – Grundlagen, Springer-Verlag Berlin Heidelberg, 1995, 143

[Jat00] K. V. Jata, K. K. Sankaran, J. J. Ruschau: Friction-stir welding effects on microstructure and fatigue of aluminum alloy 7050-T7451, Metallurgical and Materials Transactions A 31, No. 9 (2000) 2181-2192


187

[Jin04] H. Jin, D. J. Lloyd: Effect of a duplex grain size on the tensile ductility of an ultra-fine grained Al–Mg alloy, AA5754, produced by asymmetric rolling and annealing, Scripta Materialia 50 (2004) 1319-1323

[Kam02] C. Kammer: Aluminium-Taschenbuch Band 1, Grundlagen und Werkstoffe, Aluminium-Verlag Marketing und Kommunikation GmbH, Düsseldorf, 2002

[Kam06] N. Kamikawa, N. Tsuji, X. Huang, N. Hansen: Quantification of annealed microstructures in ARB processed aluminum, Acta Materialia 54 (2006) 3055-3066

[Kam07] N. Kamikawa, T. Sakai, N. Tsuji: Effect of redundant shear strain on microstructure and texture evolution during accumulative roll-bonding in ultralow carbon IF steel, Acta Materialia 55 (2007) 5873-5888

[Kim05] H.-W. Kim, S.-B. Kang, N. Tsuji, Y. Minamino: Elongation increase in ultra-fine grained Al–Fe–Si alloy sheets, Acta Materialia 53 (2005) 1737-1749

[Kim06] Y. S. Kim, S.H. Kang, D. H. Shin: Effect of rolling direction on the microstructure and mechanical properties of ARB processed commercially pure 1050 aluminium alloy, Materials Science Forum 503-504 (2006) 681-686

[Kle06] M. Kleinera, S. Chatti, A. Klaus: Metal forming techniques for lightweight construction, Journal of Materials Processing Technology 177 (2006) 2-7

[Klö06] C. C. Klösters: Einfluss der Oberflächenkonditionierung auf die mechanischen Eigenschaften von ultrafeinkörnigen Blechen hergestellt mit kumulativem Walzen, Lehrstuhl für Allgemeine Werkstoffeigenschaften, Friedrich-Alexander Universität Erlangen-Nürnberg, Diplomarbeit, 2006

[Kei05] M. Keigler, H. Bauer, D. Harrison, A.K.M. De Silva: Enhancing the formability of aluminium components via temperature controlled hydroforming, Journal of Materials Processing Technology 167 (2005) 363-370

[Koc99] C. Koch: Bulk behavior of nanostructured materials. Editors R. W. Siegel, E. Hu, M. C. Roco: Nanostructure science and technology, a worldwide study. Final Report by WTEC Panel 1999, 93

[Kog02] K. Koglin, H. Waltl, T. Possehn: Neuere Entwicklungen in der Blechumformung, Frankfurt: MAT INFO Werkstoff-Informationsgesellschaft mbH (2002) 21-34

[Kra04] G. Krallics, J. G. Lenard: An examination of the accumulative roll-bonding process, Journal of Materials Processing Technology 152 (2004) 154-161

[Lan74] K. Lange: Lehrbuch der Umformtechnik, Band 2: Massivumformung, Springer Verlag, Berlin 1974

[Lap07a] R. Lapovok, P. W. J. Mc Kenzie, P. F. Thomson, S. L. Semiatin: The influence of post-ECAP annealing on the properties of ultrafine-grained 5005 aluminium alloy sheet, International Journal of Materials Research 98 (2007) 325-331

[Lap07b] R. Lapovok, P. McKenzie, P. Thomson, S. Semiatin: Processing and properties of ultrafine-grain aluminium alloy 5005 sheet, Journal of Materials Science 42 (2007) 1649-1659


188

[Lau04] T. Laumann: Umformverhalten und Fügbarkeit von ultrafeinkörnigen Werkstoffen hergestellt mit dem ARB-Verfahren, Friedrich-Alexander-Universität Erlangen-Nürnberg, Institut für Werkstoffwissenschaften Lehrstuhl 1: Allgemeine Werkstoffeigenschaften, Diplomarbeit 2004

[Le04] H. R. Le, M. P. F. Sutcliffe, P. Z. Wang, G. T. Burstein: Surface oxide fracture in cold aluminium rolling, Acta Materialia 52 (2004) 911-920

[Lee02a] S. H. Lee, Y. Saito, N. Tsuji, H. Utsunomiya, T. Sakai: Role of shear strain in ultragrain refinement by ARB process, Scripta Materialia 46 (2002) 281-285

[Lee02b] S. H. Lee, Y. Saito, T. Sakai, H. Utsunomiya: Microstructures and mechanical properties of 6061 aluminium alloy processed by ARB, Materials Science and Engineering A 325 (2002) 228-235

[Lee02c] J. C. Lee, H. K. Seok, J. Y. Suh: Microstructural evolutions of the Al strip prepared by cold rolling and continuous equal channel angular pressing, Acta Materialia 50 (2002) 4005-4019

[Lee04] Y. B. Lee, D. H. Shin, K.-T. Park, W. J. Nam: Effect of annealing temperature on microstructures and mechanical properties of a 5083 Al alloy deformed at cryogenic temperature, Scripta Materialia 51 (2004) 355-359

[Li04] Y. J. Li, X. H. Zeng, W. Blum: Transition from strengthening to softening by grain boundaries in ultrafine-grained Cu, Acta Materialia 52 (2004) 5009-5018

[Li06a] B. L. Li, N. Tsuji, N. Kamikawa: Microstructure homogeneity in various metallic materials heavily deformed by accumulative roll-bonding, Materials Science and Engineering A 423 (2006) 331-342

[Li06b] B. L. Li, N. Shigeiri, N. Tsuji, Y. Minamino: Microstructural evolution in pure copper severely deformed by the ARB process, Materials Science Forum 503-504 (2006) 615-620

[Liu97] G. Liu, L. E. Murr, C. S. Niou, J. C. McClure, F. R. Vega: microstructural aspects of the friction-stir welding of 6061-T6 aluminum, Scripta Materialia 37, No. 3 (1997) 355-361

[Low00] T. C. Lowe, R. Z. Valiev: Producing nanoscale microstructures through severe plastic deformation, Journal of the Minerals, Metals and Materials Society 52 (2000) 27-28

[Ma02] Z. Y. Ma, R. S. Mishra, M. W. Mahoney: Acta Mater. 50 (2002) 4419

[Ma05] Z. Y. Ma, R. S. Mishra: Development of ultrafine-grained microstructure and low temperature (0.48×Tm) superplasticity in friction stir processed Al-Mg-Zr, Scripta Materialia 53 (2005) 75-80

[Mah98] M. W. Mahoney, C. G. Rhodes, J. G. Flintoff, R. A. Spurling, W. H. Bingel: Properties of friction-stir-welded 7075 T651 aluminum, Metallurcial and Materials Transactions A 29, No. 7 (1998) 1955-1964

[May04] J. May: Optimierung der Verfahrensparameter für die Erzeugung ultrafeinkörniger metallischer Werkstoffe mittels Severe Plastic Deformation (SPD), Diplomarbeit, Universität Erlangen-Nürnberg (2004)


189

[May05] J. May, H. W. Höppel, M. Göken: Strain rate sensitivity of ultrafine-grained aluminium processed by severe plastic deformation, Scripta Materialia 53, Issue 2 (2005) 189-194

[Mer01] M. Merklein: Laserstrahlumformen von Aluminiumwerkstoffen-Beeinflussung der Mikrostruktur und der mechanischen Kennwerte, Meisenbach, Bamberg, 2001

[Mer02] M. Merklein, M. Geiger: New materials and production technologies for innovative lightweight constructions, Journal of Materials Processing Technology 125-126 (2002) 532-536

[Mer06] M. Merklein: Charakterisierung von Blechwerkstoffen für den Leichtbau, Maisenbach Verlag 2006, Bamberg

[Mey06] M. A. Meyers, A. Mishra, D. J. Benson: Mechanical properties of nanocrystalline materials, Progress in Materials Science 51 (2006) 427-556

[Mil62] D. R. Milner, G. W. Rowe: Fundamentals of solid-phase welding, Metallurgical Reviews 7 (1962) 433-480

[Mis05] R. S. Mishra, Z. Y. Ma: Friction stir welding and processing, Materials Science and Engineering R 50 (2005) 1-78

[Mol07] X. Molodova, G. Gottstein, M. Winning, R.J. Hellmig: Thermal stability of ECAP processed pure copper, Materials Science and Engineering A 460-461 (2007) 204-213

[Mor02] D. G. Morris, M. A. Muñoz-Morris: Microstructure of severely deformed Al–3Mg and its evolution during annealing. Non-uniform coarsening, Acta Materialia 50 (2002) 4047-4060

[Mts03] MTS, Nano Instruments INC, Product Specifications Nanoindenter XP, Oak Ridge, TN, USA, 2003

[Nov03] S. Novotny, M. Geiger: Process design for hydroforming of lightweight metal sheets at elevated temperatures, Journal of Materials Processing Technology 138 (2003) 594-599

[Neu04] S. Neumeier: Friedrich-Alexander-Universität Erlangen-Nürnberg, Institut für Werkstoffwissenschaften Lehrstuhl 1: Allgemeine Werkstoffeigenschaften, Diplomarbeit 2004

[Par01] K.-T. Park, H.-J. Kwon, W.-J. Kim, Y.-S. Kim: Microstructural characteristics and

thermal stability of UFG 6061 Al alloy fabricated by ARB process, Materials Science and Engineering A 316 (2001) 124-152

[Per04] M. T. Pérez-Prado, J. A. del Valle, O. A. Ruano: Grain refinement of Mg–Al–Zn alloys via accumulative roll bonding, Scripta Materialia 51 (2004) 1093-1097

[Per99] A. Perovic, D. D. Perovic, G. C. Weatherly, D. J. Lloyd: Precipitation in aluminium alloys AA6111 and AA6016, Scripta Materialia 41 (1999) 703-708

[Pet53] N. J. Petch: The Journal of the Iron and Steel Institute 174 (1953) 25-28

[Pla08] Plansee SE, www.plansee.at, Werkstoffbroschüre Niob, 2008

[Pol98] I. J. Polmear: Light Alloys, Metallurgy of the light metals, Second Edition, Edward Arnold Publishing, London 1998


190

[Pow61] R. W. Powell, R.P. Tye, The thermal and electrical conductivity of Ti and its alloys, Journal of the less Common Metals 3 (1961) 226-233

[Ree95] D. W. A. Rees: Plastic flow in the elliptical bulge test, International Journal of Mechanical Sciences 37, No. 4 (1995) 337-389

[Rev00] R. W. Revie: "Uhlig's Corrosion Handbook", Electrochemical Society Series, John Wiley & Sons, Inc., New York, 2. Auflage, 2000

[Rey53] M. B. Reynolds, The determination of the elastic constants of metals by the ultrasonic pulse technique, Trans. ASM 45 (1953) 839-854

[Rho97] C. G. Rhodes, M. W. Mahoney, W. H. Bingel, R. A. Spurling, C. C. Bampton: Effects of friction stir welding on microstructure of 7075 aluminium, Scripta Matererialia 36, No. 1 (1997) 69-75

[Ros95] G. I. Rosen, D. Juul Jensen, D. A. Hughes, N. Hansen: Microstructure and local crystallography of cold rolled aluminium, Acta Metallurgica Materialia 43 (1995) 2563-2579

[Rös03] J. Rösler, H. Harders, M. Bäker, Mechanisches Verhalten der Werkstoffe, B.G. Taubner Verlag, Stuttgart 2003

[Sag98] A. Sagel, H. Sieber, H.-J. Fecht, J. H. Perepezko: Acta Mater. 46 (1998) 4233

[Sai98] Y. Saito, N. Tsuji, H. Utsunomiya, T. Sakai, R. G. Hong: Ultra-fine grained bulk aluminium produced by accumulative roll-bonding (ARB) process, Scripta Materialia 39, No.9 (1998) 1221-1227

[Sai99] Y. Saito, H. Utsunomiya, N. Tsuji, T. Sakai: Novel ultra-high straining process for bulk materials development of the accumulative roll-bonding (ARB) process, Acta Materialia 47 (1999) 579-583

[Sai00] Y. Saito, H. Utsunomiya, H. Suzuki, T. Sakai: Improvement in the r-value of aluminium strip by continuous shear deformation process, Scripta Materialia 42, (2000) 1139-1144

[Sat99a] Y. S. Sato, H. Kokawa, M. Enmoto, S. Jogan, T. Hashimoto: Precipitation sequence in friction stir weld of 6063 aluminum during aging, Metallurgical and Material Transactions A 30 (1999) 3125-3130

[Sat99b] Y. S. Sato, H. Kokawa, M. Enmoto, S. Jogan: Metallurgical and Material Transactions A 30, No. 12 (1999) 2429

[Sat04] Y. S. Sato, Y. Kurihara, S. H. C. Park, H. Kokawa, N. Tsuji: Friction stir welding of ultrafine grained Al alloy 1100 produced by accumulative roll-bonding, Scripta Materialia 50 (2004) 57-60

[Sat06] Y. S. Sato, M. Urata, Y. Kurihara, S. H. C. Park, H. Kokawa, K. Ikeda, N. Tsuji: Microstructural evolution during friction stir welding of ultrafine-grained aluminium alloys, Materials Science Forum 503-504 (2006) 169-174

[Sch04] R. Schiffmann, J. Haug, J. Banhart: Evolution of Precipitates during Age-hardening of AW 6016 Alloy, Proceedings of the 9th International Conference on Aluminium Alloys (2004)


191

[Sch03] R. Schwaiger, B. Moser, M. Dao, N. Chollacoop, S. Suresh: Some critical experiments on the strain-rate sensitivity of nanocrystalline nickel, Acta Matererialia 51 (2003) 5159-5172

[Seg77] V. M. Segal: Patent der UdSSR, Nr. 575892, 1977

[Seg81] V. M. Segal: V. I. Reznikov, A. E. Drobyshevskiy, V. I. Kopylov: Russian Metall. 1 (1981) 115

[Seg95] V. M. Segal: Materials Science and Engineering A 197 (1995) 157-164

[Skr07] W. Skrotzki, I. Hünsche, J. Hüttenrauch, C. G. Oertel, H. G. Brokmeier, H. W. Höppel, I. Topic: Texture and Mechanical Anisotropy of Ultrafine-Grained Aluminum Alloy AA6016 Produced by Accumulative Roll Bonding, Texture, stress and microstructure (2008) 1-8

[Sla06a] M. Slamova, P. Homola, P. Slama, M. Karlik, M. Cieslar, Y. Ohara, N. Tsuji: Accumulative roll bonding of AA8006, AA8011 and AA5754 sheets, Materials Science Forum 519-521 (2006) 1227-1232

[Sla06b] M. Slámová, P. Homola, P. Sláma, J. Cizek, I. Procházka, M. Cieslar: Microstructure, Texture and Property Changes of High Purity Aluminium during Accumulative Roll Bonding and Conventional Rolling, Materials Science Forum 503-504 (2006) 711-716

[Sla07] M. Slamova, P. Homola, M. Karlik: Thermal stability of twin-roll cast Al–Fe–Mn–Si sheets accumulative roll bonded at different temperatures, Materials Science and Engineering A 462 (2007) 106-110

[Smi86] N. A. Smirnova, V. I. Levit, V. I. Pilyugin, R. I. Kuznetsov, L. S. Davidova, V. A. Sazonova: Phys. Met. Metall. 61 6 (1986) 127

[Su03] J. Q. Su, T. W. Nelson, R. S. Mishra, M. W. Mahoney: Microstructural investigation of friction stir welded 7050-T651 aluminium, Acta Materialia 51 (2003) 713-729

[Sua05] J. Q. Sua, T. W. Nelson, C. J. Sterling: Microstructure evolution during FSW/FSP of high strength aluminum alloys, Materials Science and Engineering A 405 (2005) 277-286

[Sut01] T. Suter, H. Böhni: Microelectrodes for corrosion studies in Microsystems,

Electrochimica Acta 47 (2001) 191-199

[Tak99] Y. Takayama, M. Yamaguchi, T. Tozaea, H. Kato, H. Watanabe, T. Izawa: Proc. of the 4th Int. Conf. on Recrystallization and Related Phenomena (Rex’99), The Jap. Inst. Metals, Sendai, Japan (1999) 321

[Tak07] N. Takata, S.-H. Lee, C.-Y. Lim, S.-S. Kim, N. Tsuji: Nanostructured bulk copper fabricated by accumulative roll bonding, Jour. Nanosci. Nanotechnol. 7 (2007) 3985-3989

[Ter07] D. Terada, S. Inoue, N. Tsuji: Microstructure and mechanical properties of commercial purity titanium severely deformed by ARB process, Journal of Materials Science 42 (2007) 1673-1681

[Tho91] W. M. Thomas, E. D. Nicholas, J. C. Needham, M. G. Murch, P. Templesmith, C. J. Dawes: Friction stir butt welding, G. B. Patent: Application No. 9125978.8, December 1991


192

[Tho97] W. M. Thomas, E. D. Nicholas: Friction stir welding for the transportation industries, Materials and Design 18, Issues 4-6 (1997) 269-273

[Thy08] ThyssenKrupp Steel AG, Thyssen Kripp Tailored Blanks, www.tailored-blanks.com, 2008

[Top07] I. Topic, H. W. Höppel, M. Göken: Deformation behaviour, microstructure and processing of ARB aluminium alloy AA6016, International Journal of Materials Research 98 (2007) 320-324

[Top08a] I. Topic, H. W. Höppel, M. Göken: Deformation Behaviour of Accumulative Roll Bonded and Friction Stir Welded Aluminium Alloys, Materials Science Forum Vols. 584-586 (2008) 833-839

[Top08b] I. Topic, H. W. Höppel, M. Göken: Steigerung des Leichtbaupotenzials flächiger Bauteile mittels ultrafeinkörniger Mikrostrukturen produziert in Walzprozessen und Fügeverfahren, Sonderforschungsbereich 396, Berichtsband zum Abschlusskolloquium „Robuste, verkürzte Prozessketten für flächige Leichtbauteile“, (2008) 283-313

[Tsu99a] N. Tsuji, Y. Saito, H. Utsunomiya, S. Tanigawa: Ultra-fine grained bulk steel produced by accumulative roll-bonding process, Scripta Materialia 40, No. 7 (1999) 795-800

[Tsu99b] N. Tsuji, K. Shiotsuki, Y. Saito: Superplasticity of ultrafine-grained Al-Mg alloy produced by accumulative roll bonding, Materials Transaction, JIM 40 (1999) 765-771

[Tsu99c] N. Tsuji, K. Shiotsuki, Y. Saito: Low temperature superplasticity of ultrafine-grained 5083 aluminium alloy produced by accumulative roll bonding, Materials Science Forum 304-306 (1999) 73-78

[Tsu02a] N. Tsuji, R. Ueji, Y. Minamino: Nanoscale crystallographic analysis of UFG IF steel fabricated by ARB process, Scripta Materialia 47 (2002) 69

[Tsu02b] N. Tsuji, Y. Ito, Y. Saito, Y. Minamino: Strength and ductility of ultrafine grained aluminum and iron produced by ARB and annealing, Scripta Materialia 47 (2002) 893-899

[Tsu03a] N. Tsuji, Y. Saito, S. H. Lee, Y. Minamino: ARB and other new techniques to produce bulk UFG materials, Advanced Engineering Materials 5 (2003) 338-344

[Tsu03b] N. Tsuji, T. Toyoda, Y. Minamino, Y. Koizumi, T. Yamane, M. Komatsu, M. Kiritani: Microstructural change of ultrafine-grained aluminum during highspeed plastic deformation, Materials Science and Engineering A350 (2003) 108-116

[Val97] R. Z. Valiev: Structure and mechanical properties of ultrafine-grained metals, Materials Science and Engineering A 234-236 (1997) 59-66

[Val00] R. Z. Valiev, R. K. Islamgaliev, I. V. Alexandrov: Bulk nanostructured materials from severe plastic deformation, Progress in Materials Science 45 (2000) 103-189

[Val02] R. Z. Valiev, I. V. Alexandrov, Y. T. Zhu, T. C. Lowe: Paradox of strength and ductility in metals processed by severe plastic deformation, Journal of Materials Research 17 (2002) 5-8

[Val04] R. Z. Valiev: Nanostructuring of metals by severe plastic deformation for advanced properties, Nature Materials 3 (2004) 511-515


193

[Val06] R. Z. Valiev, Y. Estrin, Z. Horita, T. G. Langdon, M. J. Zehetbauer, Y. T. Zhu: Producing Bulk Ultrafine-Grained Materials by Severe Plastic Deformation, Journal of the Minerals, Metals and Materials Society 58, No. 4 (2006) 33-39

[Vev08] A. Vevecka-Priftaj, A. Böhner, J. May,H. W. Höppel, M. Göken: Strain rate sensitivity of ultrafine grained aluminium alloy AA6061, submitted in Materials Science Forum (2008)

[Vol89] F. Vollertsen, S. Vogler, Werkstoffeigenschaften und Mikrostruktur, Hanser-Verlag, München, 1989

[Wan02] Y. Wang, M. Chen, F. Zhou, E. Ma: Nature 419 (2002) 912

[Wei07] W. Wei, K.-X. Wei, Q.-B. Du: Corrosion and tensile behaviors of UFG Al-Mn alloy produced by ARB, Materials Science and Engineering A 454-455 (2007) 536-541

[Wie08] Wieland-Werke AG, www.wieland.de, 2008

[Wit03] D. Witkins, Z. Lee, R. Rodrigues, S. Nutt, E. Lavernia: Scripta Mater 49 (2003) 303

[Wlo07] J. P. Wloka: Korrosionsuntersuchungen an scandiumhaltigen AlZnMgCu-Legierungen unter besonderer Berücksichtigung des Einflusses intermetallischer Phasen, Dissertation, Universität Erlangen-Nürnberg, 2007

[Wu98] H.-Y. Wu, S. Lee, J.-Y. Wang: Solid state bonding of iron-based alloys, steel-brass and aluminium alloys, Journal of Materials Processing Technology 75 (1998) 173-179

[Xin01] Z. P. Xing, S. B. Kang, H. W. Kim: Softening behaviour of 8011 alloy produced by accumulative roll bonding process, Scripta Materialia 45 (2001) 597-604

[Yam01] A. Yamashita, Z. Horita, T. G. Langdon: Improving the mechanical properties of magnesium and a magnesium alloy through severe plastic deformation, Materials Science and Engineering A 300 (2001) 142-147

[Yan04] H. Yan, J. G. Lenard: A study of warm and cold roll-bonding of an aluminium alloy, Materials Science and Engineering A 385 (2004) 419-428

[Zei95] W. Zeiger, M. Schneider, D. Scharnweber, H. Worth: Corrosion behaviour of a nanocrystalline FeAl8 alloy, Nano Structured Materials 6 (1995) 1013-1016

[Zei03] W. Zeichner: Die fünfte 5er-Generation - Eine Klasse für sich, Sonderausgabe von ATZ und MTZ 8 (2003) 12-19

[Zha06] Y.-H. Zhao, X.-Z. Liao, S. Cheng, E. Ma, Y. T. Zhu: Simultaneously Increasing the Ductility and Strength of Nanostructured Alloys, Advanced Materials 18 (2006) 2280-2283

[Zhu01] Y. T. Zhu, H. Jiang, J. Huang, T. C. Lowe, A new route to bulk nanostructured metals, Metallurgical and Materials Transactions A 32 (2001) 1559-1562

[Zwi74] U. Zwicker, Titan und Titanlegierungen, Band 21, Springer-Verlag Berlin, 1974

http://www.wieland.de/


194

ACKNOWLEDGEMENTS I would like to express my gratitude to the following people who have helped me during the course of this project. My special thanks go to:

• Prof. Dr. M. Göken, who supported and supervised this project, and who has always given many new and helpful suggestions and ideas

• Dr. H. W. Höppel, for offering me this project and proposing the wonderful topic,

which I have enjoyed thoroughly during my whole PhD career, as well for his excellent supervision concerning both scientific and practical areas, extremely helpful discussions and support with various industrial partners

• The German Research Foundation (Deutsche Forschungsgemeinschaft - DGM) for the

financial support within the framework of the collaborative research project SFB 396, “Robuste, verkürzte Prozessketten für flächige Leichtbauteile”

• Prof. Dr. M. Merklein, Detlev Staud and Uwe Vogt at the Institute of Manufacturing

Technology at the Friedrich-Alexander University Erlangen-Nürnberg for providing the necessary equipment for bulge tests and deep drawing tests, for their assistance, support and great collaboration

• Prof. Dr. W. Skrotzki and J. Hüttenrauch at the Technical University of Dresden for

conducting the texture measurements on ARB aluminium sheets, for helpful discussions and rewarding collaboration

• The fantastic Novelis Switzerland SA team including Corrado Bassi, Cyrille Bezençon,

Jürgen Timm, Dr. Rainer Kossak, Frank Schillinger and Jean-François Despois, for supplying various aluminium alloys, their enthusiastic interest in my project and the fruitful present and future project collaborations

• The European Aeronautic Defence and Space Company (EADS) and particularly Dr. J.

Vlcek, Marco Hüller and Tommy Brunzel for the interesting project collaboration on titanium and aluminium laminates as well as for performing friction stir welding of aluminium sheets

• Dr. J. P. Wloka for his help with corrosion measurements

• Prof. Dr. Sockel and Prof. Dr. H. Mughrabi for always showing interest and enthusiasm

in my work

• My friends and colleagues in the office Siphilisiwe Ndlovu, Ralf Nützel and Jens Schaufler, for daily news, exciting discussions, good food and friendship

• My friend, Björn Backes for proof reading parts of my thesis

• Michaela Prell, Benedikt Scharfe, Anja Neumann, Tina Hausöl, Christine Klösters and

Till Lauman for their assistance and a great and productive atmosphere at work

• Werner Langner, Wolfgang Maier, Günther Freimann and Richard Kosmala for their laboratory assistance

• All my colleagues and friends!

• My parents for their love and support, and my understanding boyfriend Javier who was

patiently waiting for his favourite meals for a couple of months without complaining.

Documents

Ultrafine-grained Metal Sheets produced using the Accumulative Roll Bonding Process for Light