Magnesium: Proceedings of the 6th International Conference Magnesium Alloys and Their Applications

Citation preview

Magnesium Proceedings of the 6th International Conference Magnesium Alloys and Their Applications

Edited by K.U. Kainer

Vakat

Magnesium Proceedings of the 6th International Conference Magnesium Alloys and Their Applications Edited by K.U. Kainer

Further titles of interest: K.U. Kainer (Ed.) Magnesium Alloys and Their Applications (2003) ISBN 3-527-30570-X

gggg

Magnesium Proceedings of the 6th International Conference Magnesium Alloys and Their Applications

Edited by K.U. Kainer

Editor: Prof. Dr. K. U. Kainer GKSS-Forschungszentrum Institut für Werkstoffforschung Max-Planck-Straße 21502 Geesthacht Germany

This book was carefully produced. Nevertheless, editor, authors, and publisher do not warrant the information contained therein to be free of erros. Readers are advised to keep in mind that statements, data, illustrations, procedural details or other items may inadvertently be inaccurate.

Library of Congress Card No.: Applied for British Library Cataloguing-in-Publication Data: A catalogue record for this book is available from the British Library Bibliographic information published by Die Deutsche Bibliothek Die Deutsche Bibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data is abailable in the Internet at . ISBN 3-527-30975-6

© 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Printed on acid-free paper All rights reserved (including those of translation in other languages). No part of this book may be reproduced in any form – by photoprinting, microfilm, or any other means – nor transmitted or translated into machine language without written permission from the publishers. Registered names, trademarks, etc. used in this book, even when not specifically marked as such, are not to be considered unprotected by law. Composition, Printing and Bookbinding: Druckhaus “Thomas Müntzer” GmbH, Bad Langensalza Printed in the Federal Republic of Germany

Preface Since its launch in 1986, the International Conference on Magnesium Alloys and Their Applications has become the most reputable and excepted event as forum for current development in magnesium technology. The increase in interest on magnesium alloys for use in transportation industry and the dramatic changes in the magnesium industry in the last 5 years show consequences for the market development and influence the R&D activities. The 6th International Conference and Exhibition on Magnesium Alloys and Their Applications reflects this evolution process. The aim of the conference is to show the correlation of processing, microstructure and properties for magnesium materials. The complexity of interactions must be understood to use the full potential of magnesium materials for a particular application. The conference will cover the important aspects of magnesium research and technology including: Alloy Development Microstructure Evolution Casting Wrought Alloys Post Processing Recycling Simulation, Modelling Corrosion Surface Treatment Magnesium Matrix Composites Application Research Programmes The conference returns to Wolfsburg, the home of Volkswagen Group, where it saw the biggest success. The Call for Paper has attracted more than 230 papers from 28 countries. More than 190 contributions are published in the proceeding volume. They give an excellent overview on newest developments.

K. U. Kainer September 2003

Vakat

VII

Inhaltsverzeichnis Microstructural Design for Enhanced Elevated Temperature Properties in Sand-castable Magnesium Alloys C.J. Bettles and M.A. Gibson, CRC for Cast Metals Manufacturing (CAST), CSIRO Manufacturing & Infrastructure Technology, Private Bag 33, Clayton South MDC, Clayton, VIC. 3169, Australia .......................................................

1

Thermodynamic Database for Mg-alloys: Progress and Application to Mg-Al-Ca-Si Alloy Development Rainer Schmid-Fetzer and Joachim Gröbner, University of Clausthal, Institute of Metallurgy, Clausthal-Zellerfeld, Germany .................................................. 12 Development of the New Creep Resistant Alloy AS31 Vadim V. Agalakov, JSC “AVISMA” , Berezniki, Russia ................................................ 18 Phase Formation, Precipitation and Strengthening Mechanisms in Mg-Zn-Sn Alloys S. Cohen, G.R. Goren-Muginstein, S. Avraham, and M. Bamberger, Department of Material Eng., Technion IIT, Haifa 3200 ISRAEL; G. Dehm, Max-Planck-Institut für Metallforschung, Stuttgart, Germany ....................... 25 Die Casting for High Performance – Focus on Alloy Development Dr. Per Bakke, Dr. Håkon Westengen, Hydro Aluminium a.s., Magnesium Competence Centre, Porsgrunn, Norway ........................................................................ 31 New Mg-Alloys for the Thixomolding™ Process Julio Aguilar, Martin Fehlbier, Andreas Bühring-Polaczek, Foundry Institute, Aachen University (RWTH), Germany............................................................................. 37 Structure and Morphology of Effective Obstacles in High Performance Mg – Rare Earth Base Alloys Bohumil Smola, Ivana Stulíková, Jitka Pelcová, Barry L. Mordike Faculty of Mathematics and Physics, Charles University, Prague 2, Czech Republic; Zentrum für Funktionswerkstoffe gGmbH, Clausthal-Zellerfeld; Institute of Materials Engineering and Technology, TU Clausthal, Clausthal-Zellerfeld, Germany ................ 43 Regularities in the Phase Diagrams and Behavior During Aging of the Ternary Mg-Base Alloys Containing Two Rare-Earth Metals of Different Subgroups Lazar Rokhlin, Tatyana Dobatkina, Nadezhda Nikitina, Baikov Institute of Metallurgy and Materials Science, Moscow, Russia ................................................... 49 Development of New Magnesium Alloys for Advanced Applications B. Bronfin, E. Aghion, S. Schumann, H. Friedrich, Dead Sea Magnesium Ltd, Beer Sheva, Israel; F. von Buch, Volkswagen AG, Wolfsburg, Germany ....................................................... 55

VIII Creep Resistance and Creep Kinetics of Mg-alloys W. Blum, Y. J. Li, X. H. Zeng, Universität Erlangen-Nürnberg, Inst. f. Werkstoffwissenschaften LS1, 91058 Erlangen, Germany; B. von Großmann, C. Haberling, H.-G. Haldenwanger, Audi AG, 85045 Ingolstadt, Germany......................................................................................................... 62 Creep Behaviour of MgZnRE Alloys Formed by Semi-Solid-Casting E. D. Morales, N. Hort, K. U. Kainer, GKSS Research Center Geesthacht GmbH, Geesthacht, Germany ....................................................................................................... 68 Creep Resistant Magnesium Alloys for Powertrain Applications Mihriban. O. Pekguleryuz, McGill, University, Department of Metals and Materials Engineering, University Street, Montreal, QC, Canada H3A 2B2; A. Arslan Kaya, Tubitak Marmara Research Center, P.O.B.21, Gebze, 41470 Turkey.. 74 Russian Ultralight Constructional Mg–Li Alloys. Their Structure, Properties, Manufacturing, Applications Fedor M. Elkin, Valentin G. Davydov, All-Russia Institute of Light Alloys, Moscow, Russia ............................................................................................................................... 94 The Effect of Grain Size on the Mechanical Properties of AM-SC1 G. Dunlop, Australian Magnesium Corporation, Milton, Australia; C.J. Bettles, J.R. Griffiths, K. Venkatesan, CRC for Cast Metals Manufacturing (CAST), CSIRO Manufacturing & Infrastructure Technology, Melbourne, Australia; Lihui Zheng and Ma Qian, CRC for Cast Metals Manufacturing (CAST), The University of Queensland, St Lucia, Australia.......................................................... 100 Thermal Properties of Mg-Li and Mg-Li-Al Alloys Alexandra Rudajevová, Pavel Lukácˇ, Department of Metal Physics, Charles University, Prague, Czech Republic; Stanislav Kúdela, Institute of Materials and Machine Mechanics SAS, Bratislava, Slovakia ............................................................................................................................ 106 Designing for Improved Creep Resistance in the RE – Free Magnesium Alloys Vladimir G. Tkachenko, International Center for Electronic materials Science and Applied Problems of Aerospace Technology, 3, Krzhyzhanovsky Str, Kyiv, 03680, Ukraine ....................................................................................................... 110 Ageing Characteristics and Creep Resistance of Mg-Y-Nd-Sc-Mn Alloy Ivana Stulíková, Bohumil Smola, Jitka Pelcová, Faculty of Mathematics and Physics, Charles University, Prague, Czech Republic; Ivana Stulíková, Bohumil Smola, Zentrum für Funktionswerkstoffe gGmbH, Clausthal-Zellerfeld, Germany; Barry L. Mordike, Institute of Material Science and Engineering, Technical University Clausthal, Clausthal-Zellerfeld, Germany ..................................................... 116 Deformation Behaviour of Mg-Li-Al Alloys at Room and Elevated Temperatures Zdeneˇk Drozd, Zuzanka Trojanová, Viera Gärtnerová, Faculty of Mathematics and Physics, Charles University, Prague, Czech Republic .................................................... 122

IX The Effect of Grain Size on the Bolt Load Retention Behaviour of AMC-SC1 C.J.Bettles, K.Venkatesan, CRC for Cast Metals Manufacturing (CAST), CSIRO Manufacturing & Infrastructure Technology, Melbourne, Australia; Lihui Zheng, Ma Qian, CRC for Cast Metals Manufacturing (CAST), The University of Queensland, St Lucia, Australia .................................................................................. 128 High Temperature Behaviour of a New Mg-Ca-Zn Alloy A. Zilberov, G. R. Goren-Muginstein and M. Bamberger, Department of Material Eng., Technion IIT, Haifa 32000, Israel .......................................................................... 134 Investigation of Die Cast Mg-Al Alloys with Additions of Sn, Si and Sb Dr. Per Bakke, Dr. Ketil Pettersen, Dr. Håkon Westengen, Hydro Aluminium a.s., Magnesium Competence Centre, Porsgrunn, Norway..................................................... 140 An Electron Microscopy Investigation on As-cast AZ91D Alloy Modified with Nitrogen A.A. Kaya, TUBITAK-MRC, MCTRI, Kocaeli, Turkey; O. Yucel, Istanbul Technical University, Istanbul, Turkey; D. Eliezer, Ben-Gurion University of The Negev Department of Materials Engineering Beer-Sheva Israel; E. Aghion, MRI, Beer-Sheva, Israel................................................................................. 151 Recrystallization Behaviour of Four Magnesium Alloys L.W.F. Mackenzie, F.J. Humphreys, G.W. Lorimer, Manchester Materials Science Centre, University of Manchester and UMIST, UK; K. Savage, T. Wilks, Magnesium Elektron Ltd, Manchester, UK .................................... 158 Dynamic Recrystallization of Mg-3Al-1Zn T. Laser, Ch. Hartig, R. Bormann, Technische Universität Hamburg-Harburg, 21073 Hamburg, Germany; J. Bohlen, D. Letzig, GKSS Forschungszentrum Geesthacht, 21502 Geesthacht, Germany ........................................................................................................................... 164 Mechanical Properties and Formability of PM Mg-Al Based Alloys K. Matsuzaki, K. Hatsukano, K.Hanada, M.Takahashi and T. Shimizu, National Institute of Advanced Industrial Science and Technology(AIST), AIST-EAST, 1-2 Namiki, TSUKUBA 305-3564, Japan ........................................................................ 170 Corrosion Properties of Conversion Plasma Coated Magnesium Alloys H. M. Nykyforchyn, M. D. Klapkiv, Karpenko Physico-Mechanical Institute of the National Academy of Sciences of Ukraine, Lviv, Ukraine; W. Dietzel, C. Blawert, GKSS-Forschungszentrum Geesthacht GmbH, Geesthacht, Germany ........................................................................................................................... 176 The effect of Variations in Melt Temperature on the Grain Structures in an AM60 Die Casting Hans I. Laukli, Otto Lohne, Lars Arnberg, Norwegian University of Science and Technology, Trondheim, Norway; Haavard Gjestland, Stian Sannes, Norsk Hydro ASA, Porsgrunn, Norway.................... 182

X Microstructural Characterisation of AZ91 Magnesium Alloy Andrzej Kiełbus, Maria Sozan´ska, Silesian University of Technology, Katowice, Poland; Lubomirc Ciz˙ek, VSB-TU Ostrava, Czech Republic ........................................................ 190 Microstructural Changes of AZ91 Magnesium Alloy after Heat Treatment Andrzej Kiełbus, Silesian University of Technology, Katowice, Poland; Lubomir Ciz˙ek, Libor Pawlica, VSB-TU Ostrava, Czech Republic................................. 196 Microstructure of Ultra Fine Grained Mg and Mg-10 wt.% Gd Prepared by HighPressure Torsion J. Cizek, I. Prochazka, I. Stulikova, B. Smola, R. Kuzel, V. Cherkaska, Charles University, Faculty of Mathematics and Physics, Prague, Czech Republic; I. Stulikova, B. Smola, Zentrum für Funktionswerkstoffe GmbH, Clausthal-Zellerfeld, Germany; R.K. Islamgaliev, O. Kulyasova, Institute of Physics of Advanced Materials, Ufa State Aviation Technical University, Ufa, Russia ............................................................ 202 Development of New Grain Structure and Tensile Properties Improving in a Hot Pressed and ECAP Processed ZK60 Magnesium Alloy A. Galiyev, R. Kaibyshev, M. Almakaev, Institute for Metals Superplasticity Problems, Ufa, Russia ....................................................................................................................... 208 Texture Formation and Texture Modelling of AZ31 Magnesium Wrought Alloy A. Styczynski, Ch. Hartig, R. Bormann, Arbeitsbereich Werkstoffphysik und – Technologie, Technische Universität Hamburg-Harburg, 21071 Hamburg, Germany; F. Kaiser, J. Bohlen, D. Letzig, GKSS-Forschungszentrum Geesthacht GmbH, 21502 Geesthacht, Germany............................................................................................ 214 Texture Development of AM20 Tensile Samples Under Load S.B. Yi, H.-G. Brokmeier, Institute of Materials Engineering, Technical University Clausthal, Clausthal-Zellerfeld, Germany; K.U. Kainer, T. Lippmann, Institute of Materials Engineering, GKSS-Research Center, Geesthacht, Germany ......................................... 220 Strengthening Phases in Extruded Quasicrystal Containing Mg-Zn-Y Alloys Alok Singh, A.P. Tsai, M. Nakamura, National Institute for Materials Science, Tsukuba 305-0047, Japan; M. Watanabe, A. KatoToyota, Motor Corporation, Toyota 471-8572, Japan................ 226 Grain Structure Characterization of AM60 Die Castings by Electron Backscatter Diffraction (EBSD) Measurements in SEM Hans I. Laukli, Otto Lohne, Lars Arnberg, Norwegian University of Science and Technology, Trondheim, Norway..................................................................................... 232 Possibilities of ECAP of Magnesium Alloy Katarzyna N. Braszczyn´ska, Institute of Materials Engineering, Technical University of Cze˛stochowa, Al. Armii Krajowej 19, 42-200 Cze˛stochowa, Poland .......................... 236

XI Electron Probe Micro Analysis of Sedimented Zirconium Particles in Magnesium Bruce Davis, Materials Science Department, Oxford University, now Magnesium Elektron, Manchester, UK; Keyna O’Reilly, Materials Science Department, Oxford University, John King, Magnesium Elektron, Manchester, UK............................................................................ 242 Superplasticity in Hot-rolled Magnesium Alloy AZ31 Sheets P. Zhang, J. Wendt, V. Šupik, J. Zhang, F. Holländer, Lehrstuhl Metallkunde und Werkstofftechnik, BTU-Cottbus, Cottbus, Germany ........................................................ 248 Extrusion of Magnesium Profiles and Microstructure Characterization Sören Müller, Ingwer Denks, Klaus Müller, Technical University Berlin, Berlin, Germany; Anke Pyzalla, Vienna Technical University, Vienna, Austria.......................................... 254 Investigating the Plastic Deformation Behaviour of Magnesium Alloy AZ31 by Plane Strain Forging of U and H Sections M.S.Yong, S.C.V.Lim, Singapore Institute of Manufacturing Technology, Singapore ......................................................................................................................... 260 Application of Isothermal Rolling for Improvement of Sheet Formability of a Magnesium Alloy A. Galiyev, R. Kaibyshev, D. Voronin, Institute for Metals Superplasticity Problems, Ufa, Russia ....................................................................................................................... 266 Influence of Al Content (. Aghion and Bromfin (11) investigated the microstructure and properties of various AZ91 ingots by light and scanning electron microscopy techniques. Clark studied Mg-9%Al alloy by TEM after aging (12). Crawley et al. (13,14) later complimented Clark’s report, and furthermore suggested crystallographic orientation relationships between the α matrix and β products in aged structures revealing that the plates form parallel to the basal planes of the hexagonal matrix. Previous studies (12-14) on the aging behaviour of die-cast AZ91 revealed that the precipitates of β phase form with various morphologies that includes discontinuous and continuous precipitation products with lamellar, and plate morphologies, respectively. It was also reported that aging process did not result in the formation of additional morphological features. A difference was noticed only in the precipitate density and the size, double aging resulting in a higher density of β plates with a higher aspect ratio. In a study on AZ91E, Bettles (15) made a distinction between the previously reported plate shaped β and what he called ‘lozange’ shaped continuous precipitation products, and also revealed that trace additions of copper inhibited the discontinuous precipitation. Assessment of the individual effects of particular types of precipitate morphologies on the properties of the material is difficult as all of them exist in the same material. Some information has been reported regarding the effect of discontinuous precipitation on corrosion properties (16). Generally this kind of cellular microstructures are considered as detrimental to the toughness of the material. When evaluating the effect of discontinuous precipitation on mechanical properties the nodule size should also be considered. Elimination of this type of precipitation reaction in Mg-based alloys would be desirable as it is also claimed to accelerate creep at high temperatures (17). Furthermore, absence of this constituent would presumably allow for more extensive continuous precipitation leading to better strengthening effect. Coexistence of a few morphologies of what is essentially β, Al12Mg17, in the same material raises questions that are worth considering. Coarse β particles at the grain boundaries are the product of a divorced eutectic reaction from the aluminum-enriched part of the liquid metal that solidifies the last in a casting (11). Unfortunately, little has been proposed towards explaining the relative conditions for the formation of the competing continuous and discontinuous precipitate morphologies. The continuous precipitation products in AZ91 are known to nucleate on dislocations and twins in grain interiors (12-14). Increased precipitation in pre-deformed samples strongly supports this idea (12). However, what leads to their different shapes is not clear. Although it is known that slower cooling rates increases the amount of discontinuous precipitation, in the case of magnesium alloys other conditions and particularly alloy modifications that may prevent the formation of this detrimental morphology need be explored in more detail. 1.1

Discontinuous and Continuous Precipitation

In solid-solid phase transformations there are two reactions that lead to identical cellular morphologies, eutectoid transformation and discontinuous precipitation. Unlike the eutec-

152 toid transformation, discontinuous precipitation does not lead to decomposition of the matrix into two new phases. It merely changes the orientation and the composition of the matrix in which the precipitation of a second phase takes place. An important characteristic of this reaction is that the compositional change across the transformation front occurs discontinuously. The reaction mechanism of discontinuous precipitation has been reviewed by a number of workers and various suggestions have been made (18-21). Although not changing the general morphology, some exceptional examples of discontinuous precipitation have also been reported to exist (22-24). It should also be pointed out that apparently in all of the previous work the discontinuous precipitation has been investigated under isothermal reaction conditions rather than continuous cooling. Discontinuous precipitation in magnesium alloys has been studied in terms of composition profiles in cellular nodules (25-27). It was shown that the growth of discontinuous precipitation is affected by defects in matrix and the habit plane of the lamella can change from basal to pyramidal planes. It has also been recognized that regions of α phase behind the transformation front, though to a lesser degree compared to the untransformed matrix, remains supersaturated. In the case of discontinuous precipitation, it is known that the boundary diffusion predominates the growth of the precipitates and that supersaturation plays an important role. High angle boundaries are considered as sites of higher potential for the formation of the cellular nodules as is the case in magnesium alloys. However, the debate on a universally acceptable theory to allow the prediction of discontinuous precipitation is yet to be resolved. Continuous precipitation is, on the other hand, a more common phenomenon. Like its discontinuous counterpart, this reaction also gives rise to the formation of only one new phase. In systems where discontinuous precipitation is absent the term ‘continuous’ usually becomes redundant. This type of reaction does not involve a moving transformation front that is common to a colony of precipitates in the matrix grain. The precipitation merely takes place at energetically favorable intragranular sites such as dislocations, twin boundaries, and stacking faults generally by establishing an orientation relationship with the matrix. Each precipitate changes the composition of the matrix in their immediate vicinity, leading to a reduction in supersaturation, without changing the crystal orientation of the host lattice. However, the entire precipitation potential is generally not exhausted in continuous cooling conditions, or after short annealing times, and the precipitates that form first may not be the equilibrium type.

2

Experimental Procedure

The chemical composition of the alloy studied is given in Table 1. Preheated AZ91 scrap from a cold chamber die casting machine was heated to 720 °C in a steel crucible. A control sample was prepared by casting an ingot from the melt before nitrogen purging was started. The melt was then purged with nitrogen gas through a plunger for 30 minutes, and 60 minutes and ingots were cast representing both cases. All ingots were about 5 x 5 x 20 cm. in size. Table 1. Chemical composition of the alloy (wt%)

Mg

Al

Zn

Mn

Fe

Cu

Ni

Si

89.96

9.1

0.8

0.16

0.0061 0.0014 0.0010 0.005

Ca

Be

0.0068 0.0004

153 SEM examinations were carried out with a JEOL 840A microscope. TEM studies were conducted with a JEOL 2010 high-resolution microscope operated at 200 kV. Metallographic samples were cut from ingots by using a low speed diamond slit. Samples for light microscopy and SEM examinations were etched with 5% HNO3 solution. Thin foils for TEM were cut from a bulk specimen using spark erosion technique followed by mechanical polishing down to 100 μ, and dimpling with a micro grinder, and finally perforated via ion milling in cooled condition.

3

Experimental Results

Fig. 1 is a SEM micrograph showing the classical morphology in the sample not treated with nitrogen. Aluminum enriched regions towards the periphery of each grain were clearly noticeable. These regions are the result of coring as the equilibrium conditions were not achieved during the solidification of ingots. As shown in Fig. 1 the microstructure of the alloy was typical of AZ91 involving coarse β (Mg17Al12) particles along the grain boundaries as well as lamellar precipitation as a product of discontinuous cellular reaction.

Figure 1. SEM micrograph showing the microstructure of the control sample. Note the coarse β and the cellular precipitation along the grain boundaries and the contrast difference due to coring in grains.

Fig. 2 shows the microstructure of the ingot that was cast after purging the molten metal with nitrogen for 30min. As seen in the SEM micrographs the morphology of grain boundary coarse β particles is very different, displaying a jagged interface and copious amount of precipitation on them. Electron diffraction in TEM revealed that these are β particles of Mg17Al12 type (Fig. 3). TEM examination also identified the precipitates on β particles as αMg. No precipitation with cellular morphology was observed in this nitrogen treated sample. Fig. 4 is an SEM micrograph revealing the microstructure of the second ingot that was cast after purging the liquid metal with nitrogen for 1hr. The microstructure is clearly different from the other two samples. The coarse precipitates do not have a jagged interface nor the precipitates on them thus differing from those of the other nitrogen treated sample. On the other hand the cellular discontinuous precipitation product is also absent in the microstructure and therefore differing from the control sample that received no nitrogen treatment at all.

154

b Figure 2. (a) SEM micrograph showing the microstructure of the ingot cast after 30mins. of purging the molten metal with nitrogen. Note that the morphology of β has changed and that the cellular precipitation is absent. (b) a higher magnification SEM image of β particle.

a

b

(110)β (110)

Mg

c Figure 3. Bright Field (a) and Dark Field (b) TEM micrographs showing the coarse β; diffraction pattern (c) revealing the phases of β as Mg17Al12 and the precipitates as αMg.

155

Figure 4. SEM micrograph showing the microstructure of the ingot cast after 60 mins. of purging the molten metal with nitrogen. Note the absence of cellular precipitation and that the morphology of coarse β resembles that of the control sample.

4

Discussion

Knowing that the solubility of nitrogen in magnesium is practically nil one would perhaps not consider the addition of nitrogen as a potentially worthwhile trial towards creating a change in microstructure. The attempt on this ground may, at first sight, seem futile, or adventurous at best. However, this lack of solubility of nitrogen in magnesium coupled with the knowledge that the lamellar morphology is a product of an interface-controlled discontinuous reaction, are what led to this seemingly adventurous attempt. It was thought that if an unlikely small species such as nitrogen can be introduced into the alloy in the molten state, and if it can remain in solid at high temperatures in some minute quantities during solidification it would constantly be rejected to the crystal boundaries before it is totally diffuses out of the alloy. Assuming that such a scenario is valid one would then expect that introduction of nitrogen might alter at least the kinetics of the interface controlled reaction that forms the lamellar morphology and therefore modify it. The results of this study may be taken to support this proposition. Indeed, this study has shown that an ingot structure can be altered if the liquid metal is purged with nitrogen prior to casting. The period of purging seem to have different effects on the development of the microstructure. Shorter purging time seemed to promote a welldeveloped divorced eutectic leading to formation of coarse β particles with numerous αMg precipitations on them. This shorter nitrogen treatment seemed to have also changed the shape of the β particles leading to a serrated appearance along the edges. Longer purging appeared to have changed this appearance reverting back to the classical β morphology as seen in the control sample. Purging the molten metal prior to casting seems to change the microstructure considerably eliminating the cellular precipitation altogether. Admittedly, the explanation of the observed effects needs more detailed studies. Based on the current work the morphological differences in the appearances of β can only be related to the length of nitrogen purging operation.

156

6

References

[1] I.J. Polmear, Light Alloys, 3rd ed., Arnold, 1995, p. 196-247. [2] M.T. Murray, W.P. Sequeira, and R. D’Allesandro, SAE Technical Paper Series, 960420. [3] W.P. Sequeira, M.T. Murray, G.L. Dunlop, and D.H. John, Int. Conf. on Automotive Alloys, TMS, February, Orlando, Florida, 1997, p. 169-183. [4] H. Rosenson, Z. Koren, M. Bamberger, and E. Aghion, Proc. of the 9th International Metallurgy and Materials Congress, vol.2, Istanbul, Turkey, June-1997, p. 1537-1542. [5] A. Beste et al., Magnesium Alloys and Their Applications, eds. B.L. Mordike and K.U. Kainer, Proceedings volume sponsored by Volkswagen AG, Wolfsburg, Germany, April-1998, p. 253-258. [6] H. Sasaki, M. Adachi, and T. Sakamoto, Conf. Proc. IMA 53, Magnesium Material Advancing to the 21st Century, Ube City, Yamaguchi, Japan, June-1996, p. 86-92. [7] A. Mwembela and H.J. McQueen, ‘Hot workability of magnesium alloy AZ91’, Light Metals 1993, p. 523-532. [8] M. Dargusch et al., Proc. of 3rd Int. Magnesium Conf., edt. By G.W Lorimer, Inst. of Materials, Manchester, UK, April-1996, p. 153-166. [9] K. Pettersen, O. Lohne, and N. Ryum, Met. Trans. A, 1990, 21A, 221-230. [10] K. Pettersen and N. Ryum, Met. Trans. A, 989, 20A, 847-852. [11] E. Aghion and B. Bromfin, Magnesium Alloys and Their Applications, ed. B.L. Mordike and K.U. Kainer, Proceedings volume sponsored by Volkswagen AG, Wolfsburg, Germany, April-1998, p. 295-300. [12] J.B. Clark, Acta Metallurgica, 1968, 6, 141-152. [13] A.F. Crawley and K.S. Milliken, Acta Metallurgica, 1974, 22, 557-562. [14] A.F. Crawley and B. Lagowski, Met. Trans., 1974, 5, 949-951. [15] C.J. Bettles, ‘Magnesium Alloys and their Applications’, ed. B.L. Mordike and K.U. Kainer, Proceedings volume sponsored by Volkswagen AG, Wolfsburg, Germany, April-1998, p. 265-270. rd [16] P. Uzan, D. Eliezer, and E. Aghion, Proc. of 3 Int. Magnesium Conf., ed. G.W Lorimer, Inst. of Materials, Manchester, UK, April-1996, p. 43-50. [17] M.S. Dargush and G.L. Dunlop, Proc. of Materials 98, Institute of Materials Engineering, Australasia Ltd., ed. Michael Ferry, Wollongong, Australia, 1998, p. 579-584. [18] E. Hornbogen, Met. Trans., 1972, 3, 2717-2727. [19] W.Gust, Proc. Conf. on Phase Transformations, Inst. of Metals, 1979, II, p.27. [20] D.B. Williams and E.P. Butler, Int. Met. Rev., 1981, 3, 153-183. [21] G.R.Purdy, Proc.Int. Conf. on Solid-Solid Phase Transformations, ed. Aaronson, Laughlin, Sekerka, Wayman, AIME, 1982, p.521-530. [22] G.R. Purdy and N. Lange, 2nd Acta-Scripta Metallurgica Conf., Germany, 1983, p.214-221. [23] M. Kawase, H. Emoto and M. Kikuchi, Conf. Proc. on Phase Transformations, Inst. of Metals, Cambridge, 1987, p.254-258. [24] W.E. Voice and R.G. Faulkner, Conf. Proc. on Phase Transformations, Inst. of Metals, Cambridge, 1987, p.262-265.

157 [25] D. Duly, M.C. Cheynet, and Y. Brechet, Acta Metall., 1994, 42, 3843-3854. [26] D. Duly, M.C. Cheynet, and Y. Brechet, Acta Metall., 1994, 42, 3855-3863. [27] D.A. Porter and J.W. Edington, Proc. R. Soc. London A, 1977, 358, 335.

Recrystallization Behaviour of Four Magnesium Alloys L.W.F. Mackenzie+, F.J. Humphreys+, G.W. Lorimer+, K. Savage* and T. Wilks* +

Manchester Materials Science Centre, University of Manchester and UMIST, UK * Magnesium Elektron Ltd, Manchester, UK

1

Abstract

The thermomechanical behaviour of four magnesium alloys, AZ31, AS11, WE43 and ZC71, has been studied using channel die compression tests at temperatures between 558 and 673 K. The recrystallization behaviour of each alloy was observed between temperatures of 558 and 798 K, using optical microscopy and electron backscattered diffraction. AZ31, AS11 and ZC71 alloys underwent some dynamic recrystallization during the deformation at 558K. In AS11 and WE43 the nucleation of recrystallized grains occurred at both second phase particles and pre-existing grain boundaries; in AZ31 and ZC71 recrystallization was found to occur predominately at grain boundaries. The channel die compression results have been compared with the microstructures produced during extrusion at temperatures between 543 and 663 K.

2

Introduction

Some magnesium alloy extrusions are known to exhibit very different tensile and compressive strengths [1]. Extrusions with a fiber texture may have anisotropic mechanical properties if crystal orientation prevents twinning during tension parallel to the extrusion axis. This results in an increase in tensile strength relative to that in compression. It has been suggested that differences in texture in magnesium alloy extrusions can be attributed to recrystallization mechanisms and that particle stimulated nucleation of recrystallization (PSN) might be responsible for a reduction in texture [2]. The understanding of recrystallization behaviour is therefore of interest when considering mechanical properties. Study of recrystallization in magnesium alloys has been limited to a few systems such as Mg-Al-Zn and Mg-Zn-Zr [3, 4, 5]. It is important to develop an understanding of recrystallization mechanisms in magnesium alloys if the microstructure-property relationships are to be understood during thermomechanical processing. Four alloys were investigated in this work: AS11, AZ31, WE43 and ZC71.

3

Experimental

The alloys were received as sand castings of dimensions 200 x 200 x 25 mm. The sand cast AS11, AZ31 and ZC71 were homogenized at 708 K for 24 hours. In order to assist

159 the break-up of the second phase the ZC71 then underwent a 15% deformation by rolling at 673 K and a further 24 hour homogenization at 708 K. The WE43 was homogenized for 24 hours at 798 K and then 16 hours at 723 K. The alloys were deformed in a channel die to a true strain of 0.4 at a true strain rate of 10-4 s-1 and temperature of 558 K. 80 mm diameter billets of AS11, AZ31 and ZC71 were extruded into 19 diameter mm bars at 573, 555 and 543 K, respectively. WE43 was extruded into 38 mm diameter bar at 608 and 663 K. The tensile and compressive strengths of each alloy were measured parallel to the extrusion axis in the as-extruded state and after a recrystallization heat treatment. Full recrystallization was obtained when AS11, AZ31 and ZC71 were homogenized for 16 hours at 708 K and WE43 for 16 hours at 798 K.

4

Results and Discussion

4.1

AS11

Figure 1a shows the microstructure of AS11 deformed in a channel die at 558 K to a true strain of 0.4 at a true strain rate of 10-4 s-1. Twinning and nucleation of new grains occurred during deformation. Recrystallization is associated with original grain boundaries, second phase and eutectic particles. In Figure 1b recrystallization and twinning have occurred during extrusion at 573 K. New grains are associated with original grain boundaries, second phase and eutectic particles. Nucleation at small second phase particles

Nucleation at grain boundary

Twinning

Nucleation at particle

100 microns

Fig 1a. AS11 deformed in a channel die at 558 K to a -4 -1 strain of 0.4 at a strain rate of 10 s showing twinning and nucleation of new grains

50 microns Fig 1b. AS11 extruded at 573 K showing twinning and nucleation of new grains

The nucleation of new grains during deformation indicates that dynamic recrystallization has occurred in AS11 during both channel die deformation at 558 K and extrusion at 573 K. The association of new grains with second phase and eutectic particles indicates that dynamic PSN is probably a recrystallization mechanism in AS11. 4.2

AZ31

Figure 2a shows AZ31 deformed in a channel die at 558 K. New grains have nucleated at original grain boundaries in a “necklace effect” similar to that reported by Ion and Hum

160 phreys in a Mg-0.8Al alloy [6]. During extrusion at 555 K the AZ31 alloy fully recrystallised. Examination of the extrusion discard, Figure 2b, showed serrated grain boundaries and the nucleation of new grains similar to that reported by Galiyev, Kaibyshev and Sakai in a Mg-Zn-Zr alloy(4). New grains also nucleated at structural inhomogeneities such as shear bands. Fine grains nucleating at original grain b d i

100 microns Fig 2a. AZ31 deformed in a channel die at 558 K to -4 -1 a strain of 0.4 at a strain rate of 10 s showing nucleation of new grains

New grains nucleating at a serrated grain boundary

10 microns Fig 2b. AZ31 discard from material extruded at 555 K showing serrated grain boundaries beginning to nucleate new grains

The nucleation of new grains indicates that AZ31 underwent dynamic recrystallization during both channel die deformation at 558 K and extrusion at 555 K. Nucleation occurred predominantly at grain boundaries, although during extrusion new grains also nucleated at other structural inhomogeneities within the microstructure. 4.3

WE43

Figure 3a is of a WE43 sample deformed in a channel die at 558 K to a true strain of 0.4 at a true strain rate of 10-4 s-1 and then heated for 15 minutes at 673 K. The micrograph shows nucleation of new grains at original grain boundaries; many of the new grains are associated with grain boundary particles. New grains have also nucleated at particles within the matrix, similarly to that reported in an Al-Si alloy by Humphreys [7]. WE43 was extruded at 608 K and 663 K. When extruded at 663 K the material fully recrystallized. At 608 K some areas of the extrusion did not fully recrystallize, thereby allowing the identification of the origin of some new grains. In Figure 3b a new grain can be seen to have nucleated at a second phase particle. Figure 4 is an EBSD map of WE43 deformed in a channel die at 558 K and then partially recrystallized after a heat treatment of 15 minutes at 673 K. The thick and thin black lines represent high and low angle grain boundaries respectively. New grains have nucleated at original grain boundaries. Nucleation is also associated with second phase particles. A high concentration of low angle grain boundaries can be observed around some particles.

161

Nucleation at grain boundaries

Nucleation at particles

Nucleation at a particle

20 microns

10 microns

Fig 3a. WE43 deformed in a channel die at 558 K to -4 -1 a strain of 0.4 at a strain rate of 10 s and partially recrystallized at 673 K showing nucleation of new grains at grain boundaries and particles

Concentration of low angle grain boundaries at particle

Fig 3b. WE43 extruded at 608 K showing nucleation of a new grain at a particle

Nucleation associated with second phase particles

Nucleation at a grain boundary 50 microns Figure 4. EBSD map of WE43 deformed in a channel die at 558 K to a strain of 0.4 at a strain rate of 10-4 s-1 and partially recrystallized during a 15 minute heat treatment at 673 K

In both WE43 deformed in a channel die at 558 K and then partially recrystallized at 673 K and WE43 extruded at 608 K new grains were often associated with second phase particles. New grains also nucleated at grains boundaries and those new grains were often associated with grain boundary particles. This indicates that PSN is a mechanism of recrystallization in WE43. In Figure 4, EBSD has been used to highlight a concentration of low angle grain boundaries associated with a second phase particle; this could be associated with the formation of a particle deformation zone which is a precondition for PSN [7].

162 4.4

ZC71

Figures 5a and 5b show the microstructure of ZC71 deformed in a channel die at 558 K at a strain rate of 10-4 s-1. New grains have formed in a “necklace effect” at old grain boundaries. When extruded at 543 K the material was fully recrystallized with a very fine grain size; the origin of the recrystallized grains was not identified. Nucleation at grain boundary

20 microns 80 microns Figure 5a ZC71 deformed in a channel die at 558 K at strain rate of 10-4 s-1 showing nucleation of new grains at grain boundaries

Figure 5b ZC71 deformed in a channel die at 558 K at strain -4 -1 rate of 10 s showing nucleation of new grains at grain boundaries

The ZC71 alloy underwent dynamic recrystallization when deformed in a channel die at 558 K at a strain rate of 10-4 s-1. New grains were observed to nucleate predominantly at grain boundaries, Figure 5b. 4.5

Tensile and Compressive Testing

Each alloy was tensile and compression tested in the as-extruded form and then after heat treatment to produce a fully recrystallized microstructure. The values of 0.2 % proof stress in tension and compression are given in Table 1. AS11 and AZ31 alloys exhibited some anisotropy in tensile and compressive strength, which increased after recrystallization at 708 K. The WE43 and ZC71 alloys exhibited virtually no anisotropy in the as-extruded state. After heat treatment at 798 and 708 K, respectively, the WE43 again showed no anisotropy, whereas the ZC71 showed a large anisotropy. WE43 extruded at the higher temperature of 663 K exhibited no evidence of anisotropy in any condition. Table 1. Tensile and compressive strengths in the as-extruded and fully recrystallized conditions

Alloy AS11 as extruded AS11 recrystallized AZ31 as extruded AZ31 recrystallized WE43 as extruded WE43 recrystallized ZC71 as extruded ZC71 recrystallized

Extrusion Temp / K 573 543 608 555

Tensile Compressive Strength / MPa Strength / MPa 146 125 140 100 157 136 139 110 216 219 100 104 196 204 207 95

Tensile/Compressive Ratio 1.17 1.40 1.15 1.26 0.99 0.96 0.96 2.18

163 With fiber texture, crystal orientation allows twinning to take place during compression parallel to the extrusion axis but not in tension. This increases the tensile strength relative to the compressive strength, which may explain the anisotropy observed in AS11, AZ31 and ZC71. WE43 may have a random texture similar to that reported in previous research1.

5

Conclusions

The recrystallization behaviour of AS11, AZ31, WE43 and ZC71 has been studied during channel die deformation and hot extrusion. In AZ31 and ZC71 recrystallization occurred predominantly at grain boundaries, whereas in AS11 and WE43 new grains were observed nucleating both at grain boundaries and second phase particles; it appears that PSN occurs in these alloys. Tensile and compressive testing of extruded samples revealed anisotropy of strength in AS11 and AZ31 alloys, but not in WE43. The ZC71 alloy exhibited anisotropy in mechanical properties after full recrystallization but not in the as-extruded state.

6

References

[1] P. Prangnell, E. Ball, Scripta Met. And Mat., 1994, 31, 111-116. [2] F.J. Humphreys, M. Hatherley, Recrystallisation and Related Annealing Phenomena, 1st ed., Pergamon, Chapter 8. [3] M.R. Barnett in Magnesium Alloys 2003 (Eds.: Y Kojima , T. Aizawa, K. Higashi and S. Kamado), Trans Tech Publications, Switzerland, 2003. [4] A. Galiyev, R. Kaibyshev, T. Sakai in Magnesium Alloys 2003 (Eds.: Y Kojima, T. Aizawa, K. Higashi and S. Kamado), Trans Tech Publications, Switzerland, 2003. [5] X. Yang, H. Miura, T. Sakai in Magnesium Alloys 2003 (Eds.: Y Kojima, T. Aizawa, K. Higashi and S. Kamado), Trans Tech Publications, Switzerland, 2003. [6] S.E. Ion, F.J. Humphreys, S.H. White, Acta Metall., 1982, 30, 1909-1919. [7] F.J. Humphreys, Acta Met, 1977, 25, 1323-1344.

7

Acknowledgements

The authors would like to thank the EPSRC and Magnesium Elektron Ltd for their support.

Dynamic Recrystallization of Mg-3Al-1Zn T. Laser1, Ch. Hartig1, R. Bormann1, J. Bohlen2, D. Letzig2 1 2

Technische Universität Hamburg-Harburg, 21073 Hamburg, Germany GKSS Forschungszentrum Geesthacht, 21502 Geesthacht, Germany

1

Introduction

Due to their high specific strength, magnesium alloys offer a large potential in weight reduction to the automotive industry where less weight is leading to less emission and thus to more efficiency. The aim to fabricate structural parts out of Mg alloys drives the interest more and more to wrought alloys that offer a better ductility and strength compared to cast alloys. Magnesium wrought alloys can only compete against other light weight materials as long as the processing is inexpensive and well understood. Therefore, one step deformation of the feedstock that results in a fine recrystallized microstructure is a major goal. The poor cold formability of magnesium wrought alloys lets this goal only achieve at elevated temperatures accompanied with dynamic recrystallization (DRX). In several studies [1, 2, 3, 4] the dynamic recrystallization of Mg-3Al-1Zn (AZ31) during hot working has already been analyzed, but always starting from a fine grained microstructure (dm < 100 μm). In this work a coarse grained as-cast alloy was examined. The technological interest for a deformation process achieving a homogeneous microstructure with fine grain size was the main driving force.

2

Material and Experimental Procedure

The material employed in the present study was the wrought magnesium alloy AZ31. It was supplied in strand casting billets with 175 mm in diameter. The as-cast high purity material was cut in slices perpendicular to the cast axis and annealed for 20 hours at 400 °C for homogenization (air cool). Compression samples with an h/d-ratio of 1.5 were spark-eroded out of the slices with the compression axis lying parallel to the cast axis (Fig. 1). Uniaxial compression tests were performed at constant strain rates on a Zwick universal testing machine equipped with an electrical resistance furnace and a strain rate controlling unit. The tests were carried out in a temperature range between 200 °C and 400 °C in steps of 50 K. To ensure stable temperature conditions, the system with sample was held at testing temperature for approximately half an hour before deformation. The strain rate · -1 -2 -3 -4 -1 was varied in four steps: ε = 10 , 10 , 10 and 10 s . To reduce the friction between the sample and the compression plates, the top and bottom of the samples were lubricated with boron nitride. To conserve the microstructure that developed during forming, the samples were quenched in water immediately after unloading. The microstructure was studied by means of optical light microscopy (LM). With regard to uniaxial deformation only the center region of the compression samples was analyzed perpendicular to the load axis. The specimens were etched with an etching solution containing picric (effect on grain surface) and acetic (effect on grain boundary) acid.

165

Figure 1. Sample preparation out of strand cast billet (AZ31)

3

Figure 2. Initial strand cast microstructure (LM)

Results and Discussion

The initial as-cast microstructure after homogenization is shown in Fig. 2. An inhomogeneous grain size distribution with a mean grain size of 640 μm is observed. Precipitates of Mg17Al12 are homogeneously distributed in the matrix. 3.1

Flow stress behaviour

The true stress as function of the compression strain is shown in Fig. 3 for different strain rates and temperatures. The compression tests were carried out up to a maximum strain of ε = 1. In all flow curves the flow stress increases to a maximum and then decreases to a minimum for higher strains. This observed flow behavior is characteristic for hot working processes, where DRX leads to work softening [5]. The range of minimum flow stress is often termed in literature as a “steady state stress” [4, 6]. For very high strains (ε > 0.8) the flow stress may strongly increase especially at high flow stresses (i.e. low temperatures and high strain rates) due to friction. The maximum flow stress (σmax) moves to higher values of strain with increasing strain rate and decreasing temperature. At high temperatures and low strain rates, the maximum in the flow stress curve disappears and the flow behavior becomes nearly ideal plastic. The difference between the peak stress and the minimum stress is most significant for the mid temperature regime (250-350 °C).

Figure 3. Representative flow curves at different temperatures and strain rates

166 Neglecting the influence of friction, the minimum flow stress can be associated with a steady state flow stress σS [5,6]. Steady state flow commonly follows the Dorn equation: n

⎛˰ ⎞ ⎛ Q ⎞ ε& = A ⎜ V ⎟ exp ⎜ − m ⎟ G ⎝ ⎠ ⎝ RT ⎠

(1)

which has been used for a description of the measured steady state flow stresses as function of strain rate and temperature. Fig. 4 shows the strain rate as a function of the normalized1 stress for selected temperatures in a double logarithmic plot. The stress exponent · rises from n = 5.7 at 400 °C to n = 8.7 at 250 °C. At ε = 10-1 s-1 the resulting steady state flow stresses are lower than expected after a power law (Fig. 4). It has to be pointed out that creep laws may not be suitable for the description of processes where DRX is the controlling deformation mechanism superimposing the thermal activated slip of dislocations.

Figure 4. Power law behaviour in low strain rate regime, double logarithmic plot

Figure 5. Work hardening curves θ = ∂σ/∂ε, ε = 10 s , arrows indicating point of inflection ·

-3

-1

The work hardening behavior of the tests carried out at ε = 10-3 s-1 is shown in KocksMecking plots [7] in Fig. 5. After the theory of Poliak and Jones [8], a necessary condition for the onset of DRX is given by ∂/∂σ(-∂θ/∂σ) = 0, the point of inflection in the stage IV of the work hardening curve. After this condition DRX starts at stresses σ/σmax = 0.65 (T = 200 °C) up to σ/σmax = 0.9 (T = 350 °C) well below the peak stress, and at strains of ε ≈ 0.05, approximately 0.1 smaller than the corresponding peak strain (cf. Fig. 5). ·

3.2

Microstructural evolution

As already stated, all flow curves show a characteristic DRX behavior. This result can also be verified by examination of the microstructures. Figures 6a-6c show the microstructural evolu· tion for a specimen deformed at T = 300 °C and ε = 10-3 s-1 for different strains. All micrographs display the center of the specimen. From Fig. 6a it is obvious that the onset of DRX takes place at low strains before the peak flow stress is reached: fine globular DRX-grains have formed in the regions of grain boundaries. Some grains exhibit deformation twins. With increasing strain, several characteristic features develop in the microstructure (Figs. 6a-6c): (i) A flattening of the big original grains in load direction accompanied by further formation of 1

Shear modulus calculated with temperature dependence of Youngs modulus [9] for pure Magnesium.

167 deformation twins. (ii) At grain boundaries, necklaces of DRX grains are formed. Also at twin boundaries originating from deformation twinning, chains of DRX grains are visible at higher magnification (compare detail in Fig. 6b) (iii) At high strains, layers of DRX grains are clustering and consuming a major part of the microstructure. For strains ε ≥ 1.0, the necklace structure described above is always fully developed, i.e. all boundaries are decorated with chains of DRX grains. At low temperatures (Fig. 6d) deformation twinning increases drastically, fragmenting the globular grains of the initial microstructure. With an increase in temperature and a decrease in strain rate, the amount · of deformation twins decreases. At 400°C and ε = 10-4 s-1, no more twinning is observed.

a) T = 300 °C, ε& = 10-3 s-1, ε& = 0.15

b) T = 300 °C, ε& = 10-3 s-1, ε& = 0.3, detail showing onset of DRX at deformation twins

-3 -1 c) T = 300 °C, ε& = 10 s , ε = 1.0

-2 -1 d) T = 200 °C, ε& = 10 s , ε& = 1.0

Figure 6. Selected microstructures (LM), load axis from top to bottom

The total recrystallized volume fraction of each sample can be given qualitatively only because of overall inhomogeneous deformation conditions. A good quantitative approximation was found in limiting the region of analysis to the center region where uniaxial compression can be assumed. The results state that not more than 60% of recrystallized volume fraction can be obtained. This maximum of recrystallized volume fraction was · found for the mid temperature regime (T = 300 °C) and the lowest strain rate ε = 10-4 s-1. The mean size of the recrystallized grains situated in the DRX chains described above was determined. For a description of the grain size as a function of deformation conditions usually the steady state flow stress or the Zener-Hollomon parameter [6] are used. Fig. 7a shows the mean grain size dm as function of the normalized steady state flow stress σS for selected parameter sets. A mean grain size between 1 and 11 μm is observed under the cho-

168

a)

b)

Figure 7. Recrystallized grain size as function of a) normalized stress b) Zener-Hollomon parameter

sen deformation conditions. A clear flow stress dependence can be stated, the mean grain size decreases for higher stresses. For the deformation conditions illustrated by Figs. 6a-6c a mean grain size was observed being independent of the strain range (ε = 0.15 - 1.0). · The Zener-Hollomon parameter Z = ε exp(Q/RT) gives a reasonable guideline for deformation processes of technological relevance. For the determination of Z, a mean activation energy (Qm = 109 KJ/mol) was used, a value obtained from Arrhenius plots for constant stresses assuming power law behavior (Eq. 1). The mean grain size dm of recrystallized grains as a function of the Zener-Hollomon parameter is shown in Fig. 7b. The assumed slope changes for higher values of the Zener-Hollomon parameter, indicating different regimes of the evolution of DRX at high temperatures, low strain rates (low Z) and low temperatures, high strain rates (high Z). 3.3

Discussion

A qualitative similar behavior of the strain rate dependence of the flow stress was also observed for the magnesium alloy ZK60 [6] and for a fine grained magnesium alloy AZ31 [1] in compression tests. In detail, significant differences occur, especially concerning the strain hardening and the microstructural development (necklace structure). These differences may be ascribed to the coarse grained initial microstructure: The development of a necklace structure becomes much more obvious in coarse material and the resulting strain hardening is not only due to the competition between dislocation storage and dynamic recovery but also heavily influenced by the microstructural evolution [11]. An empirical relationship between strain rate, stress and temperature throughout the whole deformation range was found. Power law behavior was only observed for the low strain rate regime. At present, a complete description of the deformation behavior and microstructural evolution during DRX is lacking and only critical conditions for the occurrence of DRX can be given [8,10]. The results in the present study are in accordance with the “Poliak condition” (∂/∂σ(-∂θ/∂σ) = 0): The onset of DRX, as visible in LM, takes place after the “Poliak stress” but distinctly before the peak stress is reached. Similar as for ZK60 [6], at least two different regimes of the development of the grain size seem to be distinguishable, reflecting a change of nucleation mechanisms at low

169 temperatures and high stresses. A more detailed investigation of mechanisms for the material in the present study would require TEM observations, as done for ZK60 in [6]. The resulting necklace structure was also previously observed in AZ31 [2] and in ZK60 [6]. In [6] a theory for the formation of the necklace structure is proposed, where nonbasal slip of dislocations in the region of grain boundaries lead to accumulation of dislocations and hence to formation of high angle boundaries resulting in DRX grains. Tan and Tan [3] determined the optimum deformation conditions to receive a fine grained recrystallized structure in AZ31 to be T = 250°C at ε& = 10-4 s-1. In their study, a fine grained commercial material (12 μm) developed 85% of recrystallized volume fraction. For the coarse grained material analyzed in this study the optimum conditions (60% of recrystallized volume fraction) were found at the same strain rate but at a higher temperature: T = 300 °C. At 250 °C only 45% of recrystallized volume was observed. To receive high volume fractions of recrystallized areas, a two step deformation of the coarse grained initial microstructure might be advisable. In a first step a high fraction of deformation twins is generated at low temperature (200 °C) and low strain, whereas in a second step complete recrystallization at T = 300 °C may be achieved.

4

Conclusions

• The deformation of the studied coarse grained AZ31 alloy at elevated temperatures leads to the formation of DRX grains nucleating at grain and twin boundaries and forming a necklace structure throughout the whole deformation range. • A total recrystallized volume fraction of up to 60% can be obtained, generating DRX microstructures with mean grain sizes from 1 to 11 μm. • A rerystallized fine grained microstructure might be achieved in a two step deformation, producing high amount of deformation twins at low temperatures and high strain rates and a fully recrystallized microstructure at low strain rates in the mid temperature regime.

References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11]

A. G. Beer, M. R. Barnett, Magnesium Technology 2002, 193-198 M. M. Myshlyaev, H. J. McQueen, Mater. Sci. Eng. 2002, A337, 121-133 J. C. Tan, M. J. Tan, Mater. Sci. Eng. 2003, A339, 81-89 A. Mwembela, E. B. Konopleva, H. J. McQueen, Scripta Mater. 1997, 37, 1789-1795 F. J. Humphreys, H. Matherly, Recrystallization and Related Annealing Phenomena, Pergamon Press, Oxford, 1995, 363-392 A. Galiyev, R. Kaibyshev, G. Gottstein, Acta mater. 2001, 49, 1199-1207 H. Mecking, in Work Hardening in Tension and Fatigue (Ed.: A. W. Thompson), TMS-AIME, New York, 1977, 67 E. Poliak, J. Jonas, Acta mater. 1996, 44, 127-136 F. Förster, W. Köster, Zeitschrift für Metallkd. 1937, 4, 116-123 G. Gottstein, E. Brünger, M. Frommert, M. Goerdeler, M. Zeng, Z. Metallkd. 2003, 94, 628-635 H. Mecking, G. Gottstein in Recrystallization of Metallic Materials (Ed.: F. Haessner), Dr. Riederer Verlag, Stuttgart, 1978, 195-222

Mechanical Properties and Formability of PM Mg-Al Based Alloys K. Matsuzaki, K. Hatsukano, K.Hanada, M.Takahashi and T. Shimizu National Institute of Advanced Industrial Science and Technology(AIST) AIST-EAST, 1-2 Namiki, TSUKUBA 305-3564, Japan

1

Introduction

Mg alloys are attractive for applications where weight reduction is required because of its density which is the lowest of all construction materials. Furthermore, the advantages of the Mg alloys are an unlimited resource and good process properties as well as good recycling. To this date, most Mg alloys are produced by casting processes such as diecasting or thixomoulding and the reaming part of those are produced by using plastic forming such as forging, press. The insufficient mechanical strength and poor formability of Mg alloys mainly restrict their application area. Also, the powder metallurgical (PM) method is a useful way to produce a high strength Mg alloy except for cost and safety of handling. The major benefits of the powder metallurgy are refinement of grain, minimization of segregations, homogeneous microstructure, extend of solubility, formation of nonequilibrium phase and so on. The high strength of above 600 MPa has been reported for the MgY-Zn alloy produced by PM method [1]. For a conventional Mg alloy such as AZ91, the tensile strength is improved up to above 300MPa by PM method [2, 3], which is three times higher than that of Ingot metallurgy AZ91. The fine microstructure introduced by PM process is main reason for the improvement of the strength and is also expected to improve the formability of Mg alloy. This paper intends to describes preparation of the AZ31 alloy by hot-extrusion of gasatomization powders, and clarify their mechanical properties and formability

2

Experimental Procedure

AZ31 alloy powders were produced by gas-atomization. The ingots of AZ31 alloys were melted in a graphite crucible by an induction furnace under an argon atmosphere and atomized at a melt temperature of 1143 K into powders using a mixture gas of Ar and 0.25%O2. The ejection pressure of the melt was held at constant value of 0.3 bar and the atomization pressure was 50 bar. The obtained powder has a spherical shape with some satellite, shown in Fig.1. The powder was sieved under particle sizes of below 75 μm, followed by hot-pressed into a perform with a diameter of 39 mm and a high of 40 mm in

171 air under a pressure of 50 tonn at 753 K in 10min, and then extruded into a bar at temperatures ranging from 548 to 623 K with an extrusion ratios of 1:10 and 1:2 0. Microstructure was examined by an optical microscope and a scanning electron microscope(SEM). Hardness was measured by a Vickers micro hardness tester with a 0.49 N load. Tensile properties (ultimate tensile strength, proof stress and elongation to fracture) were obtained by a tensile test with a -4 initial strain rate of 2.6 x 10 1/s at 50Ǵ㨙 room temperature. In order to examine Fig.1. SEM Photograph of gas atomized AZ31 powders the formability, a compression test was carried out to a deformation degree of 0.6 at room temperature and temperatures ranging from 373 to 673 K at an initial strain rate of 10-4 1/s. In the test, cylindrical specimens machined from the extruded bars were used. For comparison, mechanical properties and formability of a commercial wrought AZ31 alloy with a diameter of 25 mm produced by extrusion were also examined.

3

Results and Discussion

The hot-extruded sample consists of fine equiaxed grains and shows a homogeneous microstructure. The average grain size of the sample extruded at 673 K with an extrusion ratio of 1:20 is 10 μm and slightly increases with an increase in extrusion temperature and a decrease in extrusion ratio. The X-ray diffraction analysis reveals that the basal plain in hcp-Mg of hot-extruded AZ31 powders is slightly parallel to the direction of extrusion and no significant deference in the X-ray diffraction pattern is observed for the extrusion condition. The relative density of the extruded sample reaches 98%, indicating that a fully dense structure is obtained. Table 1 summarized the tensile mechanical properties at room temperature with a Vickers hardness of hot-extruded AZ31 powders. For comparison, the properties of a commercial wrought AZ31 alloy are listed. The sample produced by extrusion at 573 K with an extrusion ratio of 1:2 0 has a yield stress of 250 MPa and a ultimate tensile strength of 340 MPa, whose values are higher than those of the wrought AZ31 alloy. The strength decreases with decreasing extrusion temperature and extrusion ratio. The hardness is in the range of 75 to 78 HV. The increase in strength is mainly due to the refinement of grain size. The elongation increased with increasing extrusion ratio and extrusion temperature, and reaches 21% for the sample extruded at 623 K with an extrusion ratio of 1:2 0. This is maybe due to theconnection between the grains becoming stronger with increasing extrusion ratio and temperature, leading to the improvement of ductility. The SEM observation revealed that the fracture surface of the sample extruded at 623 K with

172 an extrusion ratio of 1:2 0 shows a ductile feature. On the other hand, the sample extruded at 448K with an extrusion ratio of 1:10 is composed of ductile and brittle regions. Table 1. Mechanical properties of AZ31alloys produced by PM

Yeld stress (MPa)

Ultimate tensile strength (MPa)

264

340

13

78

215

300

15

75

250

340

18

78

220 175

310 260

21 15

76 50

Extrusion condition Extruion Temp. 548 K Extrusion ratio 1:10 Extrusion Temp. 573 K Extruion ratio 1:10 Extrusion Temp.573 K Extruion ratio 1:20 Extrsion Temp. 623 K Extruion ratio 1:20 Wrought alloy

Elongation to Fracture (%)

Hardness (HV)

Therefore, an increase in extrusion ratio is expected to improve the strength and ductility. It is found that the hot-extruded AZ31 powder shows high strength at room temperature compared to the wrought AZ31 alloy. In order to evaluate formability, a compression test was carried out using a round specimen with a diameter of 8mm and a height of 10 mm at room temperature and elevated temperatures ranging from 373 to 573 K. Figure 2 shows the flow curves obtained from the compression test for the extruded AZ31 powders at 573 K with an extrusion ratio of 1:20. At room temperature, the sample shows an yielding and then the flow stress increases with deformation and finally fractures at 0.18. The yield stress and fracture stress is 220 MPa and 470 MPa, respectively. With an increase in deformation temperature, yield stress and flow stress decrease, and formability increases. At above 423 K, the deformation degree exceeds 0.6. Figure 3 shows the initial and compressed samples at various temperatures. The sample deformed at RT and 373 K shows a shear failure along a plane inclined at an angle of 45 °with respect to the compression axis. At above 423 K, the samples are deformed successfully to a deformation degree of 0.6 without defect. The samples extruded at deferent temperatures with an extrusion ratio of 1:20 show a similar deformability at above 423 K. This means that the sample with an extrusion ratio of 1:20 has good formability even at a temperature as low as 423 K. For the samples with an extrusion ratio of 1:10, however, deformed samples show a distinct deferent fracture feature as shown in Fig. 4, although the similar flow curves are observed. At room temperature, shear fracture along a plane inclined at about 45° to the compression axis and the cracking parallel to the compressive axis are observed.

Nominal compressive stress (MPa)

173

600

PM AZ31

500 400

Hor-extruded 573K, 1:20

373K

423K

R.T. 473K 523K

300 200

573K

100 0

0.1

0.2

0.3

0.4

0.5

0.6

Deformation degree Fig. 2. Flow curves at various temperatures for the AZ31 alloy produced by the hot extrusion of gas atomized powders

Initial

Compressed

R.T.

373K

423K

473K

573K 10mm

Fig. 3. External appearance of uncompressed and compressed AZ31 alloy prepared by extrusion at 573 K with an extrusion ratio of 1: 20

The cracking parallel to the compressive direction takes plces for the all samples with an extrusion ratio of 1:10 and occurs during the deformation even at 673 K, indicating the hot extruded AZ31 alloy powder with an extrusion ratio of 1:10 has a poor formability. SEM micrographs shown in Fig. 5 reveals that a region of shear fracture shows a smooth surface and in the region at the crakes parallel to the compressive direction the fracture takes places at interface between particle. The smooth shear fracture surface is observed for the sample with an extrusion ratio of 1:20 and the interparticle fracture surface is observed for the sample with an extrusion ratio of 1:10. These reasons are as follows. In the present study, the AZ31alloy powders were produced by gas atomization using an Ar gas containing 0.25% oxygen as atomization gas in order to passivate the powders. Thus, the powder seems to be covered with a rather thick oxide film. The extrusion with an extrusion ratio of 1:10 is insufficient to break the oxide films on the powder surface and a strong

174

Initial

Compressed

R.T

373K

473K

523K

573K 10mm

Fig. 4. External appearance of uncompressed and compressed AZ31 alloy prepared by extrusion at 573 K with an extrusion ratio of 1:10

connection between particles is not obtained. With a higher extrusion ratio, the powder is subjected to large shear deformation during the hot extrusion, leading to the formation of a new surface and thus a strong connection between particles. Therefore, it is said that the extrusion ratio of above 1:20 is needed to produce PM AZ31 alloys with good formability. One of the other methods to obtain the formability is to perform a successive PM operation from the powder production through consolidation under a clean atmosphere [4]. In the case of a wrought AZ31 alloy, a fracture occurs at a deformation degree 50Ǵ㨙 of 0.18 at room temperature and the deformation to a degree of 0.6 without cracks is achieved at temperatures above Fig. 5. Fracture surface of AZ31 alloy produced by hot523 K. This suggests that the formability extrusion of gas atomized powders at 573 K with an of PM AZ31 alloy is superior to that of extrusion ratio of 1:10 wrought AZ31alloy. The improvement of formability is also believed to be due to the refinement of grain Figure 6 shows the change in the hardness for the deformed PM and wrought AZ31 alloys as a function of deformation temperature. The hardness of 75 HV is obtained for the PM alloy after deformed at 423 K, at which the sample has an ability to deform to above 0.6 in deformation degree without defects. By increasing deformation temperature, hardness slightly decreases but retains 70 HV at 673 K. On the other hand, the wrought alloy after deformed at above 573 K shows a hardness of 40 HV. This means that the PM

175

4

After deformed 100

No cracks

Hardness (HV)

AZ31 alloy shows a high strength even after deformations. It allows us to produce a high performance Mg parts by PM method. Finally, it is said that the AZ31 alloy produced by PM processing shows a high strength at room temperature and good formability at elevated temperatures above 423 K. Moreover, the excellent formability at temperatures as low as 423 K is expected to make them useful in near net shape forming of high strength Mg alloys such as hot forging.

50

PMAZ31 573K,1:20

No cracks

Wrought AZ3㧝alloy 0

RT 373

423

473

523

573

Deformation temperature (K㧕 Fig. 6. Change in the hardness as a function of deformation temperature for the PM and Wrought AZ31 alloy

Conclusion

The gas atomized AZ31 powders was extruded into a bar. The obtained AZ31 alloy consists of fine grain with an average grain size of 10 μm, with a high tensile strength of 340 MPa and an elongation of 18% at room temperature. The strength slightly decreases but elongation increases to 21% with increasing extrusion temperature. The sample extruded at an extrusion ratio of 1:20 has good formability at temperatures above 423 K and deforms to a deformation degree of 0.6 without fracture. On the contrary, the decrease in extrusion ratio to 1:10 causes a cracks in the cylindrical surface during compression and reduces formability. The extrusion ratio plays an important role in formability. It is concluded that a high strength AZ31 alloy with good formability is obtained by hot-extrusion of gas-atomized powders at a high extrusion ratio.

References [1] Y. Kawamura, K. Hayashi, J. Koike, A. Inoueand T. Masumoto: Mater. Sci. Forum, 2000, 350-351. [2] G. Nussbaum, P. Sainfort, G. Regazzioni and H.Gjestland: Script Metall., 23 (1989) 1079-1084. [3] C. F. Chag, S. K. Das, D. Raybould, R. L. Bye and E. V.Limoncelli: Light Metal Age (1989) p. 12. [4] A. Kato, H. Horikiri, A. Inoue and T. Masumoto: Mate. Sci. & Eng., A179/180 (1994) 707-711.

Corrosion Properties of Conversion Plasma Coated Magnesium Alloys H. M. Nykyforchyn*, W. Dietzel**, M. D. Klapkiv*, C. Blawert** * Karpenko Physico-Mechanical Institute of the National Academy of Sciences of Ukraine, Lviv, Ukraine ** GKSS-Forschungszentrum Geesthacht GmbH, Geesthacht, Germany

Abstract If a voltage of several hundred volts is applied to a metal-electrolyte system, electric discharges of an intensity of nearly 107 V/m occur which result in a plasma developing in the discharge channels leading to oxide deposition on the metal surface. The oxide-ceramic coatings thus obtained on magnesium alloys are monophase in contrast to those on aluminium alloys, and the alloying elements contained in the magnesium essentially influence the microhardness and thickness of the coatings. In the tests reported here oxideceramic coatings having a thickness of 20 to 140 μm and a microhardness of 10 GPa were produced on the surfaces of three magnesium alloys. The corrosion resistance of these coatings was investigated in acidic (pH 3.1), near neutral (pH 7.1) and alkaline (pH 11.8 and 13.4) solutions using polarisation experiments. It was found that a change in pH from acidic to alkaline conditions shifted the corrosion potential from -1200…-1500 mV to 220…-280 mV. In general, the oxide-ceramic coatings created on magnesium alloys by an electrolyte plasma reaction decreased the corrosion current density by more than one order of magnitude in chloride containing solutions but practically had no beneficial effect in case of alkaline environment.

Introduction A major factor that limits the application of magnesium alloys is their comparatively low resistance against corrosion and wear [1]. One way to overcome this problem is the application of a suitable coating. Metal coatings however are of limited applicability since due to the extreme negative electrochemical potential of the magnesium they act as a cathode with respect to the magnesium substrate in the case of cavities or surface damages. The synthesis of oxides in a low-temperature electrolytic plasma allows it to cover surfaces of magnesium and its alloys with multifunctional oxide-ceramic coatings in the same manner as has previously been shown for aluminum alloys [2]. In an attempt to identify the underlying mechanisms of this synthesis and to optimise the technological process it is essential to know the composition and the parameters of the plasma and to further clarify the role of the oxides. In previous work the plasma luminescence at high potentials, i.e. 40-400 V, was investigated, and it was shown that in the sub-sparking domain this resulted from oxide luminescence whereas in the sparking domain where current

177 densities are less than 0.5 kA/m2 the luminescence is also induced by some electrolytes and by the base-metal constituents [3-5]. On the other hand, the results of these investigations can not directly be used to determine the most appropriate plasma producing parameters and/or to elucidate the role of the oxides. Moreover, the spectra of multicomponent electrolytes are what might be described as “contaminated”, which makes an exact identification of the source of the radiation impossible.

Experimental Procedure Three magnesium alloys were investigated having the compositions Mn 0.1-0.5, Al 4-6, Zn 1.3, Li 7-10, Cd 4 (designated “IMB2”), Mn 0.1-0.4, Al 0.5-1, Zn 2.3, Li 10-11.5 (“MA18”) and Zn 0.8, Y7.1-7.9, Cd 0.63, Zr 0.5 (“BMD10”), respectively. Oxide-ceramic coatings were produced by spark and microarc discharges using a set-up which is schematically shown in Figure 1 and which in principle consists of a current source and an electrolytic bath containing the specimen. Oxide-ceramic coatings were produced at cathodic to anodic current density relations of Ic/Ia = 1 using two different electrolytes: A KOH solution containing an addition of sodium silicate (solution “N1”) and a similar solution which in addition contained a chromium compound (solution “N2”). The thickness of the coatings thus obtained was varied from 20 μm to 140 μm by varying the process time. Figure 2 shows a magnesium specimen during the coating process surrounded by a plasma in the electrolyte bath at initial, intermediate, and final stage of the process, and a coated sample.

Figure 1. Principle of set-up (schematic, left) and time dependence of the voltage applied (right) for producing oxide-ceramic coatings: 1- specimen (working electrode), 2- electrolyte, 3- bath, 4- discharge channel.

The electron number density in the plasma, ne, was determined from the broadening of the spectral lines due to the linear Stark effect. The corresponding theory takes into account the broadening, Δλ, of the spectral lines caused by a slowly varying electric field and by fast electrons [6]. This theory ensures a high accuracy of the calculations and, due to the fact that the profiles of Balmer lines can be measured relatively easily, this method is convenient for the determination of the electron number density, ne. The complete method for calculating the electron number density and temperature is described elsewhere [7].

178 The corrosion properties of the thus obtained coated samples were investigated using a potentio-dynamic test technique and a chlorine-silver reference electrode. The corrosion environments consisted of 0.3% HCl solution with pH = 3.1, a 0.3% NaCl solution with pH = 7.1, and a KOH solution with pH-values of 11.8 and 13.4.

Figure 2. Magnesium specimen surrounded by a plasma in an electrolyte bath at initial, intermediate and final stage of the coating process (top to bottom, left), and coated sample (right).

Results and Discussion The spectrum obtained from an electrolyte plasma is a superposition of the continuous spectrum of electron radiation and line or band spectra of various atoms, molecules and radicals. The continuous spectrum appears as a result of free-free (bremsstrahlung radiation) and bound-free (collision-radiative recombination of electrons) transitions. The radiation energies of bound-free transitions are in the same order of magnitude as the radiation energies of spectral lines located near the edges of the series, and correspond to highly excited states of atoms close to the ionisation energy. These states are split according to the linear Stark effect. Micro-fluctuations of the field caused by the fields of neighbouring charged particles or by an external electric field broaden the corresponding levels and lines. The line emission spectrum (bound-bound transitions) characterises the electronic transitions in atoms and ions between certain energy levels and contains information on the physical parameters of individual atoms and ions and of the plasma as a whole. The kind of the corresponding elements and their degrees of ionisation can be determined from the wavelengths of the emission lines. The intensity of the lines and their half widths specify the temperature and the number density of the electrons. The experimental data shown in Figure 3 in terms of the relative radiation intensity, J, indicate that the entire emission spectrum consists of the continuous spectrum of the electron radiation and of individual lines of atoms and ions of magnesium, hydrogen, sodium,

179

J [relative units]

oxygen and the OH-radical. The parameters of the identified lines were used to determine the number density and temperature of the electrons [8]. The number density of the plasma electrons determined from the broadening of the Hα hydrogen lines was found to be ne ≈ 3.3×1022 m-3. In previous studies of oxide synthesis in an electrolyte plasma it had been assumed that the plasma temperature lies in the interval of 1000-2300 K [9]. However, the presence of partially fused-in spots in the coating leads to the assumption that the plasma temperature must be significantly higher. An analysis of the changes in the relative intensities of the spectral lines during the formation of the coatings with various thickness suggests that the plasma temperature is indeed much higher. The electron temperature determined from the ratio of intensities of the Mg I (383.8 nm) line and the Mg I (518.3 nm), Mg I (517.2 nm) Mg II (448.1 nm) lines was found to be 104 K. The maximum difference in the temperatures obtained for these lines taking the time dependence of the electron density into account was smaller than 400K.

λ [nm] Figure 3. Emission spectrum of an electrolytic plasma at a current density of 3 kA/m . 2

The analysis of the anodic and cathodic polarisation curves obtained in the potentiodynamic tests enables the determination of the potentials and corrosion currents of the 0 0 , icor ) and with the two coatings N1 alloys under investigation in the initial state ( Ecor 1 1 2 2 ( Ecor , icor ) and N2 ( Ecor , icor ) (E2cor, i2cor), respectively. These data are shown in Table 1. The corrosion potentials measured for the magnesium alloys are rather negative. In the NaCl solution they are shifted by 145 - 315 mV to negative values compared to the HCl solution. The corrosion rates of the uncoated magnesium alloys were higher in the 0.3% NaCl solution than in the HCl environment. Here, it should be pointed out that the concentration of Cl ions is about 6 times higher in the NaCl solution. This increase in Cl ion content locally destroys the oxide films promoting the so-called negative differentialeffect and hence increases the corrosion rate. Similar data have been reported for the corrosion of pure magnesium in chloride solutions [10]. The alloy IMB2 shows the highest corrosion resistance in both environments, despite of the fact that its potential is not the lowest. This may be due to the positive effect of Li, Cd, Zn as alloying elements. The

180 higher corrosion rate of the alloy BMD10 may be attributed to the presence of the cathodic phase of the cubic intermetallic alloy Mg24Y5. Table 1. Summary of the corrosion data obtained from potentio-dynamic tests. 0 Ecor mV -1377 -1365 -1209 -1556 -1518 -1522 -795 -964 -908 -449 -339 -499

Solution Alloy 0.3% HCl BMD10 IMB2 MA18 0.3% NaCl BMD10 IMB2 MA18 KOH, BMD10 pH11.8 IMB2 MA18 KOH, BMD10 pH13.4 IMB2 MA18

1 Ecor mV -1327 -1147 -1435 -1429 -1331 -1456 -829 -229 -587 -242 -280 -226

2 Ecor mV -1294 -1283 -1283 -1490 -1444 -1506 – – – – – –

0 icor , A/m 2 0.037 0.012 0.024 0.140 0.024 0.110 0.00611 0.00135 0.00344

0.0325 0.0239

1 icor , A/m 2 0.00263 0.00017 0.00388 0.00157 0.00125 0.01200 0.00408 0.00228 0.00528 0.0254 0.0143 0.0293

2 icor , A/m 2 0.00067 0.00120 0.00392 0.00240 0.00273 0.00101 – – – – – –

Note: The indices 0, 1 and 2 refer to alloys without coating and with coatings formed in the electrolytes “N1” and “N2”, respectively.

In general, the corrosion potential is becoming more noble with the oxide-ceramic coatings, and the corrosion rates of alloys with coatings decrease by 1-2 orders of magnitude. The extent to which the current density decreases is determined by the thickness of the coatings, their porosity, and further depends on the chemical composition of the alloy. Without coating and with the coating “N1”, the alloy IMB2 shows the highest corrosion resistance in both environments. With the coating “N2” the alloy BMD10 has the best resistance in the HCl solution, whereas the alloy MA18 has the best resistance in the NaCl solution. 0 Ecor, mV

a

-400

10

3

-1

6 2

5

1

-1200

icor, A/m

-800

4

-1600 2

4

6

8

10

12

14 pH

b

1

2

3

10

-2

10

-3

10

-4

5

2

6

4

4 2

6

8

10

12 pH

Figure 4. Dependence of potential Ecor (left) and current icor (right) on pH for the alloys MA18 (1, 4), IMB2 (2, 5), and BMD10 (3, 6) without coating (1, 2, 3) and with the coating N1 (4, 5, 6)

The effect of the coating thickness and the porosity on the corrosion current density is caused by anodic dissolution at the pore tips with associated with cathodic depolarisation of hydrogen at the side faces of the pores where the negative charges are accumulated.

181 The dependence of the potential and the corrosion current on the pH of the solution is not unequivocal (Figure 4): In the chloride environment the potential decreases and the corrosion current increases with increasing pH. On the other hand, in alkaline solutions the potential increases significantly and the corrosion current decreases (at pH 11.8) and which can be attributed to passivation processes. The positive effect of the coating is vanishing in this case. However, the corrosion current decreases again for pH 13.4 in spite of the potential also further increasing.

Conclusions The procedure outlined here allows to obtain oxide-ceramic coatings with a corrosion resistance in chloride containing environments which is by 1 to 2 orders of magnitude higher than that of uncoated material. The number density of the electrons was determined 22 -3 to be ne ≈ 10 m , and the electron temperature during the synthesis of oxides in electrolyte plasma of spark discharges is Te ≈ 104 K. An increase of the chlorine ion content in the environment decreased the corrosion resistance of the coated alloys. On the other hand, a passivation of the magnesium alloys in alkaline solutions brought the corrosion currents of the uncoated materials close to the levels which were observed for the same alloys with oxide-ceramic coatings.

References [1] H. Godard, W. Jepson, H. Bothwell, R. Kane, The Corrosion of Light Metals, Wiley, N.Y. 1967, p. 311. [2] M. D. Klapkiv, H. M. Nykyforchyn, V. M. Posuvailo, Surface and Coating Technology 1998, 100-101, 219. [3] S. Ikonopisov, A. Girginov, M. Maskova, Electrochim. Acta 1989, 5, 631. [4] K. Shimizu, S. Tajima, N Baba, Electrochim. Acta 1979, 24, 35. [5] A. Günterschultze, H. Betz;, Z. Phys. 1931, 4, 681. [6] H. R. Griem, A. C. Kolb, K. Y. Shen, Phys. Rev. 1959, 116, 4. [7] M. D.Klapkiv, H. M. Nykyforchyn, V. M. Posuvailo, Mater. Sci. 1994, 3, 333. [8] H. R. Griem; Plasma Spectroscopy, McGraw-Hill, New York 1963. [9] T. B. Van, S. D Bronn, G. P. Wirtz, Ceramic Bull. 1977, 56, 563. [10] R. Tunold, H. Holtan, M.-B. H. Berge et al., Corr. Sci. 1977, 17, 353.

The effect of variations in Melt Temperature on the Grain Structures in an AM60 die Casting Hans I. Laukli1, Otto Lohne1, Lars Arnberg1, Haavard Gjestland2, Stian Sannes2 1

2

Norwegian University of Science and Technology, Trondheim; Norsk Hydro ASA, Porsgrunn

Abstract The grain structures in cold chamber high pressure die castings (HPDC) of magnesium are complex and vary with position in the casting. A fraction of the metal solidifies in the shot sleeve before injection and the remaining melt rapidly freezes in the die cavity. The final grain structures can be described as a result of three different phenomena: a) Melt solidified in the die, b) fragments of dendrites formed at the sleeve wall and c) crystals formed close to the sleeve wall during pouring and subsequent settling. The initial melt temperature and the conditions prevailing in the shot sleeve affect the contribution of each mechanism to the final grain structure. In the present work the grain structures in thin walled AM60 die castings produced at three different melt temperatures and at various shot sleeve temperatures, have been investigated. The grain morphologies are observed to depend strongly on initial melt superheat and the three different solidification mechanisms can to different extent contribute to the final grain structure.

Introduction In the cold chamber high pressure die casting process (HPDC) a significant heat loss is experienced during filling of the melt into the shot sleeve. Occasionally all superheat has dissipated prior to the injection stage [1], thus initiating solidification. Shot sleeve solidification comprises the formation of a solid skin at the shot sleeve wall and equiaxed dendritic crystals in the melt. The skin can result in cold flakes in the casting [2]. A mixture of primary crystals, termed externally solidified grains [3] or floating crystals [4], and residual melt is injected into the cavity. The relatively cold die walls and thin walled geometry ensures rapid freezing which generates a bimodal grain size distribution [4-7]. The general understanding of floating crystals have been limited to being recognized as externally solidified crystals. The mechanisms by which the crystals are formed in HPDC have been questioned [6]. A review of the different mechanisms of nucleation of equiaxed crystals in aluminium alloy castings has been reported [8], and analogous mechanisms have been suggested to pertain to the solidification of magnesium alloys [9]. Generally, the solidification of alloys initiates with crystal nucleation on suitable substrates [10] at a sufficient undercooling of the melt. These crystals can then grow into dendrites with different order branches depending on the dendrite tip radius, the local thermal gradient, solute liquid diffusion and the constitutional conditions present. When

183 the crystals start to interact, i.e. the thermosolutal diffusion fields encapsulating the crystals impinge on each other, peripheral growth is substituted by crystal coarsening [11]. The mechanisms of nucleation of equiaxed crystals in the absence of grainrefiner additives are briefly as follows: Equiaxed dendritic crystals can form during pouring of metal into a permanent mould, as a result of nucleation and growth in the thermally undercooled region adjacent to the wall [12]. These crystals grow with a necked shape due to the constraints set up by the wall restricting solute diffusion at the root. The crystals are easily separated from the wall by convection. A similar mechanism is the nucleation and growth of equiaxed crystals, termed free chill crystals, in the thermally undercooled zone adjacent to the mould walls when pouring into a cold permanent mould [13]. During growth of solid at the wall, solute is rejected at the crystal interface creating a solute enriched zone. This results in a region with lower liquidus temperature than the bulk of the liquid. Simultaneously, heat is extracted from the growing solid and a thermal gradient is established adjacent to the growing interface. A constitutionally undercooled zone is therefore generated in which suitable substrates can nucleate crystals [14]. When a layer of material solidifies at a mould wall, dendrites are rapidly formed growing into the liquid opposite and parallel to the direction of heat flow. The dendrite trunks develops branches in which segregated melt is interdendritically entrapped. These dendrites can be fragmented through remelting due to constitutional gradients, thermal conduction, solute solid diffusion or ripening phenomena [15]. The dendrite fragments then float or sink depending on the alloy constitution and convective conditions present. Another proposed mechanism is that due to heat radiation at the melt surface a thermally undercooled zone is generated adjacent to the melt surface in which dendrites can form. These crystals can be detached by breaking off [16]. A critical criterion for the survival of crystals is the magnitude of superheat in the melt. The present paper presents an investigation of the influence of melt temperature on the grain structures in thin walled AM60 box-shaped castings [17] produced in a Bühler 420ton cold chamber HPDC machine.

Experimental Approximately 15 boxes were cast to obtain a quasi-steady state temperature in the shot sleeve. These initial castings were thus produced with constant melt temperatures at the initiation of filling, at qualitatively increasing shot sleeve temperatures. At steady state a minimum of 30 castings were produced for each melt temperature with process parameters as listed in Table 1 (details in [6]). The steady state shot sleeve temperature has been estimated to be 250 °C directly below the pouring hole and 300 °C at the end [6]. Table 1. Process parameters used for producing AM60B die castings. (Tliquidus of the AM60 alloy is 619 °C o

o

-1

Tm [ C]

Tdie [ C]

Fill fraction

v2 [ms ]

P3 [bar]

640, 680, 710

200

~0.5

4.5

600

[18]

).

Samples were prepared for optical microscopy [19] taken from two positions in front of one of the gates, A) close to the gate and B) farther away, and C) in an area midway between the gates in the bottom of the box [4, 6]. Area fraction measurements of the floating crystals were performed with an equivalent circle diameter (ECD) threshold of 10 μm, derived from image analysis [20].

184

Results Effect of melt and shot sleeve temperature on the grain size distribution The micrographs in Figure 1 are obtained from a casting produced with an initial melt temperature of 640 °C. The microstructure is generally coarser closer to the gate (position A), with the primary crystals distributed over the cross section. In position B the coarse crystals are principally centered with fewer and finer floating crystals distributed from the center to the surface. A more random distribution and fewer coarse crystals are present in position C.

Figure 1. Optical micrographs of etched cross sections (2.5mm thickness) of a casting produced with low superheat, Tm = 640 °C. Micrograph A: flow vertical. B: flow horizontal. C: flow horizontal.

The measurements of area fraction as function of thickness for the three melt temperatures are displayed in Figure 2. Consistent with the micrographs in Figure 1, the area fraction of floating crystals is decreasing with distance from the gate. In position A (640 °C), the floating crystals covers 35% of the area in 50% of the cross section thickness. At 680 °C this drops to 30% in 30% of the thickness. At 710 °C the area fraction of floating crystals is approximately 10% in 10% of the section thickness. A maximum in area fraction is present in the center of position B for all temperatures. At 640 °C the area fraction is approximately 40%, at 680 °C it is 35% and at 710 °C 13%. A slight increase in area fraction is observed closer to the surface at 640oC. In position C a maximum in area fraction is observed at 640 °C, i.e. 25%. At 680 °C and 710 °C the area fraction in position C is approximately 2-3%. A C

0 Tm=640oC

10

20

30

40 50 60 [%thickness]

70

80

90 100

45 40 35 30 25 20 15 10 5 0

A B C

0 Tm=680oC

10

20

30

40 50 60 70 [%thickness]

80

90 100

45 40 35 30 25 20 15 10 5 0

A

Areafraction [%]

Areafraction [%]

B

Areafraction [%]

45 40 35 30 25 20 15 10 5 0

B C

0 Tm=710oC

10

20

30

40 50 60 [%thickness]

70

80

90 100

Figure 2. Area fraction of floating crystals in A, B and C for Tm = 640 °C (left), 680 °C (center) and 710 °C (right).

In Figure 3 micrographs from position B from a) a pre-steady state casting and b) a steady state casting are displayed. A coarser microstructure is present in Figure 3 a). a)

b)

Figure 3. Micrographs from position B. Tm = 710 °C. a) Lower Tshot sleeve and b) at a quasi-steady state Tshot sleeve.

185 Effect of superheat on the crystal morphology The castings produced at a low superheat (640 °C) revealed two distinct crystal morphologies. The floating crystals are observed to be coarse dendritic or rosette like, as observed in Figure 4 a), or globular in shape, Figure 4 b). The latter dominates in position C. At higher melt temperatures (680 and 710 °C) the crystal morphology is generally elongated dendrite trunks or arms (Figure 4 c) or branched dendrites (Figure 4 d). The branched crystals dominate in the center of the cross sections and closer to the gate.

Figure 4. Floating crystal morphologies; a) coarse dendritic/rosette like, b) globular (Tm = 640 °C), c) elongated trunks/arms, d) branched dendritic (Tm = 680 and 710 °C).

Discussion Effect of melt and shot sleeve temperature on the grain size distribution In magnesium HPDC up to 20% of the metal can be solidified in the shot sleeve prior to injection as reported in [5], and a decreasing fraction of floating crystals from the gate has been observed [21]. This is analogous to the observations in Figure 1 and in Figure 2 and has been proposed to result from the non-isothermal conditions in the shot sleeve. Supported by work on plunger movement and wave behavior in the shot sleeve, the larger fraction of coarse crystals observed close to the gate in the casting has been proposed to originate from closer to the plunger. The lower fraction of crystals observed farther from the gate has been deduced to originate from closer to the runner end [4, 6, 18]. Presolidification is most likely susceptible to shot sleeve volume and fill ratios. Measurements of the area fraction of floating crystals (with an ECD threshold of 6.7μm) close to 0.5 has been reported [5] using a small diameter shot sleeve with a fill fraction of 0.35. It can be deduced that with a low melt temperature a larger undercooling can be rapidly established and a larger number of crystals are nucleated (640 °C, Figure 2). Nucleation, growth and remelting of crystals depend on the degree of melt undercooling, and at higher melt temperatures the undercooling adjacent to the walls is reduced and fewer crystals are formed. Crystals that are nucleated later remelt due to the superheated liquid and hence fewer floating crystals are observed in the casting (Figure 2, 680 °C and 710 °C). The floating crystals are occasionally centered in the cross section thickness, e.g. position B in Figure 2, which can be explained by a solid skin is rapidly formed at the die walls during filling. The skin alters the flow behavior locally by reducing the effective cross section thickness. The melt, or mush, arriving later is forced towards the center. Thus, a larger area fraction of floating crystals is observed in the center [6, 18]. Occasionally, the skin may not be sufficiently thick to fully accommodate for the centered crystals and crystal migration can be an additional mechanism acting as proposed in [22]. Presently it is not clear if the observed plug flow is solely due to any one of these mechanisms, and work is in progress to increase the understanding of semisolid flow.

186 It has been proposed that velocity and superheat is critical for the distribution of floating crystals in the cross section thickness [6]. The microstructure in position B (Figure 1) deviate from that reported and can partially be attributed to the different filling phase velocities, i.e. 4.5 ms-1 in contrast to 6.5 ms-1 [6]. A lower velocity can result in a greater fraction of metal freezing at the die walls during filling. Hence, the crystals distributed between the surface and the center are locked in this position. This can not be observed at higher melt temperatures as fewer floating crystals survive in the shot sleeve, and the crystals that first enter the cavity are less susceptible to be deposited closer to the surface as less metal freeze at the die walls. The observed effect of decreased fraction of floating crystals with increased shot sleeve temperature can be deduced to result from the reduced number of crystals formed and the increased likelihood of remelting with superheated melt (Figure 3). To test the hypothesis that the thermal conditions in the shot sleeve affects the formation of floating crystals, a few castings were produced with Figure 5. Micrograph from position A coating in the interior of the shot sleeve. The coating was in a casting produced with coated shot sleeve. The coating procedure inter- expected to exaggerate the effects observed in Figure 3. rupted steady state and hence gener- As depicted in Figure 5, only a few floating crystals ated a slightly lower shot sleeve tem- are present in position A (close to the gate) despite that perature. Initial T = 680 °C. a longer cycle time was inevitably generated during coating. m

Effect of superheat on the crystal morphology The diversity of dendrite morphologies observed is proposed to result from the variable cooling conditions and different nucleation mechanisms that prevail in the shot sleeve. At lower melt temperatures a less inclined temperature gradient and a larger thermal undercooling is established adjacent to the shot sleeve wall in which crystals can nucleate. During filling the crystals can be formed at the wall and eventually separate according to the theory outlined in [12], or resemble the nucleation of free chill crystals in the thermally undercooled region adjacent to the shot sleeve wall [13]. The “wall crystals”, with a nearglobular shape [12], are transported into the bulk of the liquid during filling by the flow of melt. The crystals survive as the melt is sufficiently cold and branching is restricted due to lack of undercooling. At a later stage a large number of floating crystals has formed and grown to the point of impingement followed by dendrite coarsening, which results in rosette like dendrites (Figure 4 a). The first wall crystals that were formed possess a globular morphology (Figure 4 b) and are distributed closer to the runner end of the shot sleeve and therefore end up far from the gate in the casting [4, 6, 18], e.g. in position C (Figure 1, right micrograph). As the globular crystals are associated with a lower melt temperature and dominates at lower solid fractions at low superheat, e.g. position C, they are most likely not formed as a result of the high shear rates encountered in HPDC, as proposed in [5]. To obtain a globular microstructure shear is commonly applied for order of magnitudes longer times than the crystals are exposed to in HPDC [23].

187 At higher melt temperatures the temperature gradient adjacent to the shot sleeve wall during filling is steeper, and wall crystals are not expected to form as easily as the thermally undercooled region is reduced. Some crystals can form, but these are disposed to shrink and remelt as they are carried into the bulk of the superheated liquid. At a later stage crystals nucleate on the wall and grow as dendrites into the melt. Dendrites can be nucleated in the constitutionally undercooled zone at the growth front [14] or be fragmented due to dendrite arm remelting at the roots [15]. The latter mechanism is more probable as the morphology of many of the crystals resembles dendrite fragments (Figure 4 c and d). A cascade effect can be triggered as more fragmentation is initiated. The dendrite finds itself in superheated liquid, and arms are easily detached. In-situ solidification experiments, by means of X-ray imaging, have manifested the ease by which dendrite arms are fragmented during columnar dendritic growth in aluminium alloys, and convection is a major contributor in the fragmentation process [24]. Forced convection generated by the plunger movement can enhance fragmentation through dendrite arm remelting. Forced and free convection is set up by the melt flow during and after pouring and thermosolutal gradients, respectively. The crystals remain branched dendritic till injection is initiated, as an insufficient number of crystals are present to impinge on each other and generate coarsening. It has been reported that dendrite fragments results from the high shear rates encountered in the gate [5]. If this mechanism pertains, a high fraction of branched dendritic crystals would be expected in the runner system. However, a large fraction of dendrite fragments is present on either side of the gate, as observed in the micrograph in Figure 6 obtained from a position in front of the gate. Metallic dendrites are not expected to break mechanically in a casting operation, but can be plastically deformed [25]. Figure 6. Microstructure of gate-section with dendrite fragments present.

The solidification in HPDC is highly non-idealised. Large differences in grain sizes, distributions and crystal morphology can be observed, depending on position in the casting, process conditions, thermal and constitutional variations, flow behaviour, volume fraction of pre-solidified metal and alloy type. An additional aspect can be the growth of floating crystals in the cavity, which is not within the scope of the present work. However, it can be argued that a higher melt temperature would generate a coarser microstructure due to a higher fraction liquid. Analogous, a low melt temperature and larger fraction of floating crystals would generate less melt to freeze, and coarsening would not occur. However, the floating crystals are coarser in the latter case. A higher melt temperature can improve the efficiency of the intensification pressure, and pressure effects (ClaussiusClapeyron) could be introduced generating a finer grain structure.

Conclusions The initial melt temperature and the thermal conditions in the shot sleeve affect the distribution of floating crystals:

188 1. A larger area fraction of floating crystals is obtained with a low superheat and colder shot sleeve, which is attributed to the large number of floating crystals formed and the increased survival rate. 2. A lower area fraction floating crystals is obtained with a larger melt superheat or altered heat transfer conditions in the shot sleeve. This is proposed to result from fewer floating crystals are formed and fewer survive. The floating crystal morphology is affected by the melt superheat: 1. With a low melt temperature wall mechanisms are proposed to dominate. This results in floating crystals with a coarse dendritic (rosette like) or globular morphology. 2. Branched and elongated dendrites are obtained with higher melt temperatures. This is attributed to extensive dendrite fragmentation in the shot sleeve.

Acknowledgements The present work was funded by the project NorLight Shaped Castings and partners: the Norwegian Research Council, Alcoa Automotive Castings, Scandinavian Casting Center ANS, Elkem Aluminium ANS, Fundamus AS, Hydro Aluminium Metal Products, Norsk Hydro ASA, NIMR, NTNU and SINTEF (project responsible). The authors thank the partners for financial support.

References [1] Z.W. Chen, M.Z. Jahedi, Int. J. C. Met. R., 1998, 11, 129-138 [2] D. Rodrigo, M. Murray, H. Mao, J. Brevick, C. Mobley, R. Esdaile, NADCA Trans., 1999, pp. 219-225. [3] D. StJohn, A.K. Dahle, T. Abbott, M.D. Nave, M. Qian, Proc. Magnesium Techn., 2003, pp. 95-100. [4] H.I. Laukli, O. Lohne, H. Gjestland, S. Sannes, L. Arnberg, NADCA Trans., 2002, pp. T02-035. [5] W. Sequeira, PhD thesis, The University of Queensland (Brisbane), 2000. [6] H.I. Laukli, O. Lohne, S. Sannes, H. Gjestland, L. Arnberg, Int. J. C. Met. R., 2003. Submitted. [7] A. Bowles, J.R. Griffiths, C.J. Davidson, Proc. Magnesium Techn., 2001, pp. 161168. [8] J. Hutt, D. StJohn, Int. J. C. Met. R., 1998, 11, 13-22 [9] A.K. Dahle, Y.C. Lee, M.D. Nave, P.L. Schaffer, D.H. StJohn, J. Light Met., 2001, 1, 61-72 [10] L. Bäckerud, M. Johnsson, Proc. Light Met., 1996, pp. 679-685. [11] W. Kurz, D.J. Fisher, Fundamentals of Solidification, 1998, pp. 85-89. [12] A. Ohno, Solidification - The sep. theory and its practical applic., 1987, pp. 51-60, 78. [13] B. Chalmers, J. Aust. Inst. of Metals, 1963, 8, 255-263 [14] W.C. Winegard, B. Chalmers, Trans. ASM, 1954, 46, 1214-1224

189 [15] K.A. Jackson, J.D. Hunt, D.R. Uhlmann, T.P. Seward, Trans. AIME, 1966, 236, 149158 [16] R.T. Southin, Trans. AIME, 1966, 239, 220-225 [17] S. Sannes, H. Gjestland, H. Westengen, H.I. Laukli, Proc. Mg. Alloys & Appl., 2003. [18] S. Sannes, H. Gjestland, H. Westengen, H.I. Laukli, O. Lohne, SAE Trans., 2003, pp. 03M-192. [19] H.I. Laukli, O. Lohne, L. Arnberg, Poster in Proc. Mg. Alloys & Appl., 2003. [20] Adobe Photoshop, 7.0, Adobe, 2002 [21] P.D.D. Rodrigo, V. Ahuja, Proc. Int. Conf. Mag. Sci. Techn., 2000, pp. 97-104. [22] A.K. Dahle, S. Sannes, D.H. St. John, H. Westengen, J. Light Met., 2001, 1, 99-103 [23] Y. Ito, M.C. Flemings, J.A. Cornie, Rheol. & microstr. of AlSi6.5 in Nature and prop. of SSM, 1991. [24] R.H. Mathiesen, L. Arnberg, K. Ramsøskar, T. Weitkamp, C. Rau, A. Snigirev, Met. Mat. Trans. B, 2002, 33B, 613-623 [25] A. Hellawell, S. Liu, S.Z. Lu, JOM, 1997, 18-20

Microstructural Characterisation of AZ91 Magnesium Alloy Andrzej Kiełbus*, Maria Sozańska*, Lubomirc Ciżek** * Silesian University of Technology, Katowice, Poland **VSB-TU Ostrava, Czech Republic

Abstract Casting magnesium alloys make up about 85÷90% of all products in Europe fabricated with the application of Mg. The most often used alloys are Mg-Al-Zn-Mn-based ones. Magnesium alloys are widely used in the automotive industry, particularly for elements of the car interior, car body, chassis and driving gears. The basic alloy, which individual motorcar elements are made of, is AZ91. The paper presents microstructural charakterisation of AZ91 (Mg-9Al.-0,6Zn) alloy in its initial state (after casting). Melting was performed in a 250 kg induction furnace by applying modification with SPEFINAL T 200 and refining with Emgesal Flux 200. The casting temperature was 740 °C. The research methodology covered microscopic investigation as well as LM, SEM and an X-ray microanalysis.

Introduction General characterisation of magnesium alloys Magnesium and its alloys are more and more often applied in different spheres of life, including the aircraft and motor vehicle as well as metallurgical, chemical and electricalchemical industries. This is connected, among other things, with the advantageous dependence between tensile strength of 160 to 365 MPa, modulus of elasticity (45 GPa) and density (1.74 g/cm3). Magnesium alloys have a high value of tensile strength/density compared to other alloys. Moreover, magnesium shows relatively good electric and thermal conductivity as well as high damping capacity [1, 2]. The basic magnesium alloys include ones which contain manganese, aluminium, zinc, zirconium and rare-earth elements which allow obtaining suitable properties. Manganese does not cause any increase of tensile strength, however, it does slightly increase the yield point. It also brings about an increase of resistance to the action of sea water. The quantity of manganese in magnesium alloys is limited by its relatively low solubility in magnesium. Manganese content in alloys with an Al addition does not exceed 0.3% and 1.5% in alloys without Al addition. Aluminium enhances both tensile strength and hardness, and improves casting properties of an alloy. The best ratio of mechanical to plastic properties is obtained with a 6% Al content. An addition of zinc in combination with Al aims at improving tensile strength at a room temperature, however 1% of Zn with a 7÷10% Al content in an alloy enhances hot cracking. Zirconium is added to alloys which contain

191 zinc, rare-earth elements, thorium and their combinations, for the purpose of structure refinement. It should not be used in alloys containing aluminium and manganese, since it forms stable compounds with them which are removed from the solid solution. Rare-earth elements are added to manganese alloys as a mischmetal or didymium. Mischmetal contains cerium, lanthan and neodymium, whereas didymium is a mixture of neodymium and praseodymium. An addition of rare-earth elements enhances magnesium alloys’ strength at a room temperature and what is more, it reduces porosity of casts [3]. In general, magnesium alloys can be divided into two groups [1, 3]: I – those containing 2-10% Al with minute quantities of Zn and Mn. They cost of production is low and they are characterised by good corrosion resistance as well as a considerable decrease of mechanical properties with an increase of temperature. The following alloys are rated among this group: • Mg-Al-Mn alloys; • Mg-Al-Zn-Mn alloys; II – those containing a wide range of elements (mostly Mn, Zn, Th, Ag and Si instead of Al), but always with an adequate content of Zr which has a considerable influence on the increase of mechanical properties. These alloys can work at elevated temperatures, however, the price of alloying additions along with a special production technology considerably raises the cost of their fabrication. Within this group of alloys, those most often applied are: • Mg-Zr alloys; • Mg-Zn-Zr alloys; • Mg-rare-earth elements-Zr alloys; • Mg-Ag- rare-earth elements-Zr alloys; • Mg-Y- rare-earth elements -Zr alloys; Casting magnesium alloys make up about 85÷90% of all products fabricated in Europe with the application of Mg. Those most frequently applied are AZ91 and AZ31 alloys which contain 9 and 3% Al respectively, and ca. 1% Zn; next are AM50 and AM60 alloys with 5 and 6% Al contents and with a Mn addition. These alloys are characterised by good casting properties, in particular during high-pressure casting, as well as good mechanical properties. They are used for casting such structural components of cars as: wheel rims (Fig. 1a), instrument panels, steering wheels (Fig. 1b) and seat frames (Fig. 1c), as well as components of equipment in other fields, e.g. telephones (Fig. 2a), camcorders (Fig. 2b) and gardening tools (Fig. 2c). a)

b)

c)

Fig. 1. Car components made of magnesium alloys: a) Ford wheel rim of AZ91D alloy; b) steering wheel of AM50A alloy; c) seat frame of AM60B alloy.

192 The maximum solubility of aluminium in magnesium at an eutectic temperature (437 °C) is 14%, whereas an eutectic mixture (α + Mg17Al12 intermetallic phase) occurs at ca. 33% Al content (Fig. 1). The content of aluminium in all industrial alloys of Mg with Al is not higher that the boundary solubility of Al in Mg. The equilibrium structure of these alloys is characterised by 100% presence of a solid solution, whereas the unbalanced structure, additionally metastable in casting alloys, shows the presence of an eutectic already at a 2% Al content [4]. a)

b)

c)

Fig. 2. Components used in different sectors of economy, made of magnesium alloys: a) fragment of a cell phone made of AZ91D alloy; b) camcorder casing; c) garden shears made of AZ91D alloy. 700

10

0

20

Magnesium, wt% 40 50 60

30

70

80

90

660°C

650°C

L

600

500 (Al)

450°C 18.6

458°C

452°C

λ

410°C

400

Mg17Al12

437°C 62.3

88.5

69

(Mg)

ε Al3Mg2

Temperature, °C

100

300

200

100

0

Al

10

20

30

40 60 50 Magnesium, at%

70

80

90

100

Mg

Fig. 3. Mg-Al binary phase diagram [3]

Experimental procedure Material for research The object of the research was the AZ91D casting alloy intended for wheel rims, cast into a sand mould. The casting practice included melting in a 250 kg induction furnace with the application of modification with SPEFINAL T 200 as well as refining with Emgesal Flux 200. The casting temperature was 740 °C. The chemical composition of the analysed alloy is presented in Table 1.

193 Table 1. Chemical composition of AZ91D alloy in wt.-%

Mg

Al

Zn

Mn

Si

Cu

Fe

Be

Remainder

9,15

0,6

0,24

0,03

0,01

0,01

0,0001

Research methodology The microsections for structure examination were subjected to grinding with sandpaper of 250 to 1200 granulation. Next, they were mechanically polished with the use of diamond pastes and chemically, in a solution of 90 ml CH3OH + 10 ml HNO3. For etching, two reagents were used with the following chemical compositions: –10 ml fluoric acid (48%) + 90 ml water; –2 ml fluoric acid + 2 ml nitric acid + 96 ml water. The observation was performed with: • REICHERT light microscope of MeF2 type; the results of the observation were recorded with a NIKON digital apparatus, COOLPIX 990 model; • HITACHI S-4200 scanning microscope with a cold cathode, equipped with an Xradiation detector EDS-VOYAGER of NORAN INSTRUMENTS. Research results As a result of the microscopic examination performed, it was found out that the casting AZ91D alloy is characterised by a solid solution structure α with α+β eutectic and β phase (Mg17Al12) at grain boundaries (Figs 4÷6). The eutectic obtained assumes the form of a so-called abnormal eutectic, where the initiating particles dissolved in the solid solution precipitate during the casting process at the boundaries of the initiating solid solution. Furthermore, the occurrence of Laves’ phase in the form of Mg2Si was proved as well as in the form of precipitations of probably a MnAl4 phase (Fig. 6). The Mg2Si phase has a light blue colour and is characterised by a regular, multilateral shape with smooth edges. The precipitations containing manganese and aluminium are characterised by an irregular shape with a rough surface; they very often take the form of spines. The closer the precipitation surface, the higher the ratio of Al to Mn content. The microanalysis results of the chemical composition of individual phases are shown in Table 2 (Figs 5, 6). a)

b) α+β

Mg2Si

Mg17Al12

α

Fig. 4. Microstructure of the AZ91 alloy cast: solid solution α + eutectic (α+β) + intermetallic phase Mg17Al12 and Mg2Si precipitations a) LM 150×; b) LM 400×;

194

a)

b)

c)

Fig. 5. Microstructure of AZ91 alloy. α + (α + β) + β a) SEM magn. 1500×; b) spectrum of α-solution chemical composition c) spectrum of β-phase chemical composition - Mg17Al12 Table 2. Chemical composition of identified AZ91 alloy phases

Phase

Chemical element [at.- %]

α phase β phase Mg2Si phase MnAl4 phase

Mg 89,41 63,39 62,39 1,37

Al 9,75 35,65 78,17

Zn 0,84 0,69 -

Si 37,61 -

Mn 20,47

Conclusions The research conducted was aimed at determining the microstructure of the casting AZ91D magnesium alloy which was cast into a sand mould. It has been shown that the analysed alloy has a solid solution structure α with an abnormal α + β eutectic formed as a result of precipitation, during casting, of β phase particles, with the β phase dissolved primarily in the solid solution; the alloy is also characterised by massive precipitations of β phase (Mg17Al12) at the solid solution grain boundaries. In addition, precipitations (light blue) of Mg2Si phase were observed, characterised by a regular, multilateral shape with smooth edges. Some acicular precipitations, of an uneven surface, of MnAl4 phase were found as well.

195

a)

b)

c)

d)

Fig. 6. Precipitations of Mg2Si and AlMn4 phases in the AZ91 alloy matrix a) SEM 500×, b) SEM 1000× c) spectrum of Mg2S phase chemical composition d) spectrum of AlMn4.phase chemical composition

References [1] L. Cizek et al.: ”Structure and properties of the selected magnesium alloys.”, 10th International Scientific Conference “Achievements in Mechanical & Materials Engineering”, Zakopane 2001. [2] A. Luo: ”Magnesium: current and potential automotive applications.”, Journal of Materials, February 2002. [3] Avedesian M., Baker H.: „Magnesium and Magnesium Alloys.”, ASM Speciality Handbook, 1999 [4] Dahle K., A., Lee Y. C., Nave M., D., Schaffer P., L., St Johnson D.H.: „Development of the as-cast microstructure in magnesium-aluminum alloys.”, Journal of Light Metals 1, 2001, 61-73

Microstructural Changes of AZ91 Magnesium Alloy After Heat Treatment Andrzej Kiełbus*, Lubomir Ciżek**, Libor Pawlica** * Silesian University of Technology, Katowice, Poland **VSB-TU Ostrava, Czech Republic

Abstract Mg-Al-Zn based alloys are the most commonly used magnesium alloys for structural components. AZ91 alloy is a widely used casting magnesium alloy. The paper presents microstructural characterisation of AZ91 (Mg-9Al.-0,6Zn) alloy after casting and heat treatment. The casting temperature was 740°C and heat treatment was performed at 415 °C/1÷48 h with aging at 165 °C for 8h. The microstructure of the casting alloy consists of α-Mg phase matrix with a primary brittle β phase (Mg17Al12) at grain boundaries. We also observed precipitations of Mg2Si and MnAl4 phases. The microstructural changes of this alloy have been investigated using light microscopy.

Introduction History and properties of magnesium alloys [1÷3] Magnesium alloys belong to the lightest structural alloys. They are characterised by low density: ~1,8 g/cm3, tensile strength Rm = 300÷350 MPa, elongation A5 = 20% and hardness ~100HB. Mainly for these reasons, magnesium alloys have a widespread application in the motor vehicle and aircraft industries as well as in household appliances and office machines. In the past, magnesium alloys played a significant role. During the First and Second World Wars they were used in the nuclear, metal and military aircraft industries. After the Second World War, the interest in magnesium alloys fell considerably. Only in the 90 s, about 17 kg components made of magnesium alloys were used for the production of VW Beetle. Requirements connected with the fumes emission limitation necessitated renewed growth of interest in those alloys. In 1994, magnesium consumption reached 228 thousand tons and in 1998, already 360 thousand tons, at 3.6 $ for one kilogram. It is forecast that within the nearest 10 years the consumption of magnesium alloys will grow at a 7% rate a year. As all materials, magnesium alloys have a number of advantages and disadvantages. Their advantages include first of all: low density, relatively high specific strength, good castability, very good for pressure die casting, good weldability in a controlled atmosphere,good corrosion resistance. The most important disadvantages are: low Young’s modulus, limited possibilities of cold workability, relatively low hardness, limited strength and creep resistance at elevated temperatures, great shrinkage during solidification, high chemical reactivity,in some cases, limited corrosion resistance. Improvement of the properties of magnesium alloys by conventional techniques (e.g. heat treatment) is not possible. The only solution is to apply reinforcing fibres, i.e. to produce a composite.

197 Heat treatment of magnesium alloys[1, 4] Magnesium alloys are subjected to heat treatment mostly for the purpose of improvement of their mechanical properties or as an intermediary operation, to prepare the alloy to other specific treatment processes. The type of heat treatment depends on the chemical composition of the alloy, its form (casting or after plastic working) and on the anticipated service parameters. Solution heat treatment of magnesium alloys enhances their strength, with maximum ductility and resistance to dynamic loads. Aging of solution heat treated alloys allows obtaining maximum hardness and yield point, with a decrease of ductility, whereas aging treatment without prior solution heat treatment, as well as annealing, result in a decrease of casting stresses and partial increase of mechanical properties when under tension. Annealing significantly decreases the mechanical properties and causes improvement of plastic properties, thus facilitating further treatment. A change of the heat treatment basic parameters has an influence on a change of the properties. For instance, extension of the annealing time considerably raises the yield point (with a decrease of ductility). The basic magnesium alloy heat treatment processes and their designations are presented in Table 1. Table 1. Designation of basic processes of magnesium alloys’ heat treatment

Designation O T2 T3 T4 T5 T6 T7 T8 T9 T10

Process Recrystallizing – only alloys for plastic working Annealing – only casting alloys Solution treatment and cold plastic working Solution treatment Aging treatment Solution treatment and aging Solution treatment and stress relief annealing Solution treatment, cold plastic working, aging Solution treatment, aging, cold plastic working Aging treatment, cold plastic working

The most frequently applied heat treatment processes for magnesium alloys are solution treatment and aging treatment. During solution treatment of Mg-Al-Zn alloys, the components subjected to treatment should be inserted in a furnace at ca. 260 °C and then, slowly heated with furnace to the solution treatment temperature to avoid melting the eutectic components and thus, to minimize voids formation. The heating rate from 260 °C to the solution treatment temperature depends on the size and chemical composition of the component subject to treatment as well as on its weight and differences in thickness of individual fragments of the cast. Most often, however the heating time is 2 hours. Some components, made of other magnesium alloys, can be inserted in the furnace at the solution treatment temperature. In the cases when the microstructure is incorrect or when a cast has been annealed too much during slow cooling after solution treatment, reheat treating is required. Majority of magnesium alloys subjected to reheat treating do not show an excessive grain growth. In the case of Mg-Al-Zn alloys, repeated solution treatment should not last longer than 30 minutes. After solution treatment, magnesium alloys are usually cooled in the open air. In some cases, however, other cooling agents are used, such as: water, glycol or oil. The two latter are mostly used to reduce casting stresses. The main problems which occur during and after heat treatment include: oxidation, formation of fusion voids, warping, grain growth

198 and incorrect properties. Oxidation occurs in the case of heat treatment without the application of neutral or protective atmosphere. It may lead to local weakening of the component treated, and even to burning of metal in the furnace. Voids are formed when inappropriate rate of heating from 260 to 370 °C is applied for Mg-Al-Zn alloys, as well as in consequence of exceeding the solution treatment temperature required for these alloys and for alloys with Zn and rare-earth elements additions. The phases situated at grain boundaries relocate along those boundaries and create long, narrow void regions. This process is usually connected with the growth of grain size. Warping occurs when the component subjected to treatment has no support or as a result of its non-uniform heating during heat treatment. The grain size growth usually occurs in AM100A, AZ81A, AZ91C and AZ92A alloys towards the end of solution treatment. The disadvantageous properties of a component subject to treatment occur as a result of too low or too high process temperature, bad circulation, too slow cooling after solution treatment or inappropriate time of holding thick components.

Experimental procedure Material for the research The object of the research was an AZ91D alloy designed for car wheel rims, cast into a sand mould. The chemical composition of the analysed alloy is shown in Table 2. Table 2. Chemical composition of the AZ91D alloy in wt-%.

Mg Rest

Al 9,15

Zn 0,6

Mn 0,24

Si 0,03

Cu 0,01

Fe 0,01

Be 0,0001

The casting technology included melting in a 250 kg induction furnace with the application of modification with SPEFINAL T 200 as well as refining with Emgesal Flux 200. The casting temperature was 740 °C [5]. After casting, heat treatment (solution and aging treatment) was performed. In order to protect the specimens against oxidation, the process was conducted in the atmosphere. The solution and aging treatment parameters are presented in Table 3. Table 3. Parameters of the heat treatment applied.

Designation O 1 2 3 4 5

Solution treatment Temperature Time As-cast 415 18 415 18 415 18 415 18 415 18

Cooling

Aging treatment Temperature Time

Cooling

air water furnace air water

168 168

air air

8 8

Research results As-cast alloy The casting AZ91D alloy is characterized by a solid solution structure α with eutectic α + β and β (Mg17Al12) phase at grain boundaries. Moreover, the occurrence of Laves’ phase in the form of Mg2Si and precipitations, probably of MnAl4 phase, has been proved (Fig. 1). The eutectic regions constitute ca. 24% and hardness of this material is 53HV.

199

a)

b)

Fig. 1. Microstructure of the as-cast AZ91 alloy – solid solution α + (α + β) eutectic + Mg17Al12 intermetallic phase and Mg2Si precipitations. LM 150×; b) LM 400×

Solution treatment 415 °C/18 h/air After solution treatment followed by air cooling, numerous precipitations were found in the alloy structure, of β (Mg17Al12) phase and MnAl4 phase in the α solid solution matrix. No occurrence of eutectic regions (Fig. 2) was detected. The alloy hardness (54 HV) did not change compared to the as-cast condition, and the mean area of the solid solution 2 grain was 1505 μm .

a)

b)

Fig. 2. Microstructure of the AZ91 alloy after solution treatment 415 °C/18h/air – a) LM 150×; b) LM 400×;

Solution treatment 415 °C/18 h/water After solution treatment followed by cooling in the water, similarly to air cooling, precipitations were found in the alloy structure of β (Mg17Al12) phase and MnAl4 phase in the α solid solution matrix (Fig. 3). No occurrence of eutectic regions was detected. The alloy hardness was the same as in the as-cast condition, i.e. 54HV. Solution treatment 415 °C/18 h/furnace After solution treatment followed by cooling in the furnace, the alloy is characterized by α solid solution structure with α + β eutectic regions; few precipitations of β (Mg17Al12) phase were observed at grain boundaries. Some precipitations of Mg2Si (Fig.4) phase were observed, as well. The structure of this material resembles that of the as-cast alloy, however, hardness after cooling with the furnace increases up to 65 HV; also, the number of eutectic regions increases (to ~36%) comparing to the as-cast condition.

200

a)

b)

Fig. 3. Microstructure of the AZ91 alloy after solution treatment 415 °C/18 h/water a) LM 150×; b) LM 400×;

a)

b)

Fig. 4. Microstructure of the AZ91 alloy after holding at 415 °C/18 h/furnace a) LM 150×; b) LM 400×;

Solution treatment 415 °C/18 h/air + aging 168 °C/18 h/air The aging treatment applied after solution treatment with air cooling caused β phase precipitation in the form of pseudoeutectic regions. Furthermore, the β phase (Mg17Al12) precipitated at solid solution grain boundaries (Fig. 5). Hardness of this material increased to 60 HV compared to that after solution treatment, whereas the solid solution area was simi2 lar and amounted 1532 μm .

a)

b)

Fig. 5. Microstructure of the AZ91 alloy after solution treatment 415 °C/18 h/air + aging 168 °C/18 h/air a)LM 150×; b) LM 400×;

201 Solution treatment 415 °C/18 h/water + aging 168 °C/18 h/air Aging treatment of an alloy solution treated and cooled in water caused precipitation of a larger number of β phase in the form of pseudoeutectic regions compared to the condition after air cooling (Fig. 6). Hardness of the AZ91 alloy after water solution treatment and aging is higher that that of the alloy after air solution treatment and aging, and amounts 65 HV, with the mean area reduced to 1065 μm2.

a)

b)

Fig. 6. Microstructure of the AZ91 alloy after solution treatment 415 °C/18 h/water+aging 168 °C/18h/air a)LM 150×; b) LM 400×;

Conclusions The subject of the research conducted was an evaluation of the influence of the heat treatment process on the microstructure and hardness of the AZ91 alloy. It was found that immediately after casting, an AZ91D alloy has an α solid solution structure with α+β eutectic and β (Mg17Al12) phase at grain boundaries. After solution treatment (both in the water and in the air), a reduction of pseudoetectic regions was observed as well as of the number of β phase precipitations. Hardness of the investigated as-cast and solution treated alloy remained unchanged, amounting 54 HV. Application of aging treatment caused precipitation of β phase along the solid solution grain boundaries and in the form of pseudoeutectic regions, which enhanced the increase of the alloy hardness.

References [1] Avedesian M., Baker H.: „Magnesium and Magnesium Alloys.”, ASM Speciality Handbook, 1999 [2] L. Cizek et al.: “Structure and properties of the selected magnesium alloys.”, 10th International Scientific Conference “Achievements in Mechanical & Materials Engineering”, Zakopane 2001. [3] A. Luo: “Magnesium: current and potential automotive applications.”, Journal of Materials, February 2002. [4] Stratton P.F., Chamg E.K.: “Protective atmospheres for the heat treatment of magnesium alloys. [5] Pawlica L., Čížek L., Greger M., Pavelcová J., Kiełbus A.: “Study of the formability of magnesium alloy AZ91.”, XI Sem.naukowe pt.: Nowe technologie i materiały w metalurgii i inżynierii materiałowej, Katowice, 2003.

Microstructure of Ultra Fine Grained Mg and Mg-10 wt.% Gd Prepared by High-Pressure Torsion J. Cizek1), I. Prochazka1), I. Stulikova1), 2), B. Smola1), 2), R. Kuzel1), V. Cherkaska1), R.K. Islamgaliev3), O. Kulyasova3) 1)

Charles University, Faculty of Mathematics and Physics, Prague, Czech Republic Zentrum für Funktionswerkstoffe GmbH, Clausthal-Zellerfeld, Germany 3) Institute of Physics of Advanced Materials, Ufa State Aviation Technical University, Ufa, Russia 2)

1

Introduction

Applications of Mg-based alloys at elevated temperatures are limited by the low melting point of Mg. This difficulty can be at least partially overcome by an addition of rare earth elements. It has been shown that Mg-based alloys with rare earth elements can be designed for operating temperatures above 300 °C [1, 2]. Mg-Gd represents one of the promising systems for novel Mg-based hardenable alloys with high creep resistance at elevated temperatures [3]. The precipitation of a metastable c-based centered orthorhombic β’ phase in Mg-Gd alloys with Gd concentration above 10 wt.% results in considerable age hardening [3]. Material hardening can be achieved also by decrease of the mean grain size. The grain boundaries represent obstacles for dislocation motion. It makes plastic deformation more difficult and increases material hardness. The yield stress is inversely proportional to the mean grain size through the well-known Hall-Petch relation. Recently it has been demonstrated that ultra fine grained (UFG) metals with grain size around 100 nm and no residual porosity can be produced by severe plastic deformation (SPD) [4]. A number of preparation techniques based on SPD have been developed so far. Among them, high-pressure torsion (HPT) and equal channel angular pressing (ECAP) are used most frequently. The smallest grain sizes have been achieved by HPT. The UFG materials prepared by HPT exhibit only weak texture and homogeneous structure, so that they are convenient for experimental studies. The UFG samples prepared by ECAP exhibit usually larger grain sizes than HPT made specimens. On the other hand, ECAP allows for a production of massive UFG samples, which makes it suitable for possible future industrial applications. A number of UFG metals exhibit favorable mechanical properties which consist in combination of very high strength and significant ductility as was demonstrated e.g. on HPT deformed Ti [5]. For this reason, it is highly interesting to examine microstructure and physical properties of UFG Mg-based light alloys. Following this purpose, microstructure investigations of UFG Mg and UFG Mg-Gd alloy were performed in the present work. The extraordinary properties of UFG materials are closely related with defects (grain boundaries, dislocations) introduced by SPD. Positron lifetime (PL) spectroscopy is a well-established non-destructive technique with high sensitivity to open volume defects [6]. It enables identification of the defect types present in the material studied and determination of defect densities [6]. Thus, PL spectroscopy represents an ideal tool for defect studies of UFG materials. It has been successfully employed in microstructure investigations of UFG Cu [7] and UFG Ni [8]. In the present work PL spectroscopy was combined with X-ray diffraction (XRD), microhardness measurements, and direct observations of microstructure by TEM.

203

2

Experimental

The samples of as-received technical purity Mg (99.95%) were investigated. In addition, another set of Mg samples was annealed at 280 °C for 30 min and then slowly cooled down inside the furnace. It led to the complete recovery of defects. Therefore, the annealed Mg was used as a defect-free reference sample. The Mg-10 wt.% Gd alloy was prepared from the technical purity Mg by squeeze casting. The as-cast material was homogenized at 500 °C for 6 hours. The homogenization annealing was finished by rapid quenching into water of room temperature. The UFG samples were prepared from the as-received technical purity Mg and the homogenized Mg-10 wt.%vGd alloy. The UFG microstructure was achieved by HPT [5]. The initial coarse-grained samples were deformed at room temperature by torsion up to the true logarithmic strain ε = 7 [5]. At the same time the samples were compressed with high pressure of 6 GPa. The UFG samples were disk shaped with diameter around 11 mm and thickness of 0.5 mm. A fast-fast spectrometer similar to that described in [9] was employed in the present work. 22 The timing resolution of the spectrometer was 170 ps (FWHM Na) at the coincidence count -1 rate 130 s . The PL spectra were decomposed into exponential components using a maximum likelihood procedure [10]. The PL measurements were accompanied by theoretical calculations of positron lifetimes using the atomic superposition method (ATSUP) [11]. Observations of microstructure were performed on the JEOL 2000 FX electron microscope operating at 200 kV. Thin foils for TEM were electropolished in a twin-jet device TENUPOL-2. X-ray studies were carried out with the aid of XRD7 and HZG4(Seifert-FPM) powder diffractometers using Cu Kα radiation. The XRD profiles were fitted with the Pearson VII function by the program DIFPATAN [12]. The microhardness was measured by the Vickers method at load of 100 g applied for 10 s using the LECO M-400-A hardness tester.

3

Results and Discussion

3.1

Coarse-Grained Samples

The reference Mg sample annealed at 280 °C for 30 min exhibits only single-component PL spectrum with lifetime τ1 = 225.3 ± 0.5 ps, see Table 1. It means the all positrons annihilate from the free state in this sample. The measured lifetime τ1 exhibits reasonable agreement with the theoretically calculated Mg bulk positron lifetime τb = 233.2 ps, see Table 2. Table 1. Lifetimes and corresponding relative intensities of the exponential components resolved in PL spectra of the studied samples. The errors (one standard deviations) are given in parentheses.

Sample

τ1 (ps)

I1 (%)

τ2 (ps)

I2 (%)

Reference Mg As-cast Mg Mg-10 wt.% Gd HPT Mg HPT Mg-10 wt.% Gd

225.3(5) 204(4) 220(4) 188(5) 210(3)

100 63(1) 90.9(6) 32.4(5) 34(2)

275(1) 301(2) 277(2) 256(3)

37(1) 9.1(7) 67.6(5) 66(2)

204 a)

b)

Fig.1: Bright-field TEM images of Mg: a) as-received sample b) HPT deformed.

A bright-field TEM image of the initial coarse-grained Mg specimen is shown in Fig. 1a. The mean grain size was found to be about 10 μm. Randomly distributed dislocations were observed inside the grains. The dislocation density ρ ≈ 5×1012 m-2 was estimated by TEM. The PL spectrum of this specimen is well fitted by two exponential components. Lifetimes and relative intensities of these components are listed in Table 1. The lifetime τ1 lies below the bulk positron lifetime for Mg. It means that it comes from free positrons. On the other hand, the second component with the lifetime τ2 represents a contribution of positrons trapped at defects. In view of the TEM observations, we consider that positrons are trapped at dislocations. The as-received Mg exhibits microhardness HV = 440 ± 40 MPa. Table 2. Positron lifetimes calculated using ATSUP method.

τ (ps) Mg – bulk Mg – monovacancy Mg – divacancy Mg – monovacancy associated with Gd atom

233.2 296.6 315.7 295.3

A bright-field TEM image of the homogenized Mg-10 wt.% Gd alloy is shown in Fig. 2a. Large coarse grains were observed. A low dislocation density below ≈ 1012 m-2 was found by TEM. The PL spectrum of this specimen is well fitted by two exponential components given in Table 1. The first component with the lifetime τ1 < τb can be attributed to free positrons, while the second one with the lifetime τ2 comes from positrons trapped at defects. The low dislocation density in the specimen approaches the lower sensitivity limit of PL spectroscopy [6]. Therefore, positrons trapped at dislocations cannot represent a noticeable contribution to PL spectrum. The second component with the lifetime τ2 represents a contribution of positrons trapped in quenched-in excess vacancies, which were ”frozen” in the sample due to the rapid quenching. This interpretation is supported by the lifetime τ2 ≈ 300 ps, which is remarkably higher than the lifetime 275 ps of po-

205 sitrons trapped in Mg-dislocations, and agrees well with the calculated lifetime of positrons trapped in Mg monovacancy given in Table 2. Monovacancies in pure Mg become mobile below room temperature [13]. Thus, free Mg monovacancies are not stable at room temperature and are quickly annealed out. Contrary to it, we observed that positrons are trapped in the monovacancies in the homogenized Mg-10 wt.% Gd sample. The explanation is that the monovacancies are bound to Gd atoms. The vacancy-Gd pairs are stable at room temperature. One can see from Table 2 that the lifetime of positrons trapped in Mg-monovacancy practically does not differ from the lifetime of positrons trapped in a vacancy-Gd pair. Thus, a free monovacancy and a monovacancy-Gd pair cannot be distinguished by PL spectroscopy. Nevertheless, an enhanced amount of Gd in the vicinity of vacancies was proved by coincidence Doppler broadening measurements [14]. Microhardness HV = 680 ± 50 MPa was measured on the homogenized Mg-10 wt.% Gdsample. It is in reasonable agreement with the hardness measured on this alloy after homogenization in [39]. a)

b)

Fig.2: Bright-field TEM image of Mg-10wt.%Gd alloy: a) homogenized 500 oC/6h b) HPT deformed.

3.2

UFG Specimens

A bright-field TEM image and an electron diffraction pattern of the HPT deformed Mg specimen are shown in Fig. 1b. Two different kinds of regions were observed: i) “deformed regions” with UFG grains (100-300 nm) and high dislocation density, and ii) “recovered regions” with substantially larger grains (≈ 1 μm) and practically free of dislocations. The “recovered regions” indicate a dynamic recovery of the microstructure during the HPT processing. However, the recovery is incomplete. The presence of the “deformed regions” with UFG grains is demonstrated also by the X-ray back-reflection pattern, which consists of continuous diffraction rings. The PL spectrum of the HPT deformed Mg was decomposed into two exponential components, see Table 1. The first component with the lifetime τ1 comes from free positrons. The lifetime τ2 of the second component corresponds well to the lifetime of positrons trapped at dislocations. Hence, in connection with the TEM results, we can conclude that positrons are trapped at dislocations inside the “deformed regions”. It should be pointed out that dislocations in the “deformed regions” are distributed relatively homogeneously, i.e. they are present in the grain interiors as well as close to the grain boundaries. It is an important difference of the HPT deformed Mg

206 from the HPT deformed Cu and Ni, where dislocations are piled up along grain boundaries, while the grain interiors remain almost dislocation-free [7,8]. The determined microhardness 600 ± 40 MPa is by 40% higher than that of the as-received Mg. A bright-field TEM image and an electron diffraction pattern for the HPT processed Mg-10 wt.% Gd alloy are shown in Fig. 2b. Contrary to the HPT deformed Mg, the Mg10 wt.% Gd has uniform UFG microstructure with the mean grain size about 100 nm. It indicates no dynamical recovery during the HPT processing. The electron diffraction pattern shows high-angle missorientation of neighboring grains. High density of homogeneously distributed dislocations was observed. The PL spectrum of the HPT deformed Mg-10 wt.% Gd alloy consists of two exponential components, see Table 1. The first component with the lifetime τ1 comes from free positrons. The lifetime τ2 of the second component is by ≈ 20 ps shorter than the lifetime of positrons trapped at Mg-dislocations. It indicates a slightly different nature of defects in the HPT deformed Mg-10 wt.% Gd. It is likely due to a segregation of Gd atoms along dislocations [14]. The specimen broke at 250-300 MPa during tensile test at room temperature without noticeable ductility. Only homogenized Mg-10 wt.% Gd, on the other hand, had ductility of 5% and the strength of 158 MPa [15]. The microhardness increases with the radial distance r being 1700 ± 300 MPa at the center of the specimen and 2300 ± 300 MPa at the margin. Similar behavior of microhardness HV was observed also in HPT deformed Cu [16] and it seems to be typical for a number of UFG metals prepared by HPT. HV reflects most probably the increase of strain with r. The mean value of HV of the HPT deformed Mg-10 wt. %Gd alloy is by 230% higher than that for the homogenized Mg-10 wt. % Gd sample. No microvoids were found in both HPT deformed specimens. It is another important difference compared to HPT deformed Cu [7] and Ni [8]. X-ray diffraction patterns of HPT deformed Mg and the Mg-10 wt.% Gd showed texture of the (00l) type which could be well described by the empirical formula Ihkl ~ Rhkl exp (-D sin 2ψ), with the angle ψ between the texture axis and particular reflection plane (Ihkl is measured integrated intensity of the corresponding diffraction profile and Rhkl is its theoretical intensity for the random grain distribution). The values of D = 5 and D = 2 for Mg and Mg-10 wt.% Gd specimens, respectively were found. For Mg-10 wt.% Gd alloy the lower broadening of (00l) profiles with respect to other peaks indicate dominating presence of 〈a〉 dislocations with Burgers vector b = 1/3.a. [ 2 1 10 ]. No significant broadening for pure Mg sample was detected so that the dislocation density should be less than about 1 × 1013 m-2. However, major contribution to the diffraction peaks comes from the “recovered regions” (see above) so that the “deformed regions” cannot be well characterized by XRD.

4

Conclusions

Microstructure of HPT deformed Mg and Mg-10 wt.% Gd was characterized. Two kinds of regions were found in the HPT deformed Mg: “recovered regions” with coarse grains and low dislocation density and “deformed regions” with UFG grains and high dislocation density. The HPT made Mg-10 wt.% Gd exhibits homogeneous UFG microstructure with high density of uniformly distributed dislocations and grain size about 100 nm. No micro-

207 voids were found. Both HPT deformed specimens exhibit significantly increased microhardness.

5

Acknowledgement

The work was financially supported by The Czech Grant Agency (contract 106/01/D049), The Grant Agency of Charles University (contract 187/2001), and by the DFG, Germany.

6

References

[1] G. Sigaudo in Proceedings of 3rd International Magnesium Conference (Ed.: G.W. Lorimer), The Institute of Materials, Cambridge 1997, p. 137. [2] D. Eliezer, E. Aghion F.H. Froes, in Magnesium 97 (Eds.: E. Aghion, D. Eliezer),Magnesium Research Institute, Ltd., Beersheva 1998, p. 343. [3] P. Vostry, B. Smola, I. Stulikova, F. von Buch, B.L. Mordike, phys. stat. sol. (a) 1999, 175, 491-500. [4] R.Z. Valiev, R.K. Islamgaliev, I.V. Alexandrov, Prog. Mat. Sci. 2000, 45, 103. [5] A.A. Popov, I.Y. Pyshmintsev, S.L. Demakov, A.G. Illarionov, T.C. Lowe, A.V. Sergeeva, R.Z. Valiev, Scripta Mater 1997, 37, 1089. [6] P. Hautojärvi, C. Corbel, in Proceedings of the International School of Physics “Enrico Fermi”, Course CXXV, (Ed. A. Dupasquier, A.P. Mills), IOS Press, Varena 1995, p. 491. [7] J. Cizek, I. Prochazka, M. Cieslar, R. Kuzel, J. Kuriplach, F. Chmelik, I. Stulikova, F. Becvar, R.K. Islamgaliev, Phys. Rev. B 2002, 65, 094106. [8] J. Cizek, I. Prochazka, M. Cieslar, I. Stulikova, F. Chmelik, R.K. Islamgaliev, phys. stat. sol. (a) 2002, 191, 391-408. [9] F. Becvar, J. Cizek, L. Lestak, I. Novotny, I. Prochazka, F. Sebesta, Nucl. Instr. Meth. A 2000, 443, 557. [10] I. Prochazka, I. Novotny, F. Becvar, Mat. Sci. Forum 1997, 255-257, 772-774. [11] M.J. Puska, R.N. Nieminen, J. Phys. F: Met. Phys. 1983, 13, 333-346. [12] R. Kuzel, DIFPATAN - program for powder pattern analysis, 1995 (http://www.xray.cz/priv/kuzel/difpatan). [13] M. Fahnle, B. Meyer, J. Mayer, J.S. Oehrens, G. Bester, in Diffusion Mechanisms in Crystalline Materials (Ed.: Y. Mishin et al.) MRS Symposia Proceedings No. 527, p. 23. [14] J. Cizek, I. Prochazka, F. Becvar, I. Stulikova, B. Smola, R. Kuzel, V. Cherkaska, R.K. Islamgaliev, O. Kulyasova, to be submitted. [15] F. von Buch, PhD. thesis, TU Clausthal, Clausthal-Zellerfeld, 1999, p. 186. [16] J. Cizek, I. Prochazka, G. Brauer, W.Anwand, R. Kuzel, M. Cieslar, R.K. Islamgaliev, phys. stat. sol. (a) 2003, 195, 335-349.

Development of New Grain Structure and Tensile Properties Improving in a Hot Pressed and ECAP Processed ZK60 Magnesium Alloy A. Galiyev, R. Kaibyshev, M. Almakaev, Institute for Metals Superplasticity Problems, Ufa

1

Introduction

Recently, it has been reported that fine-grained microstructure can be attained in Mg alloys by equal-channel angular pressing (ECAP) [1-3]. The mechanical properties of these Mg products could be significantly improved at room and elevated temperatures. For example, ECAP processed Mg alloys showed enhanced room temperature ductility and superplasticity at a low temperature and high strain rate [4-6]. To date, however, very limited data are available for relationship between ECAP processing conditions and the formed microstructures of Mg alloys as well as mechanical properties of ECAP processed specimens. No attempt was made to evaluate dependence of tensile properties of Mg alloys upon procedure of ECAP although it is well known that the microstructure and properties of materials produced by ECAP are dependent upon the processing route [7-10]. The present investigation was initiated to examine the influence of the processing route on the microstructure and mechanical properties of ZK60 Mg alloy.

2

Experimental

A ZK60 (Mg-5.8 %Zn-0.65 %Zr) alloy used in the present study was received as a commercially extruded bar of 90 mm in diameter. Bars for ECAP were machined from the extruded bar with diameters of 20 mm and lengths 80 mm. ECAP was conducted at 523K using a die having a channel angle of 90°. Samples were pressed repeatedly to 1, 2, 4 and 8 passes, using processing routes A, BC or C, in which the samples was rotated by 0°, 90° and 180°, respectively, in the same direction after each pass. The discs of 5 mm in thickness were sliced from the samples subjected to ECAP. Microstructures of the ECAP processed specimens were observed on the X plane where this plane was defined as lying perpendicular to the direction of pressing [11]. Tensile specimens were machined from the as pressed samples with guage sections of 2 × 3 × 6 mm3 and with the guage lengths lying parallel to the longitudinal axes. Tensile tests were carried out in air at room temperature at an initial strain rate 2.8×10-3 s-1 and elevated temperature 523 K with initial strain rates from 2.8 × 10-4 to 1.4 × 10-3 s-1.

209

3

Results and Discussion

3.1

Microstructures after ECAP

Typical microstructures are shown in Fig. 1 for the Mg alloy in the as extruded state and after ECAP, using the route A. In the as extruded state the microstructure was very inhomogeneous with the volume fraction of recrystallized grains of ~40% and the sizes of recrystallized grains scattered in the range of 2-20 μm. ECAP resulted in grain refinement and increase of recrystallized volume (Fig. 2a). The recrystallized grain size reduced significantly to 1-4 μm after four passes. It can be seen in Fig. 2a that recrystallization occurred more rapidly for routes BC and C, where the volume fraction of recrystallized grains increased to almost 90% after two passes, while for route A the recrystallized volume

Figure 1. Microstructures of (a) as extruded sample and samples after ECAP using route A for: (b) one pass, (c) two passes, (d, e) four passes and (f) eight passes.

210 80

Route BC

90

Route A

80

Route C

70 60 50 40 30

(a) 0

2

4

6

Recrystallized Volume [%]

Recrystallized Volume [%]

100

Route C 70

Route BC

50 40 30

8

Route A

60

Number of Pressings

(b) 0

2

4

6

Number of Pressings

8

Route BC

30

Route A

Route C 20 10

(c) 0

0

2

4

6

Number of Pressings

Recrystallized Volume [%]

Recrystallized Volume [%]

100

40

8

80 60 Fine-grained fraction Coarse-grained fraction Total volume

40

2 passes 4 passes

20 0

(d) 0

90

180

Specimen Rotation [°]

Figure 2. Changes in recrystallized volume with processing route and number of pressings: (a) total recrystallized volume, (b) fraction of coarse-grained recrystallized structure, (c) fraction of fine-grained recrystallized structure, (d) effect of the processing route for two and four passes.

90% after two passes, while for route A the recrystallized volume of ~85% was attained only after four passes. Inspection of Figs. 1c-f showed that regions of fine grains with size of ~1 μm appeared in the recrystallized microstructure after two passes for all processing routes, however the volume of these regions was strongly depended on route used and number of pressings (Fig. 2c, d). This was the reason of decreasing the average grain size after two and four passes for routes BC and C in comparison with route A (Fig. 3). The average grain size after four passes was ~3.1, 1.6 and 2.1 μm for routes A, BC and C, respectively. It should be noted that route BC was more effective in attaining finer grain size. 3.2

Tensile properties at room temperature

Stress-strain curves of the extruded sample, 1-pass ECAP sample and 4-pass ECAP samples after processing through routes A, BC and C are shown in Fig. 4. The extruded sample exhibited higher 0.2% proof stress and ultimate tensile strength compared to the ECAP specimens. However, ECAP specimens exhibited much larger elongation. This is because of the fact that the basal plane is inclined to the extrusion direction in the ECAP specimen and as result the ductility was improved [4].

211 10

Route A

Route C

Stress [MPa]

Grain Size [μm]

300

200 as extruded 1 pass 4 passes - route A 4 passes - route BC 4 passes - route C

100

Route BC 1

0

2

4

6

Number of Pressings

0 0.0

8

Figure 3. Variation of the grain size as a function of number of pressings for samples pressed using routes A, BC and C.

0.1

0.2

0.3

Strain Figure 4. Stress-strain curves of as extruded sample and samples after ECAP using routes A, BC and C.

The values of the 0.2% proof stress, ultimate tensile strength and elongation to failure are summarized in Fig. 5. It can be seen that all three samples exhibited reasonably similar values of the 0.2% proof stress and ultimate tensile strength, but elongation to failure was higher after processing through route BC. It is concluded that a microstructure formed through route BC is more effective in increasing the elongation to failure. 3.3

Tensile properties at elevated temperature

Figure 6 shows the typical stress-strain curves at 523 K and an initial strain rate of 5.6 × 10-4 s-1 of the extruded sample, 1-pass ECAP sample and 4-pass ECAP samples after processing through routes A, BC and C. It is apparent from Fig. 6 that the extruded sample and 1-pass ECAP sample with coarser-grained microstructures exhibited much higher (a)

35

Elongation to Failure [%]

Tensile Properties [MPa]

360 320

Ultimate Tensile Strength

280

Route A Route BC Route C

240 0.2% Proof Stress

200 0

1

2

3

Number of Pressings

4

(b)

30 25 20 15

Route A Route BC Route C

10 5

0

1

2

3

4

Number of Pressings

Figure 5. Variation of (a) the 0.2% proof stress, ultimate tensile strength and (b) elongation to failure as a function of number of pressings for samples pressed using routes A, BC and C.

212 70

40

Strain Rate Sensitivity

50

Stress [MPa]

0.50

as extruded 1 pass 4 passes - route A 4 passes - route BC 4 passes - route C

60

30 20 10 0

0

1

2

3

4

5

6

0.45 0.40 0.35 0.30

1 pass

0.25 0.20

as extruded

0.15

7

Route A Route BC Route C

4 passes

0.0004

0.0008

0.0012 -1

Strain

Strain Rate [s ]

Figure 6. Stress-strain curves at 523 K and an 5.6 × 10-4 s-1 of as extruded sample and samples after ECAP using routes A, BC and C.

Figure 7. Variation of the strain rate sensitivity as a function of strain rate at 523 K for as extruded sample and samples pressed using routes A, BC and C.

stresses than the 4-pass ECAP samples with the fine-grained microstructure. The elongation increased with increasing number of pressings and it was due to grain refinement attained through ECAP. This trend was confirmed in Fig. 7, where the values of strain rate sensitivity, m, increased essentially from ~0.23 for as extruded sample to ~0.37 for the 1-pass ECAP sample and to the range of 0.42-0.46 for the 4-pass ECAP samples that was a direct consequence of grain refinement introduced by ECAP. As a result significant elongations were achieved in the 4-pass ECAP samples. On the other hand, the differences in the superplastic characteristics were found in samples processed by routes A, BC and C (Fig. 8). Three samples showed m values of ~0.42, 0.43 and 0.46 and elongations of ~480, 570 and 680%, respectively. This investigation demonstrates that the superplastic ductility was a maximum when using route C with a rotation of the sample by 180° between each pass through the ECAP die. This result is not

Strain Rate Sensitivity

0.40 0.35 0.30

Route A Route BC Route C

0.25 0.20

0

1

2

3

Number of Pressings

4

Elongation to Failure [%]

700

0.45

600 500 400 300 Route A Route BC Route C

200 100 0

0

1

2

3

4

Number of Pressings

Figure 8. Variation of (a) the strain rate sensitivity at 523 K and (b) elongation to failure at 523 K and an -4 -1 5.6 × 10 s as a function of number of pressings for samples pressed using routes A, BC and C.

213 consistent with the earlier investigations for aluminum and aluminum-based alloys [9, 10] showing route BC was the optimum processing route for achieving maximum ductility. Apparently, peculiarities of the crystallographic slip of magnesium alloys lead to attainment of optimum microstructure and superplastic ductility when using processing route C. Further investigations are required for evaluation of the route influence on texture and properties of ECAP processed magnesium-based alloys.

5

Conclusions

(1) The grain size of an Mg-5.8%Zn-0.65%Zr alloy was reduced to ~3.1, 1.6 and 2.1 μm using the process of ECAP by routes A, BC and C, respectively, for 4 passes at 523 K. (2) Temperature room ductility was a maximum when using processing route BC in which the sample was rotated by 90° in the same sense between each consecutive pressing through the ECAP die. (3) Optimal superplastic ductility was achieved after pressing through route C with a rotation of the sample by 180° between each pass through the ECAP die: for example, elongations of ~480, 570 and 680% were obtained at 523 K and an initial strain rate of 5.6 × 10-4 s-1 in the samples processed by routes A, BC and C, respectively.

6

References

[1] Mabuchi, M.; Iwasaki, H.; Yanase, K.; Higashi K., Scripta Mater. 1997, 36, 681-686. [2] Mabuchi, M.; Ameyama, K.; Iwasaki, H.; Higashi K., Acta Mater. 1999, 47, 20472057. [3] Yamashita, A.; Horita, Z.; Langdon, T.G. Mater. Sci. Eng. 2001, A300, 142-147. [4] Yoshida, Y.; Cisar, L.; Kamado, S.; Kojima, Y., Mater. Trans., 2003, 44, 468-475. [5] Watanabe, H.; Mukai, T.; Ishikawa, K.; Higashi, K., Mater. Sci. Forum 2003, 419422, 557-562. [6] Matsubara, K.; Miyahara, Y.; Makii, K.; Horita, Z.; Langdon, T. G., Mater. Sci. Forum 2003, 419-422, 497-502. [7] Iwahashi, Y.; Horita, Z.; Nemoto, M.; Langdon, T. G., Acta Mater. 1997, 45, 47334741. [8] Ferrasse, S.; Segal, V. M.; Hartwig, K. T.; Goforth R. E., Metall. Mater. Trans. 1997, 28A, 1047-1057. [9] Oh-ishi, K.; Horita, Z.; Furukawa, M.; Nemoto, M.; Langdon, T. G., Metall. Mater. Trans. 1998, 29A, 2011-2013. [10] Komura, S.; Furukawa M.; Horita, Z.; Nemoto, M.; Langdon, T. G., Mater. Sci. Eng. 2001, A297, 111-118. [11] Furukawa, M.; Iwahashi, Y.; Horita, Z.; Nemoto, M.; Langdon, T. G., Mater. Sci. Eng. 1998, A257, 328-332.

Texture Formation and Texture Modelling of AZ31 Magnesium Wrought Alloy A. Styczynski [1], Ch. Hartig [1], R. Bormann [1], F. Kaiser [2], J. Bohlen [2], D. Letzig [2] [1] Arbeitsbereich Werkstoffphysik und –Technologie, Technische Universität Hamburg-Harburg, 21071 Hamburg, Germany [2] GKSS-Forschungszentrum Geesthacht GmbH, 21502 Geesthacht, Germany

1

Abstract

The texture formation during cold rolling of AZ31 magnesium wrought alloy has been investigated. Two initial microstructures – 1. rolled and recrystallized, 2. cast – were considered. A typical tilting of maxima by ± 15° in the (0002) pole figures was found to be characteristic for the texture after cold rolling of both examined microstructures. The measured texture evolution and plastic anisotropy (yield stresses and r-values) can be explained by a modified Taylor model assuming activity of the basal, prismatic and pyramidal slip systems and of the {0 1 12} < 0 1 11 > twinning system.

2

Introduction

The insufficient room temperature formability of magnesium wrought alloys is commonly ascribed to their anisotropic plastic behaviour [1, 2] and makes processing steps like rolling, extrusion etc. difficult, thus limiting their application for structural parts. The slip activities and critical resolved shear stresses (CRSS) in hcp magnesium have been reported for different slip and twinning systems [3-5]. According to these experimental data, the CRSS of the basal slip mode {0001} < 01 12 > is much lower than those of the non basal slip and twinning modes. It can be shown that a fulfilment of the von Mises condition requires at least the activation of one hard slip mode with c-component burgers vectors (i.e. pyramidal slip or {0 1 12} < 0 1 11 > twinning) [2,6]. The texture evolution and the plastic anisotropy, as obtained by experiments and model calculations, can be considered as a tool for the investigation of slip mode activity and for the verification of the CRSS’s in an anisotropic polycrystal. The origin of the cold rolling texture of hcp metals has been widely considered in literature [6-8]. It was assumed that for pure hcp metals the texture evolution depends strongly on the c/a ratio. For magnesium with a c/a = 1.622, near to the ideal value of 1.633, a basal texture was commonly observed [8]. The basal, prismatic and pyramidal slip modes with Burgers vector as well as the pyramidal slip mode with Burgers vector and also the twinning system {0 1 12} < 0 1 11 > are recognised in literature to be active deformation modes in pure magnesium [2-8]. However, this crude approach is not applicable in magnesium alloys. Due to variations in chemical composition, the strengths of slip and twinning systems are altered and hence changes in texture evolution occur.

215

3

Experimental Procedures and Results

Three samples of the AZ31 magnesium alloy were investigated corresponding to different initial microstructures: a commercial recrystallized sheet (A), a squeeze cast bar (B) and a commercial rolled sheet (C). From materials A and B specimens were prepared for cold rolling experiments. The dimensions of each sample amounted to 30 mm width × 6 mm thick × 50 mm length. The initial microstructures of the commercial sheet and of the squeeze cast bar are shown in Fig. 1a and Fig. 1b. The grain size was about 60 μm for the commercial recrystallized material (A) and about 350 μm for the squeeze cast bar (B). The volume fraction of the soft secondary phase Al17Mg12 amounted to less than 5% for both materials.

Fig. 1: Initial microstructures of commercial recrystallized sheet A (a) and squeeze cast bar B (b); microstructures after rolling: recrystallized sheet (c) and squeeze cast bar (d).

The rolling was carried out at room temperature on a two-high mill with a roll diameter of 200 mm. The thickness reduction for each step was fixed to a true strain ϕ = 0.1 (strain rate 0.7 s-1 < ϕ& < 1.5 s-1) and rolling was performed up to a total true strain ϕ = 0.51 for the recrystallized sheet and ϕ = 0.50 for the squeeze cast material, respectively, before the specimen began to fracture. The microstructures (Fig. 1c,d) after rolling show a high density of deformation twins, a typical feature of heavily cold deformed AZ31 alloy [9]. The textures of the rolled sheets were measured on a D8 DISCOVER X-ray Diffractometer (BRUKER AXS Inc.) using reflection geometry and Cu-Kα radiation. For each sample the {10 10}, {0002}, {10 11},{10 12},{1120},{10 13} and {1122} pole figures were recorded. The orientation distribution function (ODF) and complete pole figures were calculated using the harmonic method of Bunge [10]. The results of texture measurements are shown in Fig. 2.

216

Fig. 2: Initial textures (a: sheet A, b: cast material B) and textures after rolling (c: material A, d: material B)

The initial textures can be characterized as a strong basal texture for the recrystallized sheet and as a random texture with scattered, strong reflections from large grains for the squeeze cast bar (Fig. 2a,b). The texture after cold rolling can be completely characterized by a symmetrical tilting of the {0002} basal poles by ±15° towards the rolling direction (RD) (Fig. 2c,d). The tilting is more clearly visible for the cold rolled recrystallized sheet material than for the squeeze cast bar with initially nearly random texture. The {1120} pole figures give evidence for an almost perfect fibre texture: There are no distinct maxima visible at a polar angle α = 90° (Fig. 2cd). In terms of an ODF this texture can be described as two fibre components along the ϕ2 Euler angle (I.: ϕ1Φϕ2 = +90°,15°, ϕ2 ; II.: ϕ1Φϕ2 = −90°, 15°, ϕ2) being equivalent because of the orthorhombic rolling specimen symmetry. This fibre texture is clearly visible also in the squeeze cast material with a random initial texture. Thus it can be concluded that the basal pole tilting is merely caused by deformation mechanisms, not by the initial texture. For the measurement of the plastic anisotropy another typical commercial rolled sheet of AZ31 alloy (specimen C), 1.6 mm thick, and finally tempered under the H24 condition has been chosen for investigations. The texture of this rolled sheet was found to be almost equal with the texture of specimen A after cold rolling (c.f. Fig. 2c). The anisotropy parameter r was measured at room temperature by tensile tests of samples cut from the sheet at different angles β in the range between 0° and 90° to the rolling direction. Thereby the parameter r, defined as the ratio r = εW/εN, is obtained as a function of β, where εW is the plastic strain into the transverse direction of the tensile sample and εN the strain into the normal direction of the sheet. Results of this measurement, being presented in detail in [11], are compared in chapter 4 with theoretical calculations.

217

4

Texture Modelling and Plastic Anisotropy

Theoretical textures were calculated on the basis of a viscoplastic Taylor model [12] by usage of the LApp [13] software package. Two different initial textures: A random distribution, generated by a random generator, for the squeeze cast material and an orientation distribution, calculated from the experimental ODF before rolling, for the experimental recrystallized sheet were taken as input data. The strain rate sensitivity exponent m of the critical resolved shear stress (CRSS) was set to m = 19 for every slip system. The simulations were performed up to a total deformation of ϕ = 0.5 without strain hardening and with full Taylor constraints. A systematic analysis of possible texture types was performed taking into account all triplet

Fig. 3: Results of texture simulations for different combinations of slip and twinning systems (c.f. Table 1). A random initial texture was supposed and the total deformation grade was set to ϕ =0.5.

combinations of slip and twinning modes (c.f. Table 1) and a meaningful range of CRSS’s. Details of this analysis are described in [2] and representative results of the texture modelling are presented in Fig. 3 in form of {0002} pole figures. A typical maximum intensity splitting in the {0002} pole figures as in the experimental texture (Fig. 2) appears exclusively for the triplet combinations A, C and E (Fig. 3), allowing the slip mode, whereas it does not appear for the triplet combinations B and D, forbidding slip. This gives clear evidence that the activity of the pyramidal slip system is responsible for the basal splitting. The C combination (with the CRSS’s: τc( basal) = 1.0, τc( pyramidal2) = 3.0 and τc(twinning) = 2.0) is the only triplet combination, which does not exhibit any {0002} pole intensities along the rolling direction, as in the experimental textures. Thus, at a first glance this combination may be considered being active during cold rolling of AZ31. For a further verification of the mode combination C the theoretical yield locus and the rvalues of the tensile tested commercial sheet (c.f. chapter 3) were determined, taking into account the measured texture of this material. The same viscoplastic Taylor model as for the

218 Table 1: Combinations of slip and twinning systems used for theoretical texture calculations

designation A B C D E CE

slip and twinning combination b {0001} < 2 1 10 > + py1 {01 11} < 2 1 10 > + py2 {12 12} < 12 13 > b {0001} < 2 1 10 > +py1 {01 11} < 2 1 10 > + tw {01 12} < 01 1 1 > b {0001} < 2 1 10 > + py2 {12 12} < 12 13 > + tw {01 12} < 01 1 1 > b {0001} < 2 1 10 > + pr {01 10} < 2 1 10 > + tw {01 12} < 01 1 1 > b {0001} < 2 1 10 > + pr {01 10} < 2 1 10 > + py2 {12 12} < 12 13 > b {0001} < 2 1 10 > + pr + {01 10} < 2 1 10 > py2 + {12 12} < 12 13 > + tw {01 12} < 01 1 1 >

calculations of the theoretical texture (s.a.) was used, with identical viscoplastic parameters and CRSS’s . In Fig. 4a the yield locus for this case is shown, assuming a plane stress state (σ33 = 0). From this curve it must be concluded that combination C cannot fulfil the necessary experimental conditions for positive values of slopes on the yield locus curve at σRD = 0 and σTD = 0, which are required for positive values of the anisotropy parameter r. Such a behaviour is also expected because of the strong basal texture and the lack of slip modes accommodating plastic strains in the transverse direction.

Fig. 4 (a) Yield loci for commercial AZ31 sheet for slip and twinning combinations C and CE. (b) Comparison of the theoretical plane anisotropy parameter r (calculated by LApp [12]) with Experiment (r-values taken for 1% strain) as function of the angle β (∠ RD, Tensile direction).

Therefore the mode combination C was modified by addition of the prism slip mode (with τc( prismatic) = 3.6) resulting in mode combination CE. The yield locus for this case, also shown in Fig. 4a, gives a good accordance with experimental results: The slopes at σRD = 0 and σTD = 0 as well as the yield stresses in rolling and transverse direction (c.f. Table 2) agree reasonably well with experimental results.. Also the theoretical rolling texture for the mode combination CE fits with the measured textures (Fig. 3). Furthermore, the theoretical r-values as function of the azimuthal angle β coincident quite well with experimental measured values for a strain of 1%, as can be seen from Fig. 4b. It has to be pointed out that experimental r-values of this material can become much greater (r = 1.2-2.7 ) for a strain of 10% [11]. A reasonable cause for this behaviour is given by the anisotropic strain hardening of the different slip modes and may be explained within the scope of advanced texture models. The good agreement of textures and values for the plastic anisotropy (yield stresses and rvalues) obtained from theory with experimental results gives a strong evidence for the activity of the four slip and twinning modes: basal, prismatic, pyramidal (c + a)

219 ( {12 12} < 12 13 > ) and twinning ( {01 12} < 01 1 1 > ). This result was obtained by a calculation based on the Taylor model, giving some confidence to a “compatibility first” model. Table 2: Theoretical and Measured yield stresses for rolled commercial sheet C

Angle with rolling direction (Degrees)

Measured Yield Stress (MPa)

Theoretical Yield Stress for Combination CE

0 45 90

219 240 263

6.6 6.7 6.9

5

Conclusions

Theoretical texture predictions from a Taylor model give a good qualitative agreement with experiment. The results are quite similar with results obtained by more advanced models [6]. The yield locus and r-values of a commercial AZ31 sheet, as calculated after a viscoplastic Taylor model agree quite well with experimental values for low plastic strains obtained from tensile tests.

References [1] Kaiser F., Letzig D., Bohlen J., Styczynski A., Hartig Ch., Kainer K.U. Materials Science Forum 2003, 419-422, p. 315 [2] Styczynski A., Hartig Ch., Bohlen J., Letzig D., Scripta Mat., submitted [3] Kelley E. W., Hosford W. F., Transactions of AIME 1968, 24, p. 654 [4] Stohr J. F., Poirier J. P., Philos. Mag. 1972; 25: p. 1313 [5] Ando S., Tonda H., Materials Science Forum 2000, 350-351, p. 43 [6] Agnew S. R., Yoo M. H., Tomé C. N. Acta Mater 2001, 49, p. 4277 [7] Agnew S. R., Proc. Magnesium Technology 2002, ed. by H.I. Kaplan, TMS, 2002; 169 [8] Rollet A. D., Wright S. I., Texture and Anisotropy (Ed.: U. F. Kocks, C. N. Tomé , H. R. Wenk), Cambridge University Press, 1998, Chapter 5 [9] Lach E., Kainer K. U., Bohmann A., Scharf M., Magnesium Alloys and their Applications (Ed.: K. U. Kainer), Wiley-VCH, 2000, 354 [10] Bunge H.-J., Texture Analysis in Material Science, Cuvillier Verlag Göttingen; 1993 [11] Kaiser F., Letzig D., Bohlen J., Styczynski A., Hartig Ch., Kainer K. U., Materials Science Forum, 2003, 419-422, p. 315 [12] Aernoudt E., van Houtte P., Leffers T., Plastic Deformation and Fracture of Materials (Ed.: H. Mughrabi), VCH-Verlag, Weinheim, 1992, Chapter 3 [13] Kocks U. F., Kallend J. S., Wenk H.-R., Rollet A. D, Wright S. I.: Preferred Orientation Package – Los Alamos, 1994

Texture Development of AM20 Tensile Samples Under Load S.B. Yi1,2, H.-G. Brokmeier1,2, K.U. Kainer2, T. Lippmann2 1 2

Institute of Materials Engineering, Technical University Clausthal, Clausthal-Zellerfeld, Germany Institute of Materials Engineering, GKSS-Research Center, Geesthacht, Germany

1

Introduction

The potential applications of Mg and its alloys are rapidly increased especially in automobile industries. However, one of the great problems for industrial applications is the low formability and the high directional anisotropy of these alloys, which are caused by their crystallographic hexagonal structure. For filling up the von Mises criterion of 5 independent slip systems the activation of only basal slip is not sufficient, so that the profound investigations on easier and/or on harder conditions for the activation of other deformation systems are necessary for getting better mechanical properties. Since the crystallographic texture relates directly with the mechanical anisotropy and the activity of different deformation systems, many researchers are making effort to manifest the relationship between the texture and the deformation behavior under various loading conditions. The frequent difficulty in the interpretation of the texture development in Mg-alloys was the ambiguity in the combination of the active deformation modes [1, 2]. The present study on the in-situ texture variations during tension was carried out to contribute to a basic understanding of the relationship between the crystallographic texture and the activation of various deformation systems. Because texture reflects the deformation history and deformation mechanisms, the in-situ texture measurements give us meaningful information to the influence of initial texture on the activation of deformation systems and on the mechanical anisotropy of deformed materials. The great potential of synchrotron radiation is based on the high photon flux and the excellent brilliance. In the case of texture measurements a set of complete pole figures can be obtained in relatively short counting time when using an area detector. Thus synchrotron radiation is ideal for in-situ experiments. Hard X-rays have a penetration depth of 2.3 cm for Mg in the case of 100 keV photon energy. For realizing different initial textures, tensile samples were cut in three different directions from an extruded bar. The results of this study can be also applied to determine the texture variation of different industrial products because each tensile sample has the characteristic texture components which are often found in industrial products.

2

Experiments

The initial material was a rectangular extruded AM20 bar with a dimension of 10 × 90 mm2 in cross-section. The microstructure of the bar consists of equi-axed grains of 4.3 μm mean size. Round type tensile samples (DIN 50125) were cut in 0°, 45° and 90° to the extrusion direction to take into account the different initial textures, with 6 mm in diameter and 24 mm in gauge length. Figure 1 shows the (0002) pole figure of the initial bar and of the cutted tensile samples. As showed in figure 1 each tensile sample stands for a

221 L

(a)

(b)

(c)

(d)

T

Figure 1. The (0 0 0 2) pole figures of the rectangular extruded bar (a). The pole figure coordinate system is changed by cutting tensile samples. (b) stands for 0°, (c) for 45° and (d) for 90° tensile sample. The capital letters ‘L’ and ‘T’ on figure (a) indicate the extrusion and transverse direction, respectively. (intensity level = 1,2,3,4, Pmax = 4.99)

characteristic initial texture. The main loading axis of the samples corresponds always to the center of the pole figure. A universal testing machine (UTM), by which a sample can be loaded up to 20kN, was installed at beamline BW5, Hasylab at DESY (Germany) [3]. Figure 2 presents a schematic view of the beam line with the UTM set-up. The tensile test was carried out at room temperature with an initial strain rate of 6 × 10-4 sec-1. For pole figure measurements during loading the sample is rotated with the UTM around the omega-axis. The omega-rotation with this set-up can reach 145° because the loading shafts obstruct the incident and dif2 fracted beams. The monochromatic incident beam with a size of 1×1 mm and a wavelength of 0.12 Å was irradiated to the sample and the diffracted beam, Debye-Scherrer cone, was registered with an image plate area detector type MAR345. The image plate was positioned perpendicular to the beam. In order to get complete pole figures a set of image plate exposures were taken at every 5° in omega. Data evaluation for complete pole figures were taken from corresponded Debye-Scherrer cones after background correction using integral intensities. The orientation distribution functions (ODFs) were calculated using the iterative series expansion method with positivity correction. The (1 0 -1 0), (0 0 0 2), (1 0 -1 1) and (1 1 -2 0) pole figures were used for calculating the ODF to a degree given by Lmax = 19 [4]. 6

1

3

7

4 5

2

Axis for omega-rotation

1. Monochromator 2. Hard X-rays (0.12Å) 3. Incident slit (I) 4. Counter diode 5. Incident slit (II) 6. UTM with sample 7. Area detector

Figure 2. Schematic view of the beam line set-up with UTM installed in Hasy-Lab.

3

Results and Discussion

The flow curves during uniaxial tension show certainly that each sample has different deformation behavior (Fig. 3). The stress fall-offs on the flow curves coincided with pole

222 2

1

3

4 4

3

0°

90° 45°

2 1 1

2

3

4

Figure 3. The stress-strain curves of each specimen during in-situ loading, the stress fall-offs on curves correspond to the pole figure measuring positions. The marked points with numbers are present in this study.

figure measurements which were caused by mechanical problems. The texture measuring points at different strains correspond to the position after the yield phenomena (point 1), during strain hardening (point 2) and quasi-steady state of plastic deformation (points 3 and 4). 90° and 45° samples show low yield strengths of about 160 MPa which testify the activation of tensile twins and {0 0 0 2} basal slip with lower CRSS. The 90° sample shows higher strain hardening rate in the plastic deformation range than other samples. The 45° sample has the largest fracture strain at about 25.1 %. On the other hand, the specimen cut at 0° has the highest yield strength of 244 MPa because the activation of basal slip is restricted for this crystallographic direction. These specific mechanical behaviors can be explained by the different activities of deformation modes which relate directly with the initial texture of the tensile samples. Most of the basal planes were oriented perpendicular to loading axis in 0° sample, where Ini. (Fmax =11.6 ) P. 1 (Fmax =15.7 ) P. 2 (Fmax =16.4 ) P. 3 (Fmax =16.6 ) P. 4 (Fmax =16.6 ) o

ϕ2 = 0

ϕ2 = 30o

Figure 4. ϕ2 = 0o, 30o ODF-sections of the 0° sample, initial state and in-situ measured. The measuring points correspond to the numbers on the Fig. 3. (intensity level = 1,3,5,7,9,11,13,15)

223 Ini. (Fmax =12.1 ) P. 1 (Fmax =12.8 ) P. 2 (Fmax =10.9 ) P. 3 (Fmax =10.3 ) P. 4 (Fmax =12.0 ) o

ϕ2 = 0

ϕ2 = 30o

Figure 5. ϕ2 = 0°, 30° ODF-sections of the 90° sample, initial state and in-situ measured. The measuring points correspond to the numbers on the Fig. 3. (intensity level = 1,3,5,7,9,11)

the 45° sample has no basal plane perpendicular to the tensile axis but at 45° inclined from the tensile axis. Considering slip on the basal plane the Schmid’s factor of 45° sample is higher than in the case of 0°, therefore, this sample has the lowest yield strength. However the low yield strength of the 90° sample can not be explained by Schmid’s law. This sample had the basal planes mostly parallel to the loading axis, and this arrangement of polycrystals is very favorable for the activation of {1 0 -1 2} tensile twinning. Because the twin boundaries act as barrier against the sliding of slips during further deformation, this sample shows the highest stain hardening behavior. Figure 4 presents ϕ2 = 0° and ϕ2 = 30° ODF-sections of the initial 0° sample and those of in-situ measurements during tensile test. The initial texture is composed of {-1 2 -1 0} (Euler angle {ϕ1, Φ, ϕ2} = 90, 90, 0) and a strong texture-fibre between {0 1 -1 0} (0, 90, 30) and {0 1 -1 0} (90, 90, 30). It should be mentioned that this type of texture is common for round or bar extruded Mg. After the yield phenomena the {-1 2 -1 0} component in the initial sample disappears, and a slight increment of {0 1 -1 0} component appears (point 2 in Figure 4). During further deformation this sample shows only a slight change of texture intensity. It is well known that the texture component in which is parallel to the loading axis is stable during uniaxial deformation [5]. The strengthening of parallel to the loading axis generally relates with the activation of -slip systems. The ODFs of the 90° sample are shown in figure 5. The texture of the initial 90° sample is for the most part composed of c-axis parallel and perpendicular to tensile direction, which is mainly found on industrially extruded tubes and rectangular bars [6]. Severe changes of the texture are observed after a small amount of plastic deformation, the texture components with c-axis parallel to tensile direction become rapidly weak and {0 1 -1 0} component is newly developed. The fiber-component between {-1 2 -1 0} and {0 0 0 1} has a maximum at {-1 2 -1 3} (90,45,0). This rapid change of the texture occurs by the activation of twinning, which impose large rotations to the crystals. By increasing deformation, the {0 1 -1 0} component becomes stronger and the {1 2 -1 0} component develops with a consumption of the fiber-component. Another change is the development of the {0 1 -1 0} (60,90,30) component.

224 Since the 45° sample has a monoclinic symmetry the ODF, this sample is represented in the Euler space of ϕ1 = 0° ~ 360°, Φ = 0° ~ 90° and ϕ2 = 0° ~ 60°. Figure 6 shows the ϕ2 = 0° and 30° ODF-sections during tension. The initial texture of this sample is composed of two weak Initial ( Fmax = 12.3 )

Position. 1 (Fmax = 14.5 )

Position 2 (Fmax = 14.0 )

Position 3 (Fmax = 13.8 )

ϕ2 = 0o

ϕ2 = 30o

o

ϕ2 = 0

ϕ2 = 30o

Position 4 (Fmax = 13.8 ) o

ϕ2 = 0

ϕ2 = 30o

Figure 6. ϕ2 = 0 , 30 ODF-sections of 45° sample, initial state and in-situ measured. The measuring points correspond to the numbers on the Fig. 3. (intensity level = 1,3,5,7,9,11,13) o

o

components, {-1 2 -1 0}, {0 1 -1 0}, and two strong components, {-1 2 -1 2} (315, 55 , 0), {0 1 -1 2} (270, 45 , 30). This type of texture can be characterized by the basal planes and pyramidal planes which are inclined 45° from the loading axis. As referred to the flow curve of the 45° sample, this crystal-arrangement is very favorable for the activation of pyramidal -slip as well as basal -slip. During plastic deformation the basal planes, which were initially 45° inclined, has an extra rotation of totally about 12° in the direction of the perpendicular direction to loading axis. This rotation can be found on the ODF as the rotation of {-1 2 -1 2} component to {-2 4 -2 3} (325, 65, 0). The pyramidal plane is also rotated about 10° in direction of the loading axis, which is described on the ODF as the rotation of {0 1 -1

225 2} to {0 3 -3 4}. The strengthening of {0 1 -1 0} (0, 90, 30) occurred by the consumption of {0 1 -1 2} component, i.e. the intensity of {0 1 -1 2} is decreased from 11.2 at initial state to 5.7 at measuring point 4, while the intensity of {0 1 -1 0} is increased from 2.3 to 5.8. In view of the Schmid’s law, these final texture is more difficult for further -slip than the initial state. However the basal planes laid at about 57° to the loading axis until fracture act as the positive factor for sliding of -dislocations. Because of this factor the 45° sample shows highest fracture strain and a relatively low value of strain hardening. These experimental results from in-situ measurements can be used for the texture simulation which is necessary for more detailed study on the quantitative determination of activities of different deformation modes inclusive -slip.

4

Summary

In-situ texture measurements using hard X-rays with 100keV photon energy were carried out to examine the influence of the initial texture on the mechanical behavior and on the further development of texture in AM20. Results are summarized as follows: 1. By using hard X-rays and a UTM set up complete pole figures during loading could be taken successfully. 2. The results from the in-situ measurements present a variation of the texture that supports and gives fundamental information for texture simulation. 3. The initial texture has a strong influence on the mechanical properties of AM20. The 0° sample which has unfavorable texture components for -slip shows the highest yield o strength. The texture components of the 45 sample is favorable for the activation of -slips so that this sample has the lowest yield strength and the highest fracture strain. In case of the 90° sample the arrangement of basal planes perpendicular to tensile direction causes an activation of tensile twinning at the beginning deformation stage, so that a higher strength hardening and a dramatic change of texture occurs.

Acknowledgments This work has been funded by the German Ministry of Education and Research (BMBF) under the contract numbers 03BRE8CL and 05KS1MCA/2.

References [1] [2] [3] [4] [5] [6]

H. Friedrich and S. Schmann, Mat.-wiss. u. Werkstofftech. 2001, 32, 6 S. Zaefferer, Mater. Sci.& Eng. A 2003, 344, 20 H.-G. Brokmeier, U. Zink, T. Reinert and W. Murach, J. Appl. Cryst. 1996, 29, 501 M. Dahms and H.J. Bunge, J. Appl. Cryst. 1989, 22, 439 I.L. Dillamore and W.T. Roberts, Metall. Reviews 1965, 10, 271 R.A. Lebensohn, M.I. Gonzalez, C.N. Tome and A.A. Pochettino, J. Nuclear Mater. 1996, 229, 57

226

Strengthening Phases in Extruded Quasicrystal Containing Mg-Zn-Y Alloys Alok Singh*, A.P. Tsai*, M. Nakamura*, M. Watanabe** and A. Kato** * National Institute for Materials Science, Tsukuba 305-0047, Japan ** Toyota Motor Corporation, Toyota 471-8572, Japan

1

Introduction

There is a need for magnesium alloys with good strength at 473 K. A strengthening phase is required which provides microstructural stability at higher temperatures. A special phase, called the icosahedral quasicrystalline phase, exists in the alloy system Mg-Zn-RE, where RE is Y or a rare earth element. An icosahedral phase is unusual because it possesses fivefold symmetry and a quasiperiodic structure. Due to its quasiperiodic order, dislocations cannot move easily through it, which makes it hard and brittle. Therefore it can form effective composites with other metallic phase. In the Mg-Zn-Y system it occurs in equilibrium with the magnesium matrix [1– 3], showing a definite orientation relationship and with a coherent interface [4]. In dilute Mg-Zn-Y alloys, the icosahedral phase forms in intergranular spaces. This phase is stable up to the eutectic temperature, which occurs above 723 K. Thus it gives stability to the microstructure and retards grain growth. If a fine grain microstructure is obtained, its strength can be retained until temperatures of 473 K and above. A fine distribution of the icosahedral phase in Mg-Zn-Y alloys has been obtained by thermomechanical processing. Bae et al. [5, 6] have employed hot rolling at 673 K for magnesium alloys containing 2– 4.3 at% Zn and 0.2 – 0.7 at% Y to achieve good tensile strengths at room temperature (RT) (yield stress (0.2 offset, YS) of 220 MPa and ultimate tensile strength (UTS) of 370 MPa with 17.2 at% elongation). However, this level of strength is not retained at higher temperatures, though a high ductility is attained. We have employed extrusion as the thermomechanical means to obtain a fine microstructure of the icosahedral phase [7]. Alloy Mg95Zn4.2Y0.8 was extruded with a ratio 1 0 :1 at two different temperatures of 673 K and 523 K. Detailed microstructural study was performed with transmission electron microscopy (TEM) and correlated with tensile strength to determine the role of various phases in strengthening. Here we examine the effect of extrusion temperature and heat treatment on tensile strength.

2

Experimental Procedure

The alloy corresponding to composition Mg95Zn4.2Y0.8 was prepared by melting high purity elements in an electric resistance furnace. The alloy was homogenized by annealing at 673 K for 10 h and then extruded with a reduction ratio of 10 :1. One batch was extruded at 673 K while another at 523 K. Heat treatments were carried out by sealing the specimens under argon in silica tubes. Samples for TEM were prepared by ion milling. A

227 JEOL 2000FX-II microscope operated at 200 kV and a JEOL 4000EX-II microscope operated at 400 kV were used. Tensile tests were performed at a strain rate of 10– 3 /s on round specimens with a diameter of 4 mm and gauge length of 22 mm at RT, 473 K and 573 K on a Shimatzu Autograph machine.

Figure 1: Optical micrographs showing the grain structure and distribution of the icosahedral phase (dark) along the extrusion direction in the alloy extruded at (a) 673 K and (b) 523 K.

3

Results

3.1

Microstructure

The as-cast alloy had grain size of approximately 30 μm with the icosahedral phase at the grain boundaries. Extrusion at two different temperatures produced different grain sizes. After extrusion at 673 K the grain size was about 2 μm, while it was typically 10 μm on extrusion at 523 K, as shown in optical micrographs of Figure 1. The main feature of the microstructure was the icosahedral phase. In as cast alloy it occurred on grain boundaries. In extruded alloy it occurred as rows of intergranular particles of size about 500 nm, an example in Figure 2. The quasiperiodic nature of the icosahedral phase is brought out from this diffraction pattern. Pentagons are drawn connecting intense spots to bring out its fivefold symmetry. The diffraction spots are arranged not periodically, but quasiperiodically. Fine icosahedral phase particles were also distributed in the matrix. In alloys extruded at 673 K these particles occurred as nano-particles, well facetted on the basal and prismatic planes of the matrix shown in Fig 3a. The icosahedral phase showed a well-defined orientation relationship with the matrix. One of the icosahedral twofold axes occurred along the matrix hexagonal axis [4]. At the interface icosahedral twofold planes match basal and prismatic planes of the matrix. In samples extruded at 523 K, however, the icosahedral phase particles that occurred in the matrix were over 100 mm in size and rough shaped Figure 3b. Occasionally cubic W phase (Mg2Zn3Y3) was found to exist in the alloys [8].

228

Figure 2: An intergranular icosahedral phase particle after 673 K extrusion sample and its fivefold diffraction pattern.

Figure 3: (a) Nano-icosahedral particles in a grain of 673 K extruded sample. (b) Fragmented icosahedral phase particles distributed in the matrix after extrusion at 523 K. Fine precipitation of τ1 phase is also observed.

Complex precipitation of Mg-Zn binary as well as ternary phases occurred in the matrix. In the samples extruded at 673 K, the equilibrium phase MgZn occurred, apparently by nucleation on dislocations. In the samples extruded at 523 K, fine precipitation occurred in the matrix. These precipitates were mainly two kinds – fine rod-like precipitates, apparently of the precursor phase β1’ [9, 10], and another fine precipitate of a ternary phase τ1. Solutionizing was carried out at 673 K (below the eutectic temperature) to achieve finer precipitation in the matrix. Solutionizing at this temperature did not dissolve the icosahedral phase, but the icosahedral phase became rounded upon this treatment. Solutionizing produced a fine distribution of two phase – τ1 and tiny rods of β1’ (Figure 4). The phase τ1, from electron diffraction patterns, appears to be similar to a ternary phase τ, reported in Zn-Mg-Dy [11], where it was shown to be related to quasicrystalline structures. τ1 appears to have defect structures, and is considered to be a phase intermediate between Zn7Mg4 and τ. The τ1 precipitates exhibited a rough interface with the matrix and a strain around them. Upon further annealing at 473 K (10 h) the β1’ rods grew a few hundred nanometers in length. After tensile tests dislocation pile-up on these rods were observed. Microstructures after solutionizing are shown in Figure 5.

229

Figure 4: Precipitates after solutionizing. (a) β1’ rods along with an icosahedral phase particle in alloy extruded at 673 K. (b) τ1 phase precipitates in alloy extruded at 523 K.

Figure 5: Microstructure after solutionizing at 673 K. (a) In alloy extruded at 673 K. Two icosahedral particles are pinning the grain boundary. (b) Alloy extruded at 523 K. Interaction of dislocations with icosahedral phase particles on mechanical testing at RT. Arrow shows growth of τ1 phase over icosahedral phase. Its diffraction in inset.

3.2

Tensile Strength

Figure 6 shows the tensile strength of the extruded alloys at RT, 473 K and 573 K. Asextruded alloys showed good strength at room temperature. Extrusion at 673 K gave better strength at RT (YS 262 Mpa, UTS 332, elongation ~18%). However, the YS fell to less than half at 473 K, exhibiting no strain hardening and resulting in poor UTS. Upon solutionizing, however, substantial strain hardening was introduced, which raised the UTS from 125 Mpa to 170 MPa at 473 K, Figure 7a. Subsequent annealing at 473 K raised the YS to 145 MPa. The stress-strain curves are compared with those from a commercial extruded AZ61 alloy. A better microstructural stability of the Mg-Zn-Y alloy is indicated. In case of extrusion at 523 K, the as extruded alloy retained substantial strength at 473 K (YS 133 MPa, UTS 163 MPa). In this case too the elongation at 473 K is about 40%. The strength dropped to very low levels at 573 K in both cases, accompanied by nearly 100% elongation. Upon solutionizing, a substantial strain hardening was observed at 473 K resulting in a significant rise in the ultimate tensile strength (UTS), as observed in Figure 7b.

230

Figure 6: Tensile stress-strain curves at RT, 473 K for alloys extruded at (a) 673 K and (b) 523 K.

Figure 7: Stress-strain curves at 473 K and 573 K of alloys in as-extruded (extr.), solutionizing (soln.) and solutionized and annealed (ann.) condition. A curve for an extruded AZ61 alloy is shown for comparison.

4

Discussion

The primary strengthening in these alloys appear to be from the icosahedral phase. It forms strong interfaces with the matrix. Its occurrence at the grain boundaries provides strengthening and retards grain growth. It interacts strongly with grain boundaries and dislocation. An important point is that heat treatments at a high temperature of 673 K is possible without substantial grain growth even though the grain size is very fine in the extruded condition. No significant loss in strength at room temperature occurred after the solutionizing treatments. The higher temperature strength became better. Further strengthening is provided by precipitation within the grains. Presence of the τ1 precipitates seem to be responsible for word hardening, while lengthening of the β1’’ rods directly correlates with increase in YS. The solubility of yttrium decreases with lowering temperature and is almost nil at RT. The heat treatment and cooling rates are critical for nucleation of competing phases. Greater control is necessary in extrusion at 673 K. The extrusion at 523 K has several advantages. The icosahedral phase is distributed mor effectively in the alloy and phase equilibrium is not altered during the process, avoiding further heat treatment.

231

5

Summary

The microstructure and strength of an icosahedral quasicrystal containing Mg-Zn-Y alloy has been examined after extrusion at two different temperatures of 673 K and 523 K. The RT strengths after extrusion at these different temperatures were nearly the same; however, extrusion at lower temperature gave better higher temperature strengths in asextruded condition.

6

Acknowledgements

This work is partly supported by Japan Science and Technology Corporation.

7

References

[1] [2] [3] [4]

Z. Luo. S. Zhang, Y Tang, D. Zhao, Scripta Metall Mater. 1993, 28, 1513. A. Niikura, A.P. Tsai, A Inoue, T. Masumoto, Phil. Mag. Lett. 1994, 69, 35. A.P. Tsai, Y. Murakami, A. Niikura, Phil. Mag. A 2000, 80, 1043–1054. A. Singh, A.P. Tsai, M. Nakamura, M. Watanabe, A. Kato, Phil. Mag. Lett. (in press). D.H. Bae, S.H. Kim, D.H. Kim, W.T. Kim, Acta Mater. 2002, 50, 2343 –2356. D.H. Bae, M.H. Lee, K.T. Kim, W.T. Kim, D.H. Kim, J. Alloy Compounds 2002, 342 –445. A. Singh, M. Nakamura, M. Watanabe, A. Kato, A.P. Tsai, Scripta Mater. 2003, 49, 417 –422. A. Singh, A.P. Tsai, Scripta Mater. 2003, 49, 143 –148. J.B. Clark, Acta Metall. 1965, 13, 1281– 1289. L.Y. Wei, G.L. Dunlop, H. Westengen, Metall. Mater. Trans. A 1995, 26A: 1705 – 1716. E. Abe, A.P. Tsai, Acta Cryst. B 2000, 56, 915 – 917.

[5] [6] [7] [8] [9] [10] [11]

Grain Structure Characterization of AM60 Die Castings by Electron Backscatter Diffraction (EBSD) Measurements in SEM Hans I. Laukli, Otto Lohne, Lars Arnberg Norwegian University of Science and Technology, Trondheim

Introduction The fine-scaled microstructures in high pressure die cast (HPDC) magnesium alloys impede quantitative measurements of the grain size. The microstructure typically consists of large pre-solidified crystals [1] and fine grains that are too small to be easily resolved in the optical microscope. The grain boundaries are generally difficult to separate due to the inter-granular dispersion of divorced eutectic that consists of Al-segregated Mg and the eutectic β-Mg17Al12 phase [2]. Electron backscatter diffraction (EBSD) measurements in SEM are shown to be a more suitable method for investigations of the grain structure.

Experimental Box-shaped AM60 castings, with a wall thickness of 2.5mm, were produced at Norsk Hydro’s Magnesium Competence Center, Porsgrunn, in a Bühler SC42D 420 ton cold chamber HPDC machine (details in [1]). A complete AM60 die casting is displayed in Figure 1.

Figure 1. Complete AM60 die casting with biscuit, runners and overflows.

For investigations in the optical microscope the samples were ground, mechanically polished and etched in an acetic acid glycol solution [3]. For EBSD measurements in SEM the samples were ion milled in an Ion-Tech FAB 306 atom mill. A Jeol 840 SEM equipped with an EBSD detector was used for the grain size investigations with operation parameters as listed in Table 1. For instantaneous data processing, HKL Channel software was used [4].

233 Table 1. Ion milling and SEM* operation parameters.

Acceleration voltage [kV] 5 20*

Current Gun [mA] angle 5 15o -6* ~10

Time [h] ~4

Specimen Working distance [mm] tilt

Resolution/ step size [μm]

70o*

1*

~23*

The EBSD process When scanning the electron beam on the tilted specimen surface in the SEM, Figure 2 a), inelastic scattering occurs. Bragg diffracted electrons for particular planes are channeled differently which results in a change in intensity. Two cones for each crystallographic plane are formed and appear as straight bands, Figure 2 b), termed electron backscatter patterns (EBSP) [4]. Patterns from each grain are detected on a phosphorous screen and by a CCD-camera, and with EBSD acquisition-software the crystallographic orientation of grid points in the microstructure is determined. Qualitative EBSD measurements can be obtained in band contrast images, Figure 2 c). Dark gray appearance corresponds to a lower quality. This can result from overlapping patterns which is typically associated with grain boundaries. Lighter gray indicates a higher quality, e.g. in the center of a grain. The pattern quality can also be influenced by the surface condition of the specimen [5]. A non-uniform ion milling of the sample can result in non-indexed points as observed in the white area, marked 1, in Figure 2 d). Occasionally, grid points are misindexed due to poor pattern quality, and appear as isolated and highly misoriented pixels on the orientation map [5], e.g. the pixels marked 2 in Figure 2 d). Different crystallographic orientations combined with grain boundaries can be processed to appear as grain maps, where a specific colour indicates a grain and grain boundaries are depicted as black lines. Misorientations greater than a specific value, e.g. 15°, are defined as grain boundaries.

Figure 2. a) Secondary electron image of the ion milled specimen surface, b) EBSP of a magnesium grain, c) band contrast image and d) orientation map (resolution 0.5 μm step size).

234

Results In Figure 3 a) an optical micrograph of the HPDC AM60 microstructure is displayed. A few coarse crystals is revealed by the etching, while the fine grains are difficult to resolve. An orientation map obtained with EBSD in SEM from an equivalent microstructure is observed in Figure 3 b). The average grain size is measured to be approximately 7 μm.

Figure 3. a) Optical micrograph of AM60 HPDC microstructure revealed by etching and b) orientation map of equivalent microstructure with an average grain size of 7μm.

Grain maps from adjacent areas can be stitched together to form quantitative orientation maps of the grain structure in large areas of the microstructure. In Figure 4 (bottom image) the grain structure is revealed by means of orientation maps approximately halfway through the thickness of the micrograph in Figure 4 (top image). Relatively coarse grains are present close to the surface and finer grains are observed at an intermediate position. In the center, coarse grains are present surrounded by fine grains.

Figure 4. Optical micrograph of etched cross section (top) and stitched orientation maps from cross section (bottom).

235

Acknowledgements The present work was funded by the project NorLight Shaped Castings and partners: the Norwegian Research Council, Alcoa Automotive Castings, Scandinavian Casting Center ANS, Elkem Aluminium ANS, Fundamus AS, Hydro Aluminium Metal Products, Norsk Hydro ASA, NIMR, NTNU, and SINTEF (project responsible). The authors thank the partners for the financial support.

References [1] H.I. Laukli, O. Lohne, S. Sannes, H. Gjestland, L. Arnberg, Int. J. C. Met. R., 2003. Submitted. [2] A.K. Dahle, Y.C. Lee, M.D. Nave, P.L. Schaffer, D.H. StJohn, J. Light Met., 2001, 1, 61-72. [3] Microstructure of magnesium based diecasting alloys in Hydro Diecaster Bulletin, 3, 1999. [4] HKL Technology Channel, 5, HKL Technology, 2001. [5] F.J. Humphreys, J. Mat. Sci., 2001, 36, 3833-3854.

Possibilities of ECAP of Magnesium Alloy Katarzyna N. Braszczyńska Institute of Materials Engineering, Technical University of Cz stochowa, stochowa, Al. Al. Armii Krajowej 19, 42-200 Cz st stochowa, Poland

Abstract The development of microstructure during equal-channel angular pressing (ECAP) of commercial AZ91 alloy was investigated. In order to obtain high plastic deformation structure warm die with the channel angle Φ and the corner angle Ψ equal 90° and 120°, respectively, was used. Samples were deformed isothermally at different temperatures and at the speed of 0.83 mm/s. The obtained results demonstrate the possibility of obtaining the homogeneous microstructure, characterized by considerable grain refinement in the investigated alloy.

Introduction Several plastic deformation (SPD) techniques are known as a tool to produce fine-grained materials, which can exhibit superplastic behavior in a certain range of thermo-mechanical parameters [1-7]. Although several different procedures are currently available for imposing the severe strains necessary to attain an ultrafine-grained structure, most attention has been devoted to the two techniques of high-pressure torsion (HTP) and equal channel angular pressing (ECAP). Processing through the procedure of ECAP is a metal working process allowing to obtain ultra-fine grain in bulk materials [6-16]. This procedure differs from more conventional processes, such as rolling and extrusion; it has the ability to introduce on exceptionally high plastic strain without any concomitant change in the cross-sectional dimensions of the sample. Equal-channel angular pressing (ECAP) in which a piece of material is pressed to pass two connected channels having equal cross sections and intersecting at an angle has emerged as a promising process to obtain new structure and properties of materials [7, 12, 17-23]. In practice, the strain imposed on the sample in ECAP depends upon the two angles defined in Fig. 1: there is an angle Φ between the two channels and an additional angle Ψ delineating the outer arc of curvature at the point of intersection of the channels. The microstructure and properties of materials pressed by ECAP are strongly dependent on the plastic deformation behavior during pressing which is governed mainly by die geometry, and process variables as temperature, pressing speed and rotation of the billet around its longitudinal axis between adjacent passes. Theoretically, the equivalent strain, ε, generated in the work piece after one pass of ECAP is given by the following relation [6]: ε=

1 ⎡ ⎛Ψ Φ⎞ ⎛ Ψ Φ ⎞⎤ ⎢ 2ctg ⎜ 2 + 2 ⎟ + Ψ cos ec ⎜ 2 + 2 ⎟ ⎥ 3⎣ ⎝ ⎠ ⎝ ⎠⎦

(1)

237

Figure 1. A schematic representation of ECAP die with the channel angle Φ and the corner angle Ψ.

Recently many workers are interested in applying ECAP process to magnesium alloys which are the lightest structural metals and hence, have many potential applications [1623]. However, magnesium and its alloys generally exhibit poor workability to improve because of HCP structure, which has limited slip systems resulting in the poor ductility. Grain refinement during the ECAP of metals with cubic structures is controlled by the formation of sub-boundaries with high dislocation densities. By contrast, the deformation of magnesium may be expected to be quite different because it involves not only the slip of dislocations but also deformation twinning. In the present study, equal channel angular pressing was undertaken in order to examine possibilities of grain refined in an AZ91 magnesium alloy.

Experimental material and procedure As-received AZ91 magnesium alloy had a dendritic microstructure, typical for the gravity-cast alloy, which was characterized by very heavy segregation of alloying elements. Mg-Al alloys are prone to segregation due to relatively wide temperature spans between the liquids and the solids curves. Non-equilibrium solidification conditions caused the formation of large crystals of the (Mg) phase (depleted in aluminum) and pushing the Al admixture away into inter-dendrical spaces (Fig. 2). The details of microstructure were described in work [24]. Before the ECAP processes, the samples were heat treated in order to obtain homogenizing microstructure, characterized by higher workability. The solution annealing was carried out at 693 K for 26 h in protective argon atmosphere. The microstructure obtained after this heat treatment is shown at Fig. 3. It should be noted that this homogenizing treatment is one of main processes for magnesium alloys, however, it causes obtaining great grains. The investigated AZ91 alloy was cut into rods with about 50 mm length and 11.8 mm diameter. The ECAP die used in this investigation was designed to obtain a shear strain of ~1.15 during each pass. It contained an inner contact angle Φ equal 90° and the cor-

238 ner angle Ψ of 1209. All billets were processed at a processing rate of 0.83 mm/s and by route BC. Molybdenum disulphide (MoS2) was used as lubricant.

Figure 2. Microstructure of AZ91 as-received

Figure 3. Microstructure of AZ91 alloy obtained after solution annealing

The processing temperatures were chosen for the best combination of ductility and efficient grain refinement. For each separate pressing, the sample was sprayed with lubricant and it was then placed in the channel after the die had reached the required temperature. Before the first press, sample was heated to a required temperature. Following ECAP, microstructure was observed using specimens cut either parallel or normal to the longitudinal axis.

Results Fig. 4. shows the macroscopic images of samples, which experienced a single ECAP pass under isothermal conditions at three different temperatures. At temperature below 553 K investigated alloy exhibited failure characterized by the formation of a series of segments along the length in one pass ECAP process (Fig. 4a, b). As temperature was raised, a transition in flow from segmented to uniform samples was observed (Fig. 4c). It should be noted that the billet temperature during an ECAP operation may be above the die temperature due to the heat generated by the mechanical work. As was analyzed [20], the maximum temperature rise at high strain rates was estimated as 26 K for AZ91 alloy.

239

a)

b)

c) Figure 4. Macrographs of AZ91 samples processed via ECAP at 473 K (a), 523K (b) and 553K (c); one pass. Direction of exit channel was marked with an arrow.

Representative microstructures in AZ91 ECAP samples pressed at different conditions are shown in Fig.5-7. Fig. 5a was taken from the transverse section of the as-pressed billet while Fig. 5b was taken from the longitudinal section sample passed at 553 K. In the transverse section, the small grains surrounding some larger grains were observed. In longitudinal section, the formation of a banded structure lay essentially parallel to the shearing plane at approximately 45° to the top and bottom surface of the sample. a)

b)

Figure 5. The microstructure of AZ91 after ECAP process at 553 K; two passes; the transverse section (a) and the longitudinal section (b)

Fig. 6 shows that same evidence for a retention of the banded structure after two passes through the die at 623K; the microstructure is homogeneous consisting equiaxed grains. Fig. 7 shows that after four passes at this temperature a reasonably homogeneous microstructure of small and equiaxed grains exists. The changes of shape and size of grains shown on these figures can appear as a consequence of ‘rotation dynamic recrystalization’ (RDX) during ECAP process. This dynamic recrystallization mechanism, described in works [25-26], is different from conventional

240 dynamic recrystallization (DRX) and was observed in magnesium alloys after hot compression or hot rolling processes. It is also apparent that ECAP is especially effective in reducing the grain size when the pressing is conducted at lower temperatures, where the occurrence of grain growth is limited. However, the grain size did not change during the ECAP processes carried out at temperatures below 673 K. This indicates, that the dynamic recrystallization plays main role in the grain refinement. a)

b)

Figure 6. The microstructure of AZ91 after ECAP process at 623 K; two passes; the transverse section (a) and the longitudinal section (b)

a)

b)

Figure 7. The microstructure of AZ91 after ECAP process at 623 K; four passes; the transverse section (a) and the longitudinal section (b)

Summary 1. The ECAP process was successfully carried out isothermally for AZ91 alloy in order to obtain grain refinement. 2. At temperatures below 553 K, the ECAP process carrying out in external shape of die was practically impossible, because of destruction of samples. 3. After pressing at temperature above 553 K the grain size of AZ91 alloy was drastically decreased and above 623 K homogeneous microstructures were obtained.

241

Acknowledgements This work was supported by State Committee for Scientific Research (KBN), Poland under grant 7T08B00222. The author is grateful for the fellowship from the Foundation for Polish Science.

References [1] M. Kumachi, M. Furukawa, Z. Horita, T.G. Longdon, Mater. Sc. and Eng., A347, 2003, 223-230 [2] C. Xu, T.G. Longdon, Scrip. Mater., 48, 2003, 1-4 [3] H.S. Kim, M.H. Seo, S.I. Hong, J. of Mater. Proc. and Tech., 130-131, 2002, 497503 [4] S.J. Oh, S.B. Kang, Mater. Sc. and Eng., A343, 2003, 107-115 [5] M. Furukawa, Z. Horita, T.G. Longdon, Mater. Sc. and Eng., A332, 2002, 97-109 [6] Y. Iwashashi, J. Wang, Z. Horita, M. Nemoto, T.G. Longdon, Scrip. Mater., 35, 1996, 143-146 [7] H.S. Kim, M.H. Seo, S.I. Hong J, J. of Mater. Proc. and Tech., 113, 2001, 622-626 [8] V. Stalyarov, Y.T. Zhu, I.V. Alexandrov, T.C. Lowe, R.Z. Valiev, Mater. Sc. and Eng., A299, 2001, 59-67 [9] I. Kim, J. kim, D.H. Shin, C.S. lee, S.K. Hwang, Mater. Sc. and Eng. A342, 2003, 302-310 [10] S.L. Semiatin, D.P. DeLo, Mater. and Des., 21, 2000, 311-322 [11] Z. Horita, T. Fujinami, T.G. Langdon, Mater. Sc. and Eng., A318, 2001, 34-41 [12] S. Komura, M. Furukawa, Z. Horita, M. Nemoto, T.G. Langdon, Mater. Sc. and Eng., A297, 2001, 111-118 [13] D.h. Shin, I. Kim, J. kim, Y.S. Kim, S.L Semiatin, Acta Mater., 51, 2003, 983-996 [14] Y. Nishida, H. Arima, J.C. Kim, T. Ando, Scrip. Mater., 45 2001, 261-266 [15] Y. Wu, I. Baker, Scrip. Mater., 37, 1997, 437-442 [16] T. Mukai, M. Yamanoi, H. Watanabe, K. Higashi, Scrip. Mater., 45, 2001, 89-94 [17] W.J. Kim, C.W. An, Y.S. Kim, S.I. Hong, Scrip. Mater., 47, 2002, 39-44 [18] M. Mabuchi, H. Iwasaki, K. Yanase, K. Higashi, Scrip. Mater., 36, 1997, 691-686 [19] A. Yamashita, Z. Horita, T.G. Langdon, Mater. Sc. and Eng., A300, 2001, 142-147 [20] Y. Nishida, T. Ando, M. Nagase, S. Lim, I. Shigematsu, A. Watazu, Scrip. Mater., 46, 2002, 211-216 [21] H. Watanabe, T. Mukai, K. Ishikawa, M. Mabuchi, K. Higashi, Mater. Sc. and Eng., A307, 2001, 119-128 [22] M. Mabuchi, K. Ameyama, H. Iwasaki, K. Higashi, Acta Mater., 47, 1999, 20472057 [23] W.J. Kim, C.W. An, Y.S. Kim, S.I. Hong, Scrip. Mater., 47, 2002, 39-44 [24] Braszczyńska K.N., Zeitschrift für Metallkunde, 93, 2002, 845-850 [25] S.E. Ion, F.J. Humphreys, S.H. White, Acta Metall., 30, 1982, 1909-1914 [26] J.A. Valle, M.T. Perez-Prado, O.A. Ruano, Mater. Sc. And Eng., A35, 2003, 68-78

Electron Probe Micro Analysis of Sedimented Zirconium Particles in Magnesium Bruce Davis* and Keyna O’Reilly** *Materials Science Department, Oxford University, now Magnesium Elektron, Manchester **Materials Science Department, Oxford UniversityJohn King, Magnesium Elektron, Manchester

Introduction The grain refinement of Mg and its alloys by Zr was first achieved in 1947 by Sauerwald [1]. Zr is a potent nucleant of Mg and can result in grain sizes of the order of 20μm in some alloy systems, a factor of ten smaller than can be achieved in the grain refinement of Al alloys and better than any other refinement system available for Mg alloys. The high potency of Zr particles as nucleants for solid Mg is due to: the small size of the particles (≈1μm); the close match of the lattice parameters and similarity to the crystal structure of Mg; and the process by which Zr particles are evenly dispersed throughout the melt. The equilibrium phase diagram for the Mg-Zr binary alloy system contains a peritectic reaction at 654 °C for Zr concentrations in excess of 0.58wt%. The process of grain refinement is thought to depend on particles of α-Zr that precipitate from solution as the melt cools below the liquidus temperature. This results in a cascade of very fine particles forming throughout the melt that act as nucleants of solid Mg below the peritectic temperature. It has been stated in the literature that grain refinement occurs in Mg even when the levels of Zr are below 0.58wt% [2]. This would suggest that undissolved Zr particles, introduced from the hardener play a role in grain refinement of Mg. However, it is generally considered that un-dissolved Zr does not act as a nucleant of Mg due to some difference of the particle surface, such as an oxide film, that interferes with the chemical wetting properties [3]. This study investigates the use of sedimentation as a method of separating Zr particles that formed in molten Mg for further analysis. Electron probe micro analysis was used to characterise the chemical distributions of particles within the samples. The results obtained are presented to help explain the phenomena of grain refinement in Mg containing low levels of Zr.

Experimental Methods and Materials The technique of sedimentation utilizes the difference between the density of intermetallic compounds present in the molten metal to separate them from the bulk of the liquid by allowing the particles to settle to the bottom of the sample. It is then possible to section the layer of sediment from the material after solidification for further analysis. Sedimentation has previously been used to study the effect of undercooling on the nucleation of TiAl3 in aluminium alloys [4, 5].

243 The sedimentation process was carried out in a three-zone vertical tube furnace for accurate temperature control and to minimise the temperature gradient along the length of the samples. This was measured to be less than 1°C. Accurate temperature control was required to prevent undershooting of the temperature when the samples were cooled. The furnace was mounted on a rubber base to minimise vibration, as this is a known cause of nucleation in solidifying metals. A dynamic argon atmosphere prevented oxidation of the liquid metal during experiments. All experiments were carried out in triplicate. The sedimentation studies were carried out on the grain refinement of commercially pure magnesium. Zirconium was added to the melt in the form of a proprietary hardener called Zirmax® supplied by Magnesium Elektron. Samples for use in the sedimentation experiments were grain refined using 5wt% Zirmax®, then cast for use in the sedimentation furnace in a mild steel die. The grain refinement and casting of the samples for sedimentation was carried out in accordance with Magnesium Elektron’s recommended practice. The melt was vigorously stirred following the addition of the Zirmax® and held for ten minutes prior to casting. The casting temperature was 780 °C.

Experimental Procedure for the Sedimentation Technique. Stage one of the sedimentation experiment involved cleaning the metal at the casting temperature, 780 °C. This removed any undissolved Zr particles that may have been present in the melt. When the crucible holder reached the required temperature it was raised to the top of the furnace. Three samples, contained in graphite crucibles, were placed in the holder and lowered into the hot region of the furnace. The samples were allowed to equilibrate for 45 minutes and then held for one hour to allow any particles present to settle to the bottom of the graphite crucibles. The crucible holder was then raised to the top of the furnace and the samples placed in a quench block. The quench block was a water-cooled brass crucible holder that chilled the base of the samples where the sedimented layer was present. When fully cooled the base of the samples were cut off, removing the sedimented layer. The remaining lengths of samples where then placed back in the crucibles for the next stage of the sedimentation experiment. Stage two of the sedimentation experiments involved reheating the samples to the original sedimentation temperature, 780 °C, to dissolve any Zr particles present in the bulk of the sample. The samples were then cooled by 20 °C and held for one hour. This allowed any Zr particles formed during cooling to settle to the base of the samples. The samples were then removed from the furnace and quenched as outlined above, and the sedimented layer removed. The process was repeated, cooling the samples by a further 20 °C each time, to produce material that had been held at 760, 740, 720, 700 °C. The grain size of these samples was measured using the linear line intercept method. A control experiment was performed to establish whether sedimentation of a liquid sample resulted in all solid particles present in the melt being removed. This was achieved by cleaning a sample through sedimentation at 780 °C as in stage one, removing all heavy inclusions, i.e. undissolved Zr particles, and quenching. The sedimented layer was then removed. The sample was reheated to 780 °C and held for one hour to allow any remaining or newly introduced particles to sediment to the base of the sample. The sample was then quenched and the base of the sample removed. Finally the sample was reheated to 780 °C and cooled to 720 °C, held of one hour, then quenched. The sections containing sedimented layers were inspected using optical microscopy.

244

Electron Probe Micro Analysis, EPMA EPMA of the samples was carried using a JEOL JXA8800 microprobe. The apparatus utilised wavelength dispersive x-ray analysis with four spectrometers, allowing four elements to be detected at any one time. The accelerating voltage used was 20 keV, with a dwell time of 100 ms. For simplicity the EPMA results presented in this paper are qualitative due to the complexity of quantitative EPMA data analysis.

Results The layer of sediment obtained from the sample held at 780 °C for one hour is shown in figure 1. It can be seen that the layer is approximately 150 μm thick. Comparison of the grain structure in the bulk of the material with the structure in the sedimented layer indicates that the grain size in the bulk of the material was larger, see figure 1a.

30 μ m

300 μ m

a)

b)

Figure 1. Micrographs showing the layer of sediment produced during stage one of the technique and it’s effectiveness in use with magnesium. a) Low magnification showing presence of sediment layer at the base of the sample. b) High magnification showing the presence of particles within the layer and a refined grain size.

As Zr was removed from the samples, by the sedimentation of particles, the grain size increased steadily. The sample held at 780 °C had a grain size of 222 μm. Following further sedimentation at 760, 740, 720 and 700 °C the grain size of the samples increased to 694, 902, 966, and 1176 μm respectively. The sedimented layers produced by the control experiment are shown in figure 2. It can be seen that following the initial sedimentation at 780 °C a number of particles settled to the base of the sample, resulting in refinement adjacent to the particles. Remelting the sample and holding the liquid metal at 780 °C for a further hour did not result in anyfurther sedimentation of particles, see figure 2b). Remelting the sample by heating to 780 °C and then cooling to 720 °C did cause further particles to form in the melt and settle to the base of the sample as seen in figure 2c. 100 μ m

a)

b)

c)

Figure 2. Base of samples produced during control experiment. a) Sedimentation at 780 °C after initial casting and grain refinement. b) Further sedimentation at 780 °C. c) Further sedimentation at 720 °C.

245 EPMA of the sedimented layer formed in the sample cooled to 720 °C indicated that there was an association of Fe and Si to the Zr present in the sample, see figure 3. The elemental maps indicate that there were high concentrations of Zr, Si and Fe at the grain boundaries of the sedimented layer. There is also evidence of Zr, Si and Fe present at the centre of a few grains. The structures revealed by EPMA can be interpreted as nucleating Zr particles located at the centre of the grains and non nucleating particles that have been pushed to the grain boundaries following solidification. a)

b)

50μm

c)

d)

50μm

50μm

50μm

Figure 3. Elemental maps showing the distribution of Zr, Fe and Si in a sedimented sample. a) Zr content. b) Fe content. c) Si content, d) Back scattered electron image. The rings in d) encircle grain refining particles that can also be seen in the other three images.

To obtain further information concerning the nature of the Zr particles present within the sedimented layers of the samples a detailed EPMA study was performed to compare non-nucleating particles present at grain boundaries with a nucleating particle located at the centre of a grain. The results are shown in figures 4 and 5. EPMA of a nucleating particle is shown in figure 4a-d. The particle was approximately circular in cross-section. The secondary electron image shown in figure 4a) reveals the presence of internal substructure that appears to have a cored formation radiating from the central region. There was a distribution of smaller particles present in the surrounding matrix. The Zr map shown in figure 4b) indicates the particle to consist largely of Zr. Some smaller precipitates are also present on the Zr map identifying them as also containing some Zr. The Fe map in figure 4d) shows only trace amounts of Fe were present in the particle. It is not felt that the detected Fe was the result of interference from the Zr signal as the Fe distribution does not wholly coincide with the presence of Zr. The Si map in figure 4c) shows that some of the smaller particles contained Si. The Si map indicates that there was Si present in the immediate vicinity of the Zr particle and it is possible that the Si shown is a component of the particle. However a number of smaller particles were present that contain Si and these may be the cause of the Si detected in the EPMA map. 5μm

a)

5μm

b)

5μm

c)

5μm

d)

Figure 4. Detail EPMA of a nucleant particle showing a) Secondary electron image, b) Zr distribution, c) Si distribution, d) Fe distribution.

246 EPMA analysis of particles present at the grain boundaries within the sedimented layer of the sample are shown in figure 5. The particles were much smaller, ) basal-plane system and due to primary twinning, which occurs at the ( 10 12 ) and ( 3034 ) planes. Grain rotation, particle (inclusions) facture are less decisive in the micro-crack nucleation processes. Fig. 5 shows surface morphologies of polished RD and TD specimens close to the fracture surface. Comparing the two micrographs it is seen that in TD specimens, the twinning mechanism is dominant whereas in RD specimens dislocation slipping is the most active process. a)

RD and Loading direction

Fig. 5a. SEM surface morphology of polished RD-tensile specimen after fracture. (Image is parallel to the rolling plane)

493 b)

Loading direction RD

Fig. 5b. SEM surface morphology of polished TD tensile specimen after fracture. (Image is parallel to the rolling plane)

On both surfaces a lot of micro cracks are seen. In TD specimens the micro cracks initiate at the borders or intersections of twinned bands, see Fig. 5b. In RD-specimens the micro-cracks generate perpendicularly to the slip bands. Fractography For all specimen orientations the macroscopic fracture surface is inclined to the loading direction and thickness direction by about 45°, but laterally the fracture edge is normal to loading direction. Fig. 6a shows a typical example of SEM micrograph obtained from a RD-specimen. The fracture surface shows some steps. On the steps one can find flat areas, which have a scale of 20~50 μm, and similar to that of the grain size. These areas show traces of plastic deformation as seen in Fig.6b. Broken particles or cavity growth around particles were not found on more than 10 fracture surfaces investigated. a)

b) 45°

Fig. 6. Fracture surface morphology of tensile specimens. a) RD, general morphology and steps parallel to the free surface, b) RD, local image in the square of a), details in a flat area. (Micrographs are parallel to the macroscopic crack plane.)

494

Summary The AZ31 sheet material shows strong anisotropy in strength, ductility, deformation and fracture mechanisms, which can be summarised as follows: 1. In TD specimens the strength and ductility are higher than in RD specimens. 2. The anisotropy factor of plastic deformation, r, is large in TD specimens and small in RD specimens. No significant strain rate dependence of r was found. 3. In RD specimens plastic deformation occurs predominantly via dislocation slip-bands. In TD specimens intense twinning is observed. 4. In RD specimens, micro-cracks nucleate at the dislocation slip-bands. The micro cracks propagate perpendicular to the slip bands and 45° to the loading direction. 5. In TD specimens, micro-crack nucleation occurs at the intersections and interfaces of the twinned bands. 6. In TD and RD specimens the macroscopic fracture edge is oriented 90° to the loading direction and inclined by 45° in thickness direction (slant fracture).

Acknowledgement Many thanks are due to Kay Erdmann for his support in performing the tests. Thanks are also due to Dr. D. Letzig and Dr. J. Bohlen for their helpful discussion and for providing the test material.

References [1] Hydro Magnesium, Data Sheet, Die Cast Magnesium Alloys, 1999 [2] K. U. Kainer, F. von Buch, Mat.-wiss. u. Werkstofftech, 1999, 30, 159-167 [3] USGS (U. S. Geological Survey), Magnesium: Statistics and Information, 2001 and 2002 [4] E. Aghion, B. Bronfin, D. Eliezer, J. Mater. Proc. Techn., 2001, 117, 381-385 [5] W. J. Kim, C. W. An, Y. S. Kim, S. I. Hong, Scripta Materialia, 2002, 47, 39-44 [6] M. M. Myshlyaev, H. J. McQueen, A. Mwembela, E. Konopleva, Mater. Sci. Engng, 2002, A337, 121-133 [7] O. Chabanet, D. Steglich, J. Besson, V. Heitmann, D. Hellmann, W. Brocks, Computational Material Science, to be published, 2003 [8] J. Heerens, D. Steglich, W. Brocks, ABAQUS User Conference, Freiburg, Germany, 2001 th [9] F. Kaiser, J. Bohlen, D. letzig, K. U. Kainer et al, 6 International Conference on Magnesium, alloys and their applications, Wolfsburg, Germany, Sept. 2003

495

Unstable Plastic Deformation in Mg Alloys-post Relaxation Effect Zuzanka Trojanová1, Pavel Lukáč1, Luboš Čížek2 1 2

Faculty Mathematics and Physics, Charles University, Ke Karlovu 5, CZ-121 16 Praha 2 Technical University Ostrava, Tř. 17. listopadu 2172, CZ-708 00 Ostrava

Abstract Post relaxation effect, i.e. an increase of the flow stress at the beginning of plastic deformation after stress relaxation in comparison to the flow stress at the beginning of stress relaxation, is observed in the case of dynamic strain ageing. The post relaxation effect is sensitive to the strain and temperature. Possible changes in the internal stresses and the mobile dislocation density are considered to be responsible for the post relaxation effect.

Introduction The plastic deformation of alloys exhibit many phenomena associated with mobile solutes. Following features give an evidence of dynamic strain ageing (DSA): (1) a local maximum in the temperature dependence of the yield stress; (2) a local maximum in the temperature dependence of the stress sensitivity parameter n; (3) a stress increase after stress relaxation in the DSA temperature interval. The observed anomalies are caused by the change in the concentration solute atoms on dislocations with the strain rate and temperature. The stress relaxation (SR) technique has been demonstrated to be quite useful experimental method. The specimen is deformed to a certain stress σ0 and then the machine is stopped and the stress decreases with time t. The specimen can be again reloaded to a higher stress (load) and the test may be repeated. The time derivative σ& = dσ/dt is the stress relaxation rate and σ = σ(t) is the flow stress at time t during the SR. Stress relaxation is very often analysed under the assumption that the stress relaxation rate is proportional to the strain rate ε& according to [1] ε& = −σ& / M,

(1)

where M is the combined modulus of the specimen – machine set. In order to estimate parameters of the thermally activated processes from the SR the following theoretical relationship between stress drops rate σ& and stress σ can be used [1] ln ( −σ& ) = C + n ln σ

(2)

496 where C is a constant and n is the stress sensitivity parameter defined as ⎛ d ln ε& ⎞ n=⎜ ⎟ . ⎝ d ln σ ⎠ T

(3)

The SR are usually analysed assuming constancy of the mobile dislocation density ρm and internal stress σi. The stability of structure during the SR is a main condition for the correct determination of the activation volume and consequently for clearing up the microscopic mechanism controlling the dislocation mobility in a material. An unstable structure may influence the course of the SR. One of the reason for the structure instability (the dynamic strain ageing phenomena) in an alloy is the solute atoms diffusion towards moving dislocations during deformation: the Portevin-Le Chatelier effect, the positive dependence of the flow stress on temperature and its negative dependence on strain rate, the sharp decrease in the creep rate. The other macroscopic indication of the dynamic strain ageing is a post relaxation effect - the pronounced yield point appearing at sample reloading after the SR. The purpose of this work is to analyse SR tests performed in three magnesium alloy in the dynamic strain ageing temperature interval.

Experimental Procedure Two magnesium alloys were used in this study. Magnesium alloy AZ91 (nominal composition 9 wt% Al, 1 wt% Zn, 0.3wt% Mn, rem. Mg) was prepared by the squeeze cast technology. Thermal treatment T6 was used according to Polmear [2] (homogenisation for 18 h at 413 °C, then precipitation for 8 h at 168 °C). Magnesium alloy WE54 (nominal composition 5 wt.% Y, 2 wt% Nd, 2 wt% mixture of rare earth – Tb, Er, Dy, Gd, rem Mg) was prepared by powder metallurgical technology and extruded at elevated temperature. Figure 1: Post relaxation effect-schematic drawn.

Samples were deformed in an INSTRON 1195 machine in the temperature interval of 22-300 °C. Sequential stress relaxation tests were performed at increasing vs tress along a stress-strain curve. Duration of the SR was 300 s. Deformations at room temperature were -5 -4 -1 performed at various initial strain rates in the interval from 2.2 x 10 to 6.7 x 10 s and duration of the SR from 120 s to1500 s. SR were analysed according the equation (2). The slope of the plot ln(- σ& ) vs ln(σ/σ0) (σ0 is the starting stress of the SR) is the stress sensitivity parameter n defined by (3). As the characteristics of the post relaxation effect, the stress increment Δσ after reloading of the sample was used.

497

Experimental Results and Discussion

ln(-σ) [MPa/s]

Examples of the SR curves in a ln(- σ& ) vs ln(σ/σ0) representation obtained for WE54 alloy are introduced in Fig. 2 (for room temperature, duration of 600 s and various starting stresses of the stress relaxation test. Tem6 -5 -1 2.2x10 s perature dependence of the stress sensitivity 278 317 5 parameter n estimated for the first SR (in 349 the vicinity of the yield stress) and for the 420 4 last SR (in the vicinity of the maximum stress) are shown in Fig.3. The introduced 3 dependencies show a local maximum at 50 °C which indicates a presence of the 2 dynamic strain ageing process. Stress sen1 sitivity parameter obtained for WE54 alloy 0.96 0.97 0.98 0.99 1.00 at various starting stain rates depending on the stress is introduced in Figure 4. σ/σ0 The dynamic strain ageing is the pheFigure 2: SR curves obtained for various starting nomenon of interactions between moving stresses at room temperature for WE54 alloy dislocations and diffusing solute atoms. The impurity diffusion towards both immobile and mobile dislocations has been regarded in many works [3-5]. At size interaction impurities segregate at small times t, according to the law ⎛ Dt ⎞ cd = c0 ⎜ α 2 ⎟ ⎝ b ⎠

2/3

(5)

where cd is the solute atoms concentration, D is the diffusion coefficient for the solute atoms and α is a constant (depending on the interaction between a solute atom and the dislocation) and b has a magnitude of the lattice constant. The model developed by authors of the paper [6] considers that the dislocations are not waiting passively for the solutes to diffuse to them, but rather they capture the solutes during their bowing out process, under the applied stress, while being held at dislocation forest. The increase in the concentration of the dislocation core atmosphere is a result of the diffusion of the existing core atmosphere along with the dislocation, and the new solutes which are encountered and captured by the dislocation. Due to the increase in the core atmosphere density, both the total thermal activation energy and the mechanical threshold stress are increased. In an alloy the flow stress may be consider as a sum of two additive contributions: σ = σf + σd ,

(6)

with σf relating to a friction imposed by the solutes on the mobile dislocations, and σd relating to the dislocation-dislocation interaction.

498 400

240 200

σ02

σmax

200

n

n

300

-5 -1

2.2x10 s 5.6x10-5s-1 -4 -1 5.6x10 s

160 120

100

80

0 0

100

200

300

100

200

300

400

500

σ0 [MPa]

T [°C]

Figure 3: Temperature dependence of the stress sensitivity parameter obtained for AZ91 alloy.

Figure 4: Influence of the strain rate on the stress sensitivity parameter obtained for WE54 alloy.

The strain rate sensitivity of the flow stress has also two additive contributions: dσ f dσ = d ln ε& d ln ε&

+

d ln σ d d ln ε&

(7)

The strain rate sensitivity of the flow stress is done as

dσ d ln σ = σ = σ f m f + σ d md d ln ε& d ln ε&

(8)

Stress sensitivity parameter measured in the SR test is n = d ln ( −σ& ) /dlnσ = dln ε& /dlnσ n =

d ln ( −σ& ) d ln σ

=

σ f mf

σ . + σ d md

(9)

The strain rate sensitivity component mf has in the dynamic strain ageing regime an anomalous negative value [7] and then parameter n may have a local maximum in the temperature dependence of this parameter. Such maximum can be seen in Figure 3. As it is to see from Figure 4 parameter n depends in the DSA region of the strain rate. We can assume that during the stress relaxation test solute atoms diffuse to dislocations. Dislocations are pinned by these solute atoms. Therefore an increase in the stress is needed to move dislocations after reloading, i.e. the flow stress at the start of new deformation after stress relaxation is higher. The stress increment after reloading of the sample Δσ was measured for AZ91 alloy depending on the time for at starting stresses of the SR (Figure 5). The dependence of Δσ on the relaxation duration is well given by a power law with the slope s = 0.2, which is lower value in comparison with the prediction of the theory.

499 8

50°C 100°C 150°C reg.

10

σ0=157 MPa

Δσ [MPa]

Δσ/MPa

6 4 2

σ0=175 MPa reg. line

0

1 2

5

10

0

25

1

2

3

ε [%]

time/min

Figure 5: The time dependence of Δσ estimated for AZ91 alloy at room temperature and two starting stresses of the SR test.

Figure 6: Strain dependence of the stress increment obtained for AZ91 alloy and three temperatures.

The strain dependence of the stress increase Δσ was measured for AZ91 alloy and three temperatures (Figure 6) and for WE54 alloy at room temperature and three strain rates (Figure 7). The stress increment after SR is given by the relation [8]: Δσ(t,ε,T) = Δσm(ε,T) {1 – exp[-(t/tc)p]} ,

(10)

where Δσm(ε,T), the stress increment for t → ∞, grows with increasing size misfit of the solute and matrix atoms, increasing solute atoms concentration and lowering temperature. 16

5 Δσ [MPa]

12 Δσ [MPa]

6

2.2x10-5 5.6x10-5 5.6x10-4

8 4

4 ε = 0.5% ε = 1.5%

3

0 0

2

4

6

8

10

12

14

16

ε [%]

Figure 7: Strain dependence of Δσ for WE54 alloy -1 and three strain rates in s .

2 0

40

80

120

160

T [°C]

Figure 8: Temperature dependence of Δ obtained for AZ91 alloy.

tc is a characteristic time which depends on the strain as tc ~ ε-k . Power coefficient k was found to be for In based alloys k=1-2 [9]. Strain and temperature influence significantly the Δσ values as it can be seen from Figures 7 and 8. It is reasonable to expect that dislocation from more than one slip system form dislocation pile-ups. The strain ageing is due to migration of dissolved solute atoms towards the pile-ups. The total dislocation density is increasing with increasing strain while the number of the free solute atoms is constant. Therefore the strength of dislocation locking is smaller for higher strain degree. On the other hand Dlouhý, et al. [10] have shown that changes in the internal stress and the density of moving dislocations with the time of SR can cause the stress increment after the reloading. The stress increment is a complex function of the stress relaxation time.

500

Conclusions (1) Stress relaxation curves of Mg alloys AZ91 and WE54 were studied in the stress rep& ) vs ln σ plots. resentation i.e. ln(- σ (2) Stress sensitivity parameter n depends on the temperature and the strain rate. (3) A stress increase after stress relaxation was observed. (4) Observed post relaxation effect as well as anomalous temperature dependence of n are caused by the change in the concentration solute atoms on dislocations with the strain rate and temperature.

Acnowledgements The authors thank the Grant Agency of the Academy of Sciences of the Czech Republic for financial support under grant A2112901.

References [1] V.I. Dotsenko: phys. stat. sol. (b) 1979, 93, 11. [2] I.J. Polmear: Magnesium Alloys and Their Applications (Eds. B.L. Mordike and F. Hehmann) DGM, Oberursel, 1992, 201. [3] A.H. Cottrell, M.A. Jawson: Proc. Roy. Soc. 1949, A199, 104. [4] H. Yoshinaga, S. Morozuni: Phil. Mag. 1071, 23, 1367. [5] G.A. Malygin: phys. stat. sol. (a) 1982, 72, 493. [6] J. Cheng, S. Nemat-Nasser: Acta mater. 2000, 48, 3131. [7] S.I. Hong: Mater. Sci. Engn. 1085, 76, 77. [8] S.V. Lubenets, L.S. Fomenko: Fizika metallov metall. 1986, 62, 377. [9] S.V. Lubenets, V.I. Startsev, L.S. Fomenko: Czech. J. Phys. 1986, B36. 493. [10] A. Dlouhý, P. Lukáč, Z. Trojanová: Kovové mater.1984, 26, 688.

501

On the Reliability of Yield Strength Data Evaluated from as-cast Tensile Test Bars of Mg-alloys M. Wessén, H. Cao Jönköping University, Jönköping, Sweden

1

Introduction

Within the automotive sector, energetic efforts are today being made to decrease vehicle weight in order to minimize the emission of combustion gases. In the light of this work it is natural that the use of light metals, and especially magnesium, has increased considerably during the last 5-10 years. Some applications in cars where cast magnesium components are used today include seat frames, instrument panels, transfer cases, wheels, steering wheels and various kinds of housings. During the optimization process of new components, stress/strain analysis tools are usually used to calculate the stress state when the component is exposed to its working loads. In order to avoid any permanent plastic deformation it is of course necessary that the stress doesn’t exceed the local yield strength in any part of the component. It should also be emphasized that cast Mg-alloy components can have a very large variation in yield strength in different regions as a result of the varying cooling conditions and solidification microstructures [1-2]. Some materials have a distinctly defined yield point, e.g. most steels, whereas other materials start to plastify gradually at very low stresses. In the latter case the yield strength is most commonly defined as the stress when a permanent deformation of 0.2% remains after unloading; this is also often called the 0.2% proof stress. The plastic strain that occurs prior to the proof stress is often called “pre-yield micro strain”. Magnesium alloys usually exhibit this behaviour. As a consequence, it is difficult to define the Young’s modulus (E) in Mg-alloys, since E varies with stress at even moderate stresses. The elastic-plastic transition and the Young’s modulus in magnesium alloys AZ91 and AM60 (heat-treated samples) have been studied thoroughly by Cáceres et al [3]. When studying published stress-strain curves for cast Mg-alloys it seems that samples cast in sand moulds as well as heat treated samples do not show a well defined yield point (as discussed above), and that the curve shape is rounded in the elastic-plastic transition region; see figure 1 [4]. In the figure, curves from heat treated sand cast samples (aged at different times at 175 °C) are shown. In the same figure a curve from a 5 mm HPDC specimen is also shown for comparison. It can be observed that the 5 mm HPDC specimen differs from the other curves in terms of curve shape. Another clear example of this difference is seen in figure 2, showing some flow curves for AM-alloys at different Al-contents, published by Hydro Magnesium [5]. It is clear that these curves have a much more defined bend when the material starts to plastify. The reason for this very significant difference in initial flow behaviour has (to the knowledge of the authors) not been discussed in the literature. A tentative explanation for the differences could be related to mould constraints, which will give rise to some plastic deformation during cooling of the tensile test bars in the die.

502 If this plastic strain exceeds only about 0.2-0.5%, it is natural that the tensile test will give rise to a curve with a more significant bend than if no plastic deformation had taken place. If this is the case, it should be realised that published data for the yield strength evaluated from as-cast bars in metal dies is not necessarily representative for castings unconstrained during solidification. Consequently, an un-constrained casting may have lower yield strength than the published value, even though the microstructure is identical. In the present study, some initial experiments were performed to test this hypothesis. 400

FQ2

true stress (MPa)

300

42 h

120 h

16 h

4h T4

hpdc 5mm

200 ac

100

0 0.00

0.04 0.08 true plastic strain

0.12

Figure 1: Examples of true stress – true plastic strain flow curves of AZ91 alloy, as cast (ac), as quenched (T4) and aged for different times (h) at o 165 C. A flow curve of a die cast specimen (hpdc5mm) is included for comparison [4].

2

Figure 2: Stress-strain curves of AM alloys from HPDC as-cast tensile test bars [5].

Experimental procedure

In this study, magnesium alloy AM50 rods having a diameter of 10 mm and a length of 180 mm were produced by re-melting and solidification in a gradient furnace. The advantage of using this procedure is that the scale of the microstructure can be easily controlled simply by varying the drawing rate in the furnace. Additionally, a very low defect level is obtained, thereby resulting in reproducible samples with a very low scatter in results. The cast rods were then machined to the dimensions shown in figure 3. A more detailed description of the experimental procedure has been published elsewhere [2]. Tensile testing bars were also machined from a commercial automotive component of alloy AM50 produced by pressure die casting; see figure 4. Finally, some of the produced tensile testing bars were heated in a furnace to 410 °C for 10 minutes, and then fixed in a vice during cooling in order to simulate the mould constraint which exists in as cast tensile test bars in a metal die. All tensile testing was carried out on a Lloyd EZ50 tensile testing machine at a crosshead speed of 0.5 mm/min using an attached extensometer.

503

Figure 3: Dimensions of machined tensile test bar

Figure 4: Automotive casting, showing position from where the test bars were machined

3

Results

3.1

Tensile test bars from a commercial casting

It is relevant to study if the curve obtained in tensile testing a bar from a HPDC commercial casting has a similar shape to those shown in figure 2. In figure 5, the stress-strain curve from a test bar machined from the casting shown in figure 4 is shown (solid line), together with the curve from a test bar produced in the gradient furnace at a drawing rate of 6 mm/s (dashed line), which resulted in a microstructure similar to that of the casting. It is seen that they correspond well, and that the curve from the component does not have the well-defined bend as was obtained for the HPDC bars tested. Of course, it seems that there is also some kind of mould constraint in the automotive component (see figure 4). However, it is considerably lower than the extreme conditions in tensile test bar geometry.

Figure 5: Stress-strain curves for AM50 from commercial HPDC casting (solid line) and from sample produced in gradient furnace (dashed line)

3.2

Figure 6: Stress-strain curves for AM50 from samples produced in gradient furnace, and exposed to different degrees of pre-strain before tensile testing

Pre-straining of test bars in tensile testing machine

The simplest way to investigate the stress-strain curve shape after some degree of plastic deformation, such as that which could take place during cooling in the die, is of course to pre-strain the test bar and then unload it before the subsequent tensile testing. Test bars

504 were produced in the gradient furnace at a drawing rate of 6 mm/s. Three bars were prestrained to different deformations; 0.5 mm, 0.75 mm and 1 mm, and then drawn until failure. The resulting curves are shown in figure 6 together with a curve from a bar drawn without any pre-strain. The pre-strained curves have been displaced in the X-direction according to their respective plastic pre-strain. As expected, the pre-strained curves have a well-defined bend and after yielding they all fall on the curve for the non-pre-strained sample. Observe the striking similarity with the curves in figure 2. 3.3

Heating and subsequent fixing of tensile test bars during cooling

Another way to simulate what happens due to mould constraint during cooling in a metal die is to heat the samples and thereafter prevent the free shrinkage by fixing them in a vice after removal from the furnace. A drawback of using this method comes from the heat treatment of the microstructure which takes place during heating, and which would make any comparison with the original material impossible. This problem was avoided by heating two identical tensile test bars (gradient furnace, 6 mm/s) for 10 minutes at 410 °C, and then allowing one of the bars to contract freely, while the other was constrained during cooling. However, due to the small dimensions of the bars, the air cooling can be assumed to be rather quick, and therefore the temperature in the bar when it was fully fixed was probably considerably lower than 410 °C. The exact temperatures could not be recorded due to experimental difficulties. From the tensile testing curves of the “heated” and “heated and fixed” samples presented in figure 7, it becomes clear that the mould constraint has a significant influence on the yield strength of the material. The “heated and fixed” sample also has a significant bend in accordance with the curves from the HPDC tensile test bars in figure 2. A very rough estimate of the plastic strains induced during cooling, assuming full mould constraint, can be done by simply multiplying the coefficient of thermal expansion with the assumed cooling interval. Taking the initial temperature as 350 °C and the final temperature as 25 °C (room temperature) gives (300 – 25) * 26.1e-6 = 0.72 %, which is certainly significant, and which indeed should give a sharper yield-point on re-loading.

Figure 7: Stress-strain curves for AM50 from samples produced in gradient furnace. Samples were heated to 410 oC for 10 min prior to cooling with (dashed line) or without (solid line) contraction constraint.

505

4

Discussion

The hypothesis which initiated this work was based on the assumption that die constraint, which gives rise to plastic deformation in the casting during cooling, is the reason for the significant difference in stress-strain curve shape in the elastic-plastic transition region between sand castings and die castings. The initial tests performed all support this hypothesis. It is certainly true that plastic deformation taking place in castings during cooling in constrained moulds is not a newly discovered phenomenon; however, the consequence for the yield strength of the material has not been discussed. Part of the material’s elasticity is of course consumed, accompanied by a deformation hardening process. As a result it is natural that the actual stress when plastic deformation starts in a pre-strained component will be higher than what would be expected from a non pre-strained tensile test bar. Since most data for yield strength of Mg-alloys seem to be evaluated from as-cast HPDC tensile test bars, and that most castings don’t have such a high degree of mould constraint as the test bars, it is very likely that the real yield strength in the casting is lower than expected. From this follows also that the ejection temperature should have an influence on the yield strength; the lower the temperature is, the higher the yield strength as a result of more plastic deformation during cooling in the die. Nowadays, it seems that a majority of design engineers have accepted the fact that the mechanical properties of Mg-alloys vary in cast components as a result of the varying cooling conditions during solidification and subsequent cooling in the solid state. The reason is of course related to the different microstructures produced; i.e. mainly the amount, distribution and fineness of different phases. One complication which also has attracted significant attention during the last years is the fact that optimum mechanical properties, especially the tensile strength and elongation, are very rarely attained in commercial castings due to the existence of different kinds of defects, such as porosity, slag inclusions, segregation bands, hot tears etc. There is still, however, a major lack of knowledge about how much the property values are reduced as a function of size, amount and distribution of these structural defects. Now, considering that any plastic deformation of the casting during cooling in the die will also have an impact on the local yield strength in a component, it follows that any precise prediction of this property will require advanced simulation tools. Today, commercial tools are available which can be used to simulate microstructures and mechanical properties for some cast materials; e.g. MAGMAiron® [6] which for almost 10 years has been used industrially for cast irons. When such a tool becomes available for Mg-alloys it opens up the possibility to link these results with simulations of the stresses and strains produced during the casting process, thereby enabling accurate predictions of how the real yield strength varies in a complex casting. Finally, it should be emphasized that the phenomena discussed above relate not only to Mg-alloys. However, due to the large pre-yield micro strain for Mg as compared to e.g. Al-alloys, which have a more linear elastic behaviour, it is likely that the effect has most importance in Mg-alloy systems.

506

5

Conclusions

This study has gone some way towards understanding the complex interaction between casting process and the local yield strength which can be expected in cast Mg-alloys components. The experiments clearly indicates that any plastic deformation taking place during cooling in the die due to mould constraints, will lead to a higher evaluated yield strength from the stress-strain curve as compared to a non-constrained casting. As a consequence, there is reason to question the validity of published yield strength data for Mgalloys evaluated from HPDC as cast tensile test bars. The questions raised by this study create a need for further investigations.

6

Acknowledgements

The authors wish to express their sincere gratitude to Mr. Olof Granath and Mr. Sven Hedlund at the School of Engineering for invaluable help during the experiments, to Mr. Jan Isaksson and Hans Brantemo at Gjutal AB for valuable discussions and help, and last but not least VINNOVA, The Swedish Agency for Innovation Systems, for financially sponsoring the project.

7

References

[1] W.P. Sequeira, G.L. Dunlop, M.T. Murray, Proc. 3rd Intl. Magnesium Conf, Institute of Materials (UK), 1997, 63-73. [2] H. Cao, M. Wessén, Metallurgical and Materials Transactions, accepted 2003. [3] C. H. Cáceres, C J Davidson, J. R. Griffiths, R. Svensson, Proc. 33rd Australian Foundry Institute Convention, Noosa, Queensland, Oct. 2002. [4] C. H. Cáceres, J. R. Griffiths, C. J. Davidson, C. L. Newton, Mater. Sci. Eng. A, 2002, 325, 344-355. [5] T. K. Aune, H. Westengren and T. Ruden, Proc. Magnesium Properties and Applications for Automobiles, Society of Automotive Engineers, Inc. (USA), Technical Paper #930418, 1993, 51-57. [6] MAGMA GmbH, Aachen, Germany.

507

Understanding the Corrosion Mechanism: a Framework for Improving the Performance of Magnesium Alloys Guangling Song and Andrej Atrens CRC for Cast Metals Manufacturing (CAST), Division of Materials, The University of Queensland, Brisbane Qld Aus 4072

Abstract This paper provides a succinct but nevertheless complete mechanistic overview of magnesium corrosion. This provides an understanding of the types of corrosion exhibited by magnesium alloys, and also of the environmental factors of most importance. This deep understanding is required as a foundation if we are to produce more corrosion resistant alloys. This present analysis provides a theoretical framework for further research. There is still vast scope both for better understanding of corrosion processes as well as on the engineering usage of magnesium. We live in interesting times regarding the study of magnesium corrosion and expect to see significant advances in the next decade!

1

Introduction

Magnesium is a reactive metal and corrosion protection is of great importance [1, 2, 3] particularly for its greatest single market, the automobile industry [4] where there has been a 20% annual growth. Song and Atrens [1] and Ghali [2] have reviewed the corrosion of magnesium. This paper builds on the prior reviews and provides a succinct overview of the corrosion mechanism based on [5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15] and the literature. We provide a theoretical framework to facilitate understanding of the various corrosion processes. We start by outlining the key features of the corrosion mechanism. Then we outline how these key features have implications for all types of corrosion experienced by magnesium alloys. This mechanistic framework provides a basis for the understanding the following manifestations of magnesium corrosion. • The negative difference effect and hydrogen evolution. • General corrosion and localised corrosion. • Alloying influences on corrosion performance. • Possibilities for the corrosion performance of bulk amorphous magnesium alloys. • Measurement of the corrosion rate by the amount of hydrogen evolved. • Severe problems with the electrochemical methods for corrosion measurement. • Atmospheric corrosion. • Macro-galvanic corrosion when magnesium is coupled with a metal like steel. • Micro-galvanic corrosion associated with second phases. • Environment influences on the corrosion performance. • Stress corrosion cracking and corrosion fatigue.

508 This deep understanding is required to produce magnesium alloys much more resistant to corrosion than the present alloys. We provide a theoretical framework to inform and guide current and future research. This paper is a cut down version, the fuller version is to appear in Advanced Engineering Materials.

2

Corrosion Mechanism

The corrosion mechanism involves all physical and chemical features and processes when magnesium is exposed to an environment, as illustrated in Fig. 1. (a) At a potential more negative than Ep, a protective film covers the surface. (b) At a potential above Ep, the film is partially protective. Corrosion, at breaks in the surface film, occurs by means of the electrochemical partial reactions (1) and (2) and the chemical reaction (3). The product formation reaction, equation (5) repairs the film. (c) The chance development of localised corrosion leads to (d) & (e) the undermining and falling out of particles, even for the corrosion of pure magnesium. 2.1

Key points

i. A partially protective film covers the surface of magnesium. ii. Hydrogen evolution is intimately associated with magnesium dissolution in two separate ways. (a) An electrochemical reaction governed by equation (1) balances the magnesium dissolution reaction, equation (2). (b) Hydrogen is also produced directly in the reaction + of Mg with water, equation (3). Note that the overall reaction, equation (4) produces one molecule of hydrogen gas for each atom of magnesium dissolved. Furthermore, the overall reaction consumes H+ and produces OH-, i.e. the pH increases, which favours the formation of a magnesium hydroxide film by the precipitation reaction, equation (5). 2H+ + 2e = H2 2Mg = 2Mg+ + e 2Mg+ + 2H2O = 2Mg++ + 2OH- + H2 2Mg + 2H+ + 2H2O = 2Mg++ + 2OH- + 2H2 Mg2+ + 2OH- = Mg(OH)2

(cathodic partial reaction) (anodic partial reaction) (chemical reaction) (overall reaction) (product formation)

(1) (2) (3) (4) (5)

iii. Magnesium has a negative free corrosion potential, Ecorr, with a slightly more negative pitting potential, Ep, in solutions of practical importance like 5% NaCl. iv. The chance development of areas of localised corrosion leads to undermining and falling out of particles of magnesium, even for the corrosion of pure magnesium. 2.2

Implications

A. The negative difference effect (NDE) = i + ii + iv. This is shorthand for saying that NDE arrises from an interaction of key aspects i, ii and iv The partially protective film is potential dependent, in that there is essentially a complete film coverage over the whole surface and a low rate of corrosion for potentials below the pitting potential, at the pitting potential there are some film free areas, and the surface area free of surface film increases with increasing potential. As the potential increases above the pitting potential, there is an increasing surface area free of film on which reactions (1) and (2) can occur more freely than on the surface covered by a surface film. Furthermore, as

509 the potential increases above the pitting potential, the speed of reaction (3) is increased and concomitantly, there is an increase in the amount of hydrogen evolved. These two effects, together with particle undermining are the key elements of the negative difference effect. This effect is explained in more detail in the relevant section in the description of the corrosion of pure magnesium.

Mg+ H2

H2 O

Mg2+ + H2

RDS

H2 Mg

Mg

E>Ep (d) Fig. 1 The corrosion mechanism.

Particle undermining (c)

Mg

E ≥ Ep late stage or E>>Ep (e)

510 B. Localised corrosion = i + iv. This is shorthand for saying that localised corrosion arrises from an interaction of key aspects i and iv. The existence of a partially protective film, and the way chance determines the progression of corrosion means that “general corrosion” is not an issue. The common form of corrosion is localised corrosion, which for magnesium is much different to that of stainless steels C. Impurity tolerance limits = i + ii + iii. The lack of a surface film on the impurity particles together with the negative corrosion potential allows them to be efficient cathodes for hydrogen discharge, thereby providing significant micro-galvanic acceleration of corrosion. This is one forms of micro-galvanic corrosion. D. Influence of Al, Zr, Y… in improving corrosion performance = i + C. Improvement in the corrosion performance of single phase alloys can be expected from an improvement in the surface film if combined with removed or passivated impurities. E. Bulk amorphous alloys and better corrosion performance = i + C + D. A step improvement of corrosion performance is expected from a combination of a more uniform protective film, a higher tolerance limit and a better alloying effect. F. Corrosion rate measurement = ii + A. Hydrogen evolution as an integral part of reaction (4) provides an elegant method by which to measure the corrosion rate of magnesium alloys. The existence of the NDE means that there are severe problems with the electrochemical methods for corrosion measurement. G. Atmospheric corrosion = i + ii. The overall reaction, equation (4) tends to increase the local pH at cathodic sites. This tendency is particularly strong in a thin surface water layer as is often associated with atmospheric corrosion, in which it is relatively easy for reaction (4) to produce a high local pH value, which tends to facilitate the precipitation of a corrosion product film. Consequently, the surface film provides good protection in many cases of atmospheric exposure. H. Macro-galvanic corrosion = iii. Macro-galvanic corrosion occurs when magnesium is coupled with a metal like steel because Mg is the most active engineering metal and consequently the corrosion potential is more negative (i.e. more active) than all the other engineering metals. I. Micro-galvanic corrosion = iii. The anode matrix has a corrosion potential lower than that of the second phases, because each second phase is formed by the reaction of magnesium with a less reactive metal. J. Stress corrosion cracking (SCC) and corrosion fatigue (CF) = i + ii + iv. Material properties are changed by hydrogen entry into the magnesium matrix through a partially protective film. The localised corrosion associated with the chance development of corrosion leads to stress concentrations, and provides easy points for hydrogen entry in film free areas.

3

Corrosion of Pure Magnesium

Magnesium dissolution in aqueous environments proceeds by an electrochemical reaction with water to produce Mg(OH)2 and H2. Magnesium has a standard electrode potential of 2.37 Vnhe. The corrosion potential is ~ -1.7 Vnhe. Magnesium forms a partially protective magnesium hydroxide film. Thermodynamics shows that Mg++ is the stable species in most aqueous solutions up to a pH of about 10, above which Mg(OH)2 is the stable species. Magnesium has a strange phenomenon, the negative difference effect (NDE) for which Song et al [6] proposed a new mechanism, Fig. 1(b).

511

4

Localised Corrosion

For magnesium, corrosion typically takes the form of localised corrosion [5, 6] as indicted by the corrosion mechanism, Fig. 1. On the surface, there is a partially protective film. The corrosion potential is more positive than the pitting potential, for both single phase alloys as well as for two phase alloys. The localised corrosion initiates as irregular pits, which spread laterally and cover the whole surface. There is not much tendency for deep pitting. The reason is that the cathodic reaction, equation (1), produces OH- ions (or equivalently the consumption of H+ ions) with an increase of the pH, stablisation of the magnesium hydroxide film and a decrease in corrosion tendency. Thus localised corrosion in magnesium has an inherent tendency to be self-limiting. This is in marked contrast to the stainless steels, where the occluded pit cell becomes more aggressive and accelerates pitting. However, unlike stainless steels, the localised corrosion in magnesium spreads laterally. It is common for such corrosion to link up and undermine particles, which fall out of the surface, Fig. 1. This particle undermining is common even in the corrosion of pure magnesium.

5

Tolerance Limits

The influence of the contaminants (Fe, Ni, Cu & Co) is also explained by the corrosion mechanism, Fig. 1. The cathodic partial reaction, hydrogen evolution, is particularly important because of the negative corrosion potential. For each of these elements a tolerance limit can be defined. When the impurity content exceeds the tolerance limit, the corrosion rate is high, otherwise the corrosion rate is low. The various mechanistic possibilities are as follows. In each case, a key part of the corrosion acceleration is easy hydrogen evolution on a new phase in contact with the magnesium matrix. • accelerated galvanic corrosion of the magnesium alloy matrix due to general dissolution of the magnesium alloy followed by the re-precipitation of metallic Fe, Ni & Cu on the alloy surface. This mechanism is similar to the acceleration of corrosion by alloyed copper for aluminium alloys. • galvanic acceleration of corrosion when any of the impurity elements Fe, Ni or Cu exceeds its solubility limit and precipitates as a separate phase, δ. • the δ phase is an active cathode for some compositions, but is passive for others. For example, addition of Mn might cause the change from δ to δ’, such that δ’ has electrochemical parameters that make hydrogen evolution difficult on its surface. (Critical parameters to determine would be Ecorr, io and βc).

6

Alloying for Corrosion Resistance

The corrosion mechanism implies that there are two important factors for the influence of alloying: films and hydrogen evolution. Alloying elements that have been associated with increased corrosion resistance include (1) Al and (2) Zr in ingot alloys and (3) passivating elements like Ni and the rare earth elements (RE) Y and Nd in amorphous alloys. Any beneficial influence of these alloying elements has been attributed to improving the protective properties of the surface film. It is particularly interesting that Ni may be beneficial in large concentrations in an amorphous alloy whereas there is the major detrimental influence of Ni in ingot alloys.

512 6.1

Ingot Alloys

The influence of aluminium is difficult to disentangle from the influence of increasing amounts of the second phase. Data [16] for the alloys of the ASxx, AMxx and AExx alloy systems shows a decrease in the corrosion rate with increasing Al content. However, these alloys are two phase and for aluminium contents above 2% there is an increase in the amount of the beta phase. Song et al [12] studied Mg-Al alloys heat treated to be single phase and also found a beneficial influence of aluminium. The mechanism by which Zr improves corrosion resistance was investigated by Song and StJohn [14] using two versions of the alloy MEZ, containing Zr grain refiner (MEZR) and without Zr additions (MEZU). MEZR contained 0.004% Fe and 0.6% Zr whereas MEZU contained 0.013% Fe and 0.005% Zr. This illustrates that one influence of Zr was to react with Fe in the melt, which removed Fe from the alloy and produced a “high purity” alloy. This was part of the explanation for the improvement in the corrosion performance. There was also an additional effect of the second phase, which was more corrosion resistant than the matrix phase. The second phase contained substantial amounts of Zr and RE elements, and the increase in corrosion performance was attributed partly to a more protective film. 6.2

Amorphous Alloys

The influence of passivating elements has been studied by [17] using alloys produced by the rapid solidification (RS). This work showed that Mg65Cu25Y10, Mg65Ni20Nd15 and Mg82Ni18 exhibited passivity over a considerable potential region. However, they were not passive in more concentrated solutions of more practical importance. Their bulk corrosion behaviour showed a similar trend, from Table 1. Table 1. Corrosion rates of selected Mg alloys.

Alloy High purity Mg Surface of die-cast AZ91D β-phase Mg17Al12 RS Mg65Ni20Nd15 RS Mg79Cu21 RS Mg65Cu25Y10 RS Mg67Ni18Nd15 6.3

Environment 1M NaCl 1M NaCl 1M NaCl 0.01M NaCl pH12 0.01M NaCl pH12 0.01M NaCl pH12 3% NaCl pH12

CR, μm/y 1,140 660 300 2.5 410 250 750

Ref [1] [1] [1] [17] [18] [18] [19]

Strategy to Improve the Corrosion Behaviour of Mg Alloys

A step improvement is to be expected for a Mg-X (or Mg-A-B-X) alloy where (1) the alloying element(s) is (are) in solid solution, and (b) the alloy forms a stable passive film based on an oxide of X rather than the Mg-based oxide/hydroxide usually formed on Mg alloys [7]. This is the approach in Fe-Cr alloys, the basis of stainless steels. In Fe-Cr alloys, the nature of the film changes from one based on Fe to a film based on Cr (see eg [20, 21]). This change in film characteristics occurs at about 12% Cr, and concomitantly the corrosion resistance increases by orders of magnitude. The corrosion behaviour is similar to that of Cr, even though there may be only 12% Cr in the alloy. It is suggested that the most exciting passivating elements to study for Mg-base alloys are Cr, Ti and Si. Cr is well known for its passivating properties in highly corrosion resis-

513 tant stainless steels (Fe-Cr alloys). Similarly Ti forms corrosion resistant Ti alloys and Si is important in imparting high corrosion resistance in high Si cast irons. The alternatives Ta, Zr, Hf, Al or Sc are not considered because of various obvious reasons. Production of Mg alloys of macroscopic size containing significant amounts of Cr, Ti or Si in complete homogeneous solid solution has become possible using bulk-amorphous alloys. As it is easy to produce Mg65Cu25Y10 as an amorphous alloy, it is probable that it is possible to produce amorphous versions of Mg65Cr25Y10, Mg65Cr25Ca10, Mg65Ti25Ca10 or Mg55Li20B10Si10C5. Such alloys are at the centre of this suggestion but have not been produced or studied. Such alloys enable the study of the influence of Cr, Ti and Si. A twostep approach is suggested. Step one should explore the properties of such alloys. If indeed there is a step increase in corrosion performance (as might be expected in analogy with the stainless steels) then step two. Step two is the optimisation of other properties, for example decreasing the density, mechanical properties, glass forming ability, processing etc.

7

Measurements of the Corrosion Rate

The corrosion mechanism, Fig. 1, has implications for the measurement of corrosion of magnesium alloys. In particular, (1) there are severe problems with the electrochemical methods for measurement of the corrosion rate, and (2) there is a particularly elegant and easy manner to measure the corrosion rate of magnesium by means of the amount of hydrogen evolved [11]. The principle is based on the overall corrosion reaction, equation (4), which indicates that the dissolution of one atom of magnesium generates one molecule of hydrogen. Therefore, the measurement of the hydrogen evolved is equivalent to the measurement of the weight loss of metal, provided both are converted into the same units. Reaction (4) indicates that the hydrogen evolution is determined by the dissolution of magnesium. The corrosion products do not influence the relationship. The hydrogen evolution directly reflects the corrosion rate of magnesium. Therefore, this method to measure the corrosion rate should be reliable, easy to use and free from many of the errors inherent in the measurement of corrosion by weight loss. A simple way in estimating the error for magnesium alloys is to take the simplest view and assume that the alloy dissolves homogeneously. As all magnesium alloys have small amounts of alloying, less than 10%, the error must be 10% or less, on the basis that the alloying elements do not themselves corrode with the concomitant liberation of hydrogen. Since the corrosion reaction for many of the alloying elements, like Al, is analogous to that for magnesium, the error will in practice be much less Comparison of the corrosion rates of various specimens measured by weight loss and by the hydrogen evolution method for various magnesium alloys, including pure magnesium, single-phase alloys and various forms of AZ91, showed that there was satisfactory agreement between the two methods [11]. This method also has the advantage that the change in corrosion rate can be monitored. This allows the study of how the corrosion rate changes with such variables as time of exposure and changing exposure conditions. The problems associated with the electrochemical methods of measuremen was illustrated by [11], which presents polarisation curves for MEZ and AZ91D in 5% NaCl (pH11). Tafel extrapolation (for both alloys of both the anodic and cathodic polarisation curves) would indicate almost identical corrosion rates. In contrast weight loss measurements indicated that the ratio of weight loss for these two alloys was MEZ:AZ91D = 25:1.

514 The reason that the Tafel extrapolations did not give correct estimates was that the corrosion of magnesium violates the requirements of the Tafel extrapolation method. The same reasons also apply to the use of polarisation resistance.

8

Atmospheric Corrosion

Overall magnesium alloys have a corrosion resistance that is not as good aluminium alloys, because the surface films on the surface of magnesium alloys are not as stable as those on aluminium alloys. This has restricted the application of unprotected magnesium alloys to benign exposure conditions like indoor (and outdoor) atmospheric exposures. It should nevertheless be emphasised that for these environmental conditions, modern magnesium alloys have corrosion rates lower than those of comparable aluminium alloys, and much lower corrosion rates than carbon steel.

9

Macro-Galvanic Corrosion

The active nature of magnesium means that galvanic effects are an issue. Magnesium is more active than all engineering alloys. Consequently magnesium is the anode and corrodes preferentially in any galvanic couple [15].

10 Micro-Galvanic Corrosion and Second Phases The possible types of behaviour are summarised by the data given in Table 2 for magnesium alloys corroding at their free corrosion potentials in 1N NaCl at pH11. High purity magnesium, taken as the standard, showed a corrosion rate of 1.1 mm/y. A higher corrosion rate was shown by the interior of die cast AZ91D and by the high-purity sand-cast AZ91. The β phase accelerated the corrosion. In contrast, the surface of die cast AZ91D had a corrosion rate lower than that of high purity magnesium. The β phase provided protection as is clear from the still lower corrosion rate shown by pure β. The corrosion rate of the β phase is nevertheless high. For aluminium alloys the corrosion properties are severely degraded by this same phase. This points to the methodology by which to increase the corrosion resistance of magnesium alloys by the alloying to produce a radically different second phase, the Ζ-phase. Much better corrosion becomes possible if this new second phase, the Ζ-phase has a corrosion resistance greater than the β phase. Such a Ζ-phase should be more passive than the β phase and also have a much lower ability to allow hydrogen evolution on its surface. Table 2. Corrosion rates under open circuit conditions in 1N NaCl [18].

Sample LP Mg (240 ppm Fe) HP AZ91 sand cast AZ91D – die cast (interior) High purity Mg AZ91D – die cast (surface) β Phase the Ζ-phase

Rate, mm/y 53 12 5.7 1.1 0.66 0.30 ?? 10-6 ??

Mechanism/comment Impurity accelerated corrosion β accelerates corrosion β accelerates corrosion Standard for comparison β protects Significant corrosion rate Needs to be more passive than the β phase and have a lower ability to liberate hydrogen on its surface

515

11 Environmental Influences The various environmental influences may be understood using the principles of the corrosion mechanism, Fig. 1. Corrosion needs a liquid water phase, so the corrosion rate is low in dry air, dry gasses, and in many organic solvents. Corrosion rates, in aqueous solutions, are low under conditions which promote the formation of a stable corrosion product film: e.g. (1) local conditions of pH above 10.5 (to produce a film of Mg(OH)2), including stagnant pure water, (2) fluoride containing solutions (insoluble MgF2 film), (3) MgSO4 in concentrated sulphuric acid, (4) oxidising conditions free of film breakdown agents like chlorides (eg a boiling solution of chromic acid) can lead to stable oxide/hydroxide films. In contrast, high corrosion rates are experienced under conditions leading to surface film beak-down (e.g chloride solutions). Corrosion rates are also high for magnesium in galvanic contact with heavy metals, or for magnesium with surface contaminated with a heavy metal (eg a magnesium surface shot blasted with iron shot), or magnesium in contact with a solution of heavy metal salts.

12 Stress Corrosion Cracking The corrosion mechanism indicates that stress corrosion cracking (SCC) is a concern. Moreover, there is the expectation that environment assisted cracking (EAC) (including SCC and hydrogen embrittlement (HE)) failures will increase with increased use of Mg alloys in load bearing applications. Magnesium alloys are known to be susceptible to EAC. For example, the authoritative ASM Handbook [22], Vol 13, “Corrosion” on SCC of magnesium alloys can be summarised as follows. Magnesium alloys containing more than 1.5% Al are susceptible to SCC. Wrought alloys appear more susceptible than cast alloys. While there is little documentation of service SCC of castings, laboratory tests can cause SCC at tensile loads less than 50% of yield stress in environments causing negligible corrosion. The low incidence of service SCC failures can be attributable to low actual stresses [22] in service in the past. However, the incidence of EAC is expected to increase because the service conditions for Mg alloys are changing, particularly in the automobile industry where magnesium components are being increasingly used in structural load-bearing applications. Furthermore, increased loading on these magnesium components is a natural progression as designers increase the loads and decrease the section sizes of components as part of the environment imperative to decrease weight. Moreover, the magnesium components in the load bearing applications are increasing in complexity, and this increasing complexity increases the probability of high loads in some parts of such components. Unanswered Issues • Are modern AMxx and Zr containing magnesium casting alloys in practice free of EAC as suggested by the early (uncorroborated) reviews in 1966? • Design allowables for safe service of magnesium alloys, and how these are controlled by the alloy and by mechanical and environment loading conditions. • Mechanical and metallurgical conditions leading to EAC for commercial cast magnesium alloys. How the conditions for EAC (initiation stress, initiation stress intensity factor, crack velocity) depend on (1) alloy chemistry variables (major alloying components such as Al, Zn, Y, RE); and (2) minor additions such as Zr, Zn, Mn, Si, RE in AZxx, AMxx, ASxx, ZExx and WExx alloys, (3) microstructural features such as volume fraction, distribution and electrochemical properties of second phases, particularly continuous eutectic phases, and (4) slip type in the matrix phase, at the isolated second phase particles and at the eutectic.

516

•

Environmental/solution/temperature conditions leading to EAC for commercial cast magnesium alloys.

13 Concluding Remarks This paper has built on the prior reviews and has provided a succinct overview of the corrosion mechanism. This understanding of the corrosion mechanism provides a basis for the understanding the various manifestations of corrosion of magnesium alloys, and provides a foundation on which to develop alloys with much greater corrosion resistance.

14 Acknowledgements The CRC for Cast Metal Manufacturing (CAST) was established under, and is supported in part by the Australian Government’s Cooperative Research Centres Scheme.

15 References [1] G Song and A Atrens, Advanced Engin Materials 1 (1999) 11-33. [2] E Ghali, in Uhlig’s Corrosion Handbook, ed RW Revie, Wiley-Interscience, (2000) p793-830. [3] Y Kojima et al eds, Magnesium Alloys 2003, Materials Science Forum 419-422 (2003)1-1037. [4] GS Cole, Materials Science Forum 419-422 (2003) 43-50. [5] G Song, A Atrens, S StJohn, J Nairn and Y Lang, Corrosion Science 39 (1997) 855. [6] G Song, A Atrens, D StJohn, X Wu and J Nairn, Corrosion Science 39 (1997) 19812004. [7] G Song, A Atrens, X Wu and B Zhang, Corrosion Science 40 (1998) 1769-1791. [8] G Song, A Atrens and M Dargusch, Corrosion Science 41 (1999) 249-273. st [9] G Song, Corrosion Science in the 21 century UMIST 7-11 July, 2003. [10] G Song and DH StJohn, Journal of Light Metals 2 (2002) 1-16. [11] G Song, A Atrens and DH StJohn, in J Hryn ed, Magnesium Technology 2001, New Orleans, TMS (2001) 255-262. [12] G Song, AL Bowles and DH StJohn, Mater Sci and Engineering, accepted for publication [13] G Song and A Atrens, Wolfsburg (1998) 415. [14] G Song and DH StJohn, J Light Metals 2(2002)1. [15] G Song, B Jonhannesson, Sarath Hapugoda, DH St John, Corrosion Science, in press. [16] O Lunder, K Nisanccioglu and RS Hansen, Paper No 930755SAE, Detroit MI, (1993) 117. [17] HB Yao, Y Li and ATS Wee, Electrochim Acta 48 (2003) 2641 [18] HB Yao, Y Li, ATS Wee, JS Pan and JW Chai, Surface Review Letters 8 (2001) 575. [19] HB Yao, Y Li, ATS Wee, JS Pan and JW Chai, Applied Surface Science 158 (2000) 112. [20] P Bruesh, K Miller, A Atrens and H Neff, Appl Phys A 38 (1985) 1-18. [21] S Jin and A Atrens, Applied Physics A 42 (1987) 149-65. [22] ASM Handbook, Vol 13, Corrosion ASM International, fourth printing (1992).

517

Potentiodynamic Studies of Some AZ and ZA Magnesium Alloys in Different Corrosive Media Mustafa Özgür Öteyaka, Anne-Marie Lafront, Edward Ghali and Réal Tremblay Université Laval, Québec, Canada

1

Introduction

It is well documented that the presence of heavy metallic impurities like Fe, Ni and Cu represents the most detrimental factor influencing the corrosion properties of commercial magnesium alloys [1-3]. The corrosion behaviour of cast Mg-Al alloys could depend considerably on the microstructure (presence of β-phase and impurities) and the environment to which they are exposed. Earlier investigation on the corrosion behaviour of AZ91 alloys proved that the β-phase could play a dual role in the dissolution behaviour since it can act as either a galvanic cathode or a kinetic barrier to dissolution [4-7]. Changes in aluminum and zinc contents and/or addition of other elements in magnesium alloys tend to modify significantly the microstructure and influence notably the corrosion resistance in specific media. In this work, electrochemical techniques were used to compare the corrosion behaviour of two groups of magnesium alloys: 1/AZ91D and AZ91E, and 2/ZA104, ZAC10403 (0,3% Ca), and ZACS1040305 (0,3% Ca and 0,5% Sr) in diverse conditions and media.

2

Experimental Procedure

2.1

Material used and specimen preparation

Specimens were prepared from cast parts of two groups of magnesium alloys. The AZ group had aluminum as principal alloying element (~9,0% Al) whereas the ZA group had zinc (~10,0% Zn). The AZ91D, AZ91E, ZA104, ZAC10403 and ZACS1040305 alloys were cast in permanent mould in accordance to the standard ASTM B93 [8]. The chemical composition of each alloy is given in Table 1. Each corrosion specimen consisted of 1 cm x 1 cm x 2 cm metallic piece which was encased in epoxy resin. The surface analyzed was polished with 600 and 1200 grit SiC abrasive papers, cleaned with alcohol and then dried before immersion in the test solution. Table 1. Chemical composition (in wt%) of the experimental magnesium alloys.

Alloys AZ91D AZ91E ZA104 ZAC10403 ZACS1040305

Al Zn 9,00 0,67 8,70 0,70 3,69 9,73 3,78 10,50 4,12 11,70

Ca ------0,31 0,31

Sr --------0,54

Mn 0,33 0,26 0,25 0,03 0,35

Si 0,010 0,200 0,015 0,001 18 ppm

Fe 0,005 0,005 0,030 0,003 189 ppm

Cu 0,030 0,015 < 0,001 < 0,001 2 ppm

Ni 0,002 0,001 0,001 0,002 2 ppm

Mg bal. bal. bal. bal. bal.

518 2.2

Testing corrosive solutions

Three different corrosive solutions were used for this study with or without agitation: 5% NaCl solution saturated with Mg(OH)2 at pH 9, distilled water saturated with Mg(OH)2 solution at pH 9, and 0,01M NaOH solution at pH 12. 2.3

Electrochemical technique parameters

The standard ASTM G5 was used to carry out potentiodynamic tests at 25 °C [9]. The potentiodynamic curves were traced at 0,1 mV/s or 0,07 mV/s scanning rate over a potential range of 1000 mV starting from the cathodic end using a potentiostat/galvanostat (Model 273 EG&G). For some experiments, the solution was stirred at 250 rpm and heated at 75 °C. Also, other scanning rates, potential ranges and cathodic polarization voltages were used. After immersion of the sample into the solution, the test was started within one minute. All potentials were given with respect to a silver chloride reference electrode Ag, AgCl/KClsat. (0,222 V vs. SHE). The reproducibility of the corrosion potential and the corrosion current values were ±10-15 mV and ±10-15% respectively.

3

Results and Discussion

In this paper, the following terms Ecorr., icorr., Ep, ip, Ecp, and icp correspond to corrosion potential, corrosion current, pitting potential, pitting current, critical passivation potential, and critical passivation current respectively. The polarization curve was started from the cathodic region after a stabilization period of one minute in the solution. The corrosion potential was determined from the potentiodynamic polarization curves, using dedicated software. It is arbitrarily admitted that the two parameters Ecp and icp can describe passivation in spite of the observed partial passivation or pseudo-passivation of the magnesium alloys in the examined media. Pitting potential was determined from the potentiodynamic curves at the rapid increase of current density in some potentiodynamic curves. 3.1

Influence of different aqueous media at 25 °C

Results from potentiodynamic studies in the three corrosive media with and without stirring for magnesium alloys are given in Table 2. In NaCl solution, the potentiodynamic tests for ZA104, ZAC10403, and ZACS1040305 alloys exhibited a passivation zone with formation of a black corrosion product acting as a barrier layer followed by dissolution of the surface. But AZ91D and AZ91E alloys did not present a good passivation zone and dissolution of the surface is accompanied by the formation of pits that have been observed microscopically. Active-passive parameters of magnesium alloys in a strong corrosive medium (5% NaCl solution) cannot be determined precisely from potentiodynamic measurements. The values of corrosion potential as determined from the polarization curves show that they are close to each other. However, if the potential of AZ91D is compared to that of ZACS1040305, it can be stated generally and statistically that the alloy containing higher quantities of Zn and small quantities of Ca and Sr have more active potentials. The minimum corrosion current rates are found for the AZ91E followed by AZ91D and somewhat inferior to that of ZAC10403 and ZACS1040305. This is not surprising since AZ91E (with less impurity) has better corrosion resistance than AZ91D. Typical corrosion

519 rates of AZ91E in salt fog have been already reported to be ~0,25 mm/year and this is about a hundred times lower than that of AZ91C standard alloy [2]. Table 2. Data from the potentiodynamic curves obtained at 0,1 mV/s scanning rate and 25 °C in different media without and with (values in parentheses) stirring for magnesium alloys.

icorr. Ep ip Ecorr. (V) (μA/cm2) (V) (μA/cm2) 5% NaCl solution saturated with Mg(OH)2 at pH 9 AZ91D -1,31 (-1,30) 3 (3) ----AZ91E -1,30 (-1,32) 2 (1) ----ZA104 -1,32 (-1,33) --- (3) ----ZAC10403 -1,32 (-1,32) 5 (2) ----ZACS1040305 -1,33 (-1,32) 4 (---) ----Distilled water saturated with Mg(OH)2 solution at pH 9 AZ91E -1,29 (-1,29) 0,1 (0,1) -0,89 (-0,72) 25 (10) ZACS1040305 -1,22 (-1,26) 0,2 (0,2) ----0,01M NaOH solution at pH 12 AZ91E -1,34 (-1,27) 0,1 (0,02) -0,73 (-0,2(1)) 15 (---) ZA104 --- (-1,16) --- (0,03) --- (-0,30) --- ( 5 ) ZAC10403 --- (-1,26) --- (0,03) --- (-0,32) --- (34) ZACS1040305 -1,29 (-1,11) 0,6 (0,01) -0,82 (-0,2(1)) 23 (---) Alloys

(1)

Ecp (V)

icp (μA/cm2)

-----------

-----------

-1,20 (-1,19) 18 (10) -----1,21 (-1,19) --- (-0,99) --- (-0,98) -1,22 (-0,95)

16 ( 2 ) --- ( 5 ) --- (23) 15 (3)

More positive than -0,2 V.

Figure 1 shows the anodic polarization curves associated with some scanned surface views of AZ91E, ZAC10403, and ZACS1040305 alloys. For AZ91E alloy, the pits appear at the beginning of the anodic polarization; they expand and cover large spots on the surface with the increase of the applied current (Figures 1a and 1b "A" and "B"). ZAC10403 and ZACS1040305 alloys exhibit the same corrosion behaviour, at the beginning of anodic polarization, round corrosion products appear on the sample surface (Figures 1a and 1b "D"). However at higher applied current densities, the corrosion products involve their surfaces (Figure 1b "E"). This corresponds to the passive zone on polarization curves which is able to prevent the corrosion rate increase of the alloy.

Figure 1. a) Anodic potentiodynamic curves of the AZ91E, ZAC10403 and ZACS1040305 alloys at scan rate of 0,07 mV/s in the 5% NaCl solution saturated with Mg(OH)2 at pH 9 and 25 °C (w/o stirring); b) surface views at different potential-current points for AZ91E (A and B) and ZAC10403 (D and E) specimens.

520 At 1000 mV potential range, a passive zone and a clear pitting corrosion potential for AZ91E alloy are observed in Mg(OH)2 solution. The alloy ZACS1040305 does not show an active-passive behaviour in Mg(OH)2 solution as seen in NaCl solution. Corrosion rate of ZACS1040305 is two times higher than that of AZ91E. It can be postulated then that the better performance of AZ91E is due to the formation of a better passive layer containing magnesium hydroxide as a barrier that obstructs dissolution [1]. The poor performance of ZACS1040305 alloy in this solution is probably due to the galvanic effect of the intermetallic phases (MgxZnyAlz) with magnesium matrix. The potentiodynamic curves of AZ91E and ZACS1040305 alloys in NaOH solution have shown good passivation zone and an obvious pitting corrosion potential for both alloys. However, it seems that the aptitude to passivation (Ecp - Ecorr.) for AZ91E is more difficult in NaOH solution at pH 12 than in Mg(OH)2 solution at pH 9 although the icp does not give the same trend and the quality of passivation is better. Both alloys have a corrosion potential more negative than that in Mg(OH)2 solution (Table 2). The alloy AZ91E demonstrates a 50 mV more active potential than ZACS1040305, while the corrosion rate of this latter is six times higher than that of AZ91E in NaOH solution. The corrosion rates of AZ91E remains constant in both solutions, while corrosion rate for ZACS1040305 increases three times in NaOH solution as compared to Mg(OH)2 solution. The corrosion rates of AZ91E are similar in the Mg(OH)2 and NaOH solutions but much less than that in NaCl solution. It seems that AZ91E has a better corrosion resistance in Mg(OH)2 because of its aluminum content (9%). The ZACS1040305 experimental alloy has more noble corrosion potential and less corrosion rate in Mg(OH)2 than in NaOH and NaCl solutions. This may indicate that ZACS1040305 corrosion resistance is increased more by the existence of the Mg(OH)2 at the interface than by a more alkaline pH [4,10-12]. 3.2

Influence of agitation

Corrosion behaviour is observed for the two groups of magnesium alloys in non-stirred and stirred NaCl solutions (Figures 1 and 2). In general, the corrosion potentials had respectively similar values (Table 2).

Figure 2. Potentiodynamic curves for magnesium alloys at scan rate 0,1 mV/s in 5% NaCl saturated with Mg(OH)2 stirred solution at pH 9 and 25 °C.

521 For AZ91E alloy, stirring shows an improving effect of the passivation zone in Mg(OH)2 solution. Also it is found that the corrosion potential is equal in stirred and nonstirred solutions. Similar corrosion rate values are found in the stirred and non-stirred solutions for both alloys (Table 2). For ZACS1040305 alloy, a weak passivation zone is observed in stirred solution. For the two groups of alloys, agitation shows a favorable influence on the passivation zone in NaOH solution. The inferior passive domain of ZAC10403 (660 mV) compared to ZA104 (690 mV) explains the no good effect of 0,3% Ca on passivation. The strontium associated with calcium has more evident influence on passivation than calcium alone. The alloy AZ91E shows a better aptitude to passivation (Ecp - Ecorr.) with a value of 80 mV compared to the group of alloys containing more zinc (Table 2). The corrosion rate (icorr.) of AZ91E and ZACS1040305 alloys in the stirred NaOH solution is generally much less important than that in the same non-stirred solution. In the group of ZA alloys, the corrosion rate of ZA104 and ZAC10403 is three times higher than that of ZACS1040305 in stirred NaOH solution. 3.3

Influence of temperature

In NaCl solution at 75 °C, AZ91E alloy presents an active-passive behaviour that is not observed clearly at 25 °C and dissolution of the surface is accompanied by pitting at both temperatures (Figure 3). The active-passive behaviour at elevated temperature is due probably to the formation of a more stable film on β-phase (Mg17Al12) because of the presence of higher amount of aluminum. The continuous and stable oxide film acts as barrier so that the dissolution of α-phase (α-Mg) is inhibited. At 25 °C, the ZACS1040305 alloy shows a relatively better active-passive behaviour than that at 75 °C. Increasing the temperature drops the corrosion potential of the alloy AZ91E towards more negative values (Δ = 40 mV). However, the corrosion potential of the ZACS1040305 alloy shifts in the opposite direction and becomes 30 mV nobler. At 75 °C, the corrosion potential of the alloy ZACS1040305 is 40 mV nobler than AZ91E alloy. The corrosion rate of the alloy ZACS1040305 at elevated temperature is half of that at ambient temperature (Table 3). Table 3. Data from the potentiody-namic curves presented in Figure 3.

Figure 3. Potentiodynamic curves for different alloys at scan rate 0,1 mV/s in 5% NaCl solution saturated with Mg(OH)2 at pH 9 and 25 °C - 75 °C (w/o stirring).

Alloys

Ecorr. (V)

icorr. (μA/cm2)

AZ91E (25 °C) AZ91E (75 °C) ZACS1040305 (25 °C) ZACS1040305 (75 °C)

-1,30 -1,34 -1,33

2 3 4

-1,30

2

522 It can be postulated that increasing temperature for the AZ group of alloys shifts the corrosion potential towards more negative values, increases the corrosion rate, and favours the active-passive behaviour. An opposite tendency is observed for the ZA alloys for which a shift of the corrosion potential to more noble values, a decrease of the corrosion rate and disappearance of the active-passive behaviour appear with an increase of the temperature. 3.4

Influence of scan rate

ZACS1040305 alloy was tested at a cathodic range of -40/250 mV for each scan in NaCl solution at 25 °C without stirring. The passivation zone appears more visible with a low scan rate but it is less evident at 0,5 mV/s (Figure 4). At low scan rate of 0,01 mV/s, the best aptitude to passivation is observed. Also the critical passivation current cited in Table 4, shows that the aptitude to passivation is the best at 0,01 mV/s (4100 μA/cm2). It is recommended to choose a scan rate between 0,01 and 0,1 mV/s for the alloy ZACS1040305 to perform the potentiodynamic tests, although the polarization studies for AZ91E alloy show very similar curves at the three scan rates used.

Figure 4. Potentiodynamic curves for ZACS1040305 alloy at different scan rates in 5% NaCl solution saturated with Mg(OH)2 at pH 9 and 25 °C (w/o stirring). Table 4. Data from the potentiodynamic curves presented in Figure 4.

Scan rate (mV/s)

Ecorr. (V)

icorr. (μA/cm2)

Ecp (V)

icp (μA/cm2)

0,01 0,1 0,5

-1,30 -1,34 -1,30

--1,3 ---

-1,30 -1,29 ---

4100 15000 ---

4

Conclusions

The electrochemical studies prove that, without agitation, the corrosion rates of AZ91E are similar in distilled water saturated with Mg(OH)2 at pH 9 and 0,01M NaOH at pH 12 solutions but much less than that in 5% NaCl solution saturated with Mg(OH)2 at pH 9.

523 The ZACS1040305 experimental alloy has more noble corrosion potential and less corrosion rate in Mg(OH)2 solution than in NaOH solution and 5% NaCl solution saturated with Mg(OH)2 . For both groups of alloys, stirring seems to have no effect on corrosion behaviour in 5% NaCl solution saturated with Mg(OH)2, but helps to increase the passive domain of the alloy AZ91E in the Mg(OH)2 solution. Generally for AZ91E and ZACS1040305 alloys in NaOH solution, agitation decreases the corrosion rate and critical passivation current and increases the passive domain. For the AZ alloys, increasing temperature shifts the corrosion potential towards more negative values, increases the corrosion rate, and favors the active-passive behaviour in 5% NaCl solution saturated with Mg(OH)2. However, the opposite tendency is observed for the ZA alloys. It is also recommended to choose a scan rate between 0,03 and 0,1 mV/s for the alloy ZACS1040305 to perform the potentiodynamic tests in 5% NaCl solution saturated with Mg(OH)2. For the studied compositions of ZA alloys, it should be noted that the addition of Sr associated with Ca improves the corrosion resistance better than the addition of Ca alone. Finally, increasing the Zn content and lowering the Al content in ZA alloys relatively to the AZ group do not seem to improve the corrosion resistance.

5

Acknowledgement

This work was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC).

6 [1]

References

M. M. Avedesian, H. Baker, Magnesium and Magnesium Alloys, ASM Specialty Handbook, OH, USA, 1999. [2] K. Nisancioglu, O. Lunder, T. K. Aune, Inter. Magnesium Ass., 1990, 43-50. [3] J. D. Hanawalt, C. E. Nelson, J. A. Peloubet, Trans. AIME, 1942, 147, 273-299. [4] H. Alves, U. Koster, E. Aghion, D. Eliezer, Materials Technology, 2001, 16, 2, 110-126. [5] R. Ambat, N. N. Aung, W. Zhou, Corrosion Science, 2000, 42, 1433-1455. [6] G. Song, A. Atrens, M. Dargusch, Corrosion Science, 1999, 41, 249-273. [7] O. Lunder, J. E. Lein, T. Kr.. Aune, K. Nisancioglu, Corrosion, 1989, 45, 741. [8] ASTM Standard B93/B93M-94b, Annual Book of ASTM Standards, 02.02, 1997, 46-49. [9] ASTM Standard G5-94, Annual Book of ASTM Standards, 03.02, 1997, 54-64. nd [10] P. Uzan, N. Frumin, D. Eliezer, E. Aghion, Proc. 2 Israeli Inter. Conf. on Magnesium Science & Technology, MRI, Beer-Sheva, Israel, 2000, 385-391. [11] O. Lunder, M. Videm, K. Nisancioglu, Corrosion Resistant Magnesium Alloy, SAE Technical Paper 950428, Detroit, USA, 1995, 57-62. [12] O. Lunder, K. Nisancioglu, R. S. Hansen, Corrosion of Die Cast MagnesiumAluminum Alloys, SAE Technical Paper 939755, Detroit, USA, 1993, 117-126.

524

Plating of Magnesium – New Developments Peter Gregg Franz Oberflächentechnik GmbH & Co. KG, Geretsried - Germany

Introduction Magnesium is more and more used in applications, which require high strength of the material combined with low weight. Especially in car industry and electronics industry we can see a boom of using magnesium diecasted parts – mobile phone chassis, complete car doors, gearboxes, internal car parts etc. are now made of Magnesium to reduce weight with no concession regarding the solidity. The special casting properties of Magnesium allow to produce big parts with low wall thickness, so there is much more flexibility regarding the geometrics compared to aluminium diecasting. Due to these casting properties Magnesium can also be used to substitute sheet metal parts in cars (car doors for example).

1

Properties of Magnesium regarding corrosion

One of the main problems with Magnesium parts is corrosion: nearly every part needs some corrosion protection, because the surface of Magnesium is highly reactive and corrodes when getting in contact with water or high humidity. On aluminium parts there is a “natural” oxide layer, which protects the substrate against the environment and gives a good corrosion resistance without any further treatment. On Magnesium parts, most applications need some corrosion protection, because this “natural” protection does not exist. Therefore the surface has to be covered by a coating, which avoids corrosion and meets further demands like abrasive properties, cosmetics, deforming properties etc.

2

Surface treatments of Magnesium

2.1

Electroplating

Process The process of electroplating of magnesium is quite similar to electroplating of Aluminium – after cleaning and activation a thin layer of zincate is put on the part and after that a copper layer of 3 to 10 micron. Onto the copper layer there is a wide range of possibilities for other metal layers or e-coat – depending of the demands and the application you can use nickel, tin, zinc, tin/zinc alloys, copper/tin alloys etc.

525 Properties of surface Depending on the top layer the properties of the surface can be varied in a wide range. If corrosion resistance and conductivity are the major demands, we recommend tin or a tin/zinc alloy, which can be chromated or plated with e-coat to increase corrosion resistance. For decorative applications, nickel plus chrome or Cu/Sn-alloy are the first choice; the properties of this plating technology are quite similar to electroplating of aluminium. The corrosion resistance of this coating can be brought to more that 100 hours salt spray test when using optimized Ni-layers, so it is possible to meet the demands of decorative parts in the automotive industry. Applications Electroplating on Magnesium parts is now mainly used for parts in the mobile phone industry, where cosmetic issues and the conductivity of the surface are most important demands. For decorative applications in the automotive industry (especially parts inside the car) this technology is also already used in serial production for high performance sports cars. Costs The costs of electroplating of magnesium are similar to aluminium plating – if there are high quantities, the plating of magnesium becomes cheaper compared to aluminium, because the process is easier to handle and control. The costs regarding waste water and environment are lower compared to aluminium plating, because the chemicals are not so critical. 2.2

Chromating (black) plus paint/e-coat

Process The process of chromate coating is quite simple: after pre-treatment the surface is chemically converted to a chromate; it is similar to chromating of aluminium or other materials. Properties of surface The chromating layer is quite thin and gives some corrosion protection, but in most cases you need additional coating to meet higher demands. The chromating itself cannot be used for visible surfaces, because it does not look good; also the abrasive properties require additional coating on the chromating. The process and the layer use Cr(6), which can cause cancer. The chromated surface itself gives a corrosion protection of about 10 hours (depending on the casting) in salt spray test. The cromated surface can be coloured matt black, which is quite interesting for applications in the optical industry. Applications Chromating is now used on quite a lot of magnesium parts in the automotive industry as a pre-coat for e-coat or powder-coat, but there is no future for this coating because Cr(6) will be forbidden for automotive applications. In the optical industry there is a market especially for black chromating, because it is easy, cheap and effective.

526 Costs Conversion coating/chromating is cheaper than electroplating; no electricity is needed in the process and the plant is quite easy to install. 2.3

Conversion coating (plus e-coat)

Process The process of conversion coating is similar to chromating, but contains no Cr(6) so it can be used for future applications in the car industry. There are several conversion coatings on the market – there is no significant difference between these coatings but the corrosion resistance of all these conversion coatings is heavily influenced by the pre-treatment: with a perfect pre-treatment it is possible to double the corrosion resistance. Properties of surface The conversion coating layer is quite thin and gives some corrosion protection, but in most cases you need an additional coating to meet higher demands. In most cases the conversion coating is combined with e-coat; but especially for decorative surfaces it is necessary to add a additional process step between conversion coating and e-coat to get a class-A surface, otherwise small defects (i.e. spots) can occur. Applications Conversion coating is now used on quite a lot of magnesium parts in the automotive industry as a pre-coat for e-coat or powder-coat. There are also applications in optical industry or mobile phone industry – in most cases there is a kind of paint as final layer to fulfil cosmetic demands and corrosion resistance. Costs Conversion coating is cheaper than electroplating; no electricity is needed in the process and the plant is quite easy to install. 2.4

Anodising ANOMAG plus paint/e-coat

Process The ANOMAG-layer is build up on the surface (not like anodising of Aluminium) and consists of a mixture of Mg-Oxides and Mg-Phosphates. The natural colour is grey; the surface can be coloured (like anodised Aluminium) and the roughness can be adjusted to get a smoother surface. Properties of surface ANOMAG is a good undercoat for lacquer and adhesives – because of the rough structure of the natural ANOMAG layer the adhesion is very good. Tests show, that the adhesion spliced parts is twice better compared to chromating or conversion coating. The corrosion resistance of ANOMAG plus e-coat shows very good results (1000 h salt pray test) and is better than chromating or conversion coating plus e-coat.

527 Applications The ANOMAG-plating is basically suitable for applications in car industry, aviation industry and for chassis parts in the electronic industry, but until today there is no breakthrough in the market because of the quite high costs compared to conversion coating. Costs The costs for ANOMAG are higher than conversion coating and the handling of the parts on the racks is quite difficult, because the process runs with high voltage and current, so good contact between part and jig is necessary. 2.5

E-coat plus electroplating surface

Process This process combines conversion coating/e-coat and electroplating: the part is first plated with conversion coating and a special e-coat; on top of this e-coat a electroplating layer of Ni plus Cr is build up to achieve cosmetic demands. The electroplating on top of the ecoat can be compared with electroplating of plastics but requires a special combination of e-coat and electroless Ni to get good results. The quality of the e-coat is crucial for the cosmetic demands – it is necessary to have a spot-free surface otherwise the whole process does not give good cosmetic results.

Properties of surface This combination gives very good corrosion resistance because of the e-coat and very good cosmetic properties, because the e-coat also levels the surface. With this process it is possible to produce class-A parts outside the car body with chromium surface. Applications The major applications are automotive parts inside or outside the car – for high performance sports cars there are already parts in serial production. If it is necessary to get a chromium surface on a magnesium part combined with more that 500 h salt spray test there is no other choice today. Costs The costs for this process are quite high, because the parts have to be put into at least 2 different plating lines. Therefore it is only suitable for high-level products with high cosmetic demands. It is possible to reduce these costs by building up a dedicated line for this process, but until now the market is too small for such a production line.

528

3

Evironment and Magnesium Coating

Regarding the environmental issues, the electroplating of magnesium causes the same problems as electroplating of aluminium: toxic chemicals are used and – depending on the required surface – heavy metals like nickel, chromium etc. are in the process and in the surface. The problems during production can be solved by a modern waste water treatment and by training the staff, so that contamination of the environment can be avoided during the process. There are some critical issues of heavy metals in the surface: nickel i.e. can cause allergic reactions, so for future developments we try to find solutions without nickel (i.e. Cu/Sn-alloys). In the chromating process the situation is app. the same because of Cr(6) in the process and the layer. n the ANOMAG-process there are no heavy metals and no chemicals, which cause environmental problems; also the coating itself contains no heavy metals and can quite easy be stripped off by chemicals. Therefore ANOMAG is environment friendly compared to other coatings and gives few problems in the waste water treatment.

4

Summary

In the future Magnesium will be used much more in automotive industry, so the demands for a cheap and corrosion resistant coating of magnesium parts will increase rapidly during the next years. There is still a lot of development work to do to find easy and inexpensive solutions, so that magnesium parts can compete with steel or aluminium. Especially for decorative parts the whole system from casting to final chromium surface needs a lot of fine tuning and coordination to get good results and high yield.

529

Surface Treatment of Magnesium Substrates Michael Walter AHC Oberflächentechnik Holding GmbH, Kerpen

Magnesium-based materials used for structural purposes are used wherever a reduction in mass has priority. With a density of 1.74 g/cm3 they partly show specific strength properties that correspond to those of aluminium or steel, or even higher. Furthermore, magnesium substrates are completely recyclable. Despite the development of high-purity alloys most magnesium-based applications require a surface treatment in order to generally improve the corrosion resistance of this non-precious material used for structural purposes (standard electrode potential of – 2.36 V). AHC Oberflächentechnik, Kerpen, has developed various processes for the surface treatment of magnesium alloys. The two most important processes are introduced in the following. We would like to start with the introduction of a chrome-free passivation process for magnesium substrates [1]. This process produces a conversion layer consisting of oxides from both the base material itself and the oxidic reaction products which derive from components of the aqueous passivation electrolyte. The chrome-free passivation layer serves as an intermediate layer for the subsequent application of paint or adhesive. The purpose of this passivation process developed at AHC Oberflächentechnik is to provide an equivalent alternative to chromate conversion coatings and it can be applied to any magnesium-based material of common use, such as die casting, cast and wrought alloys. The process includes the following steps: degreasing, pickling, passivation, drying as well as either sealing or painting. The passivation is carried out at room temperature and can be applied by dipping, splashing, spraying, painting and other commonly used methods of application. Plant designs as used for chromate conversion coatings may also be used for the new passivation process. The aqueous passivation solution consists of potassium permanganate in combination with the anions vanadate, molybdate or tungstate. By Auger-spectroscopy, MgO, Mn2O3, MnO2 and at least one oxide from the group of vanadium, molybdene or tungsten can be determined in the conversion layer. Thus it is possible to demonstrate that the reaction of the passivation solution is like that of a redox system. Permanganate, for example, is reduced to mangane oxides of low-level oxidation. The reaction mechanism of this system is similar to that of chromate conversion coatings. Within the periodic system of the elements, the used elements of the passivation substances are very close to chromium. Picture 1 shows SEM pictures of a passivated surface on AZ91 HP, treated according to one of AHC Oberflächentechnik’s processes which is ® marketed under the trademark MAGPASS-COAT . These pictures show a finely structured conversion layer, similar to that of magnesium surfaces treated with a chromate conversion coating. The chrome-free passivation process for magnesium substrates of AHC Oberflächentechnik provides an equivalent alternative to chromate for the automo-

530 tive industry, mechanical engineering in general, the optical industry, micro electronics and other industrial sectors. Within the total coating system with subsequent painting, this process helps to provide corrosion protection.

Picture 1: SEM photographs of a AZ 91 HP passivated surface treated with the MAGPASS-COAT® process (Photos: AHC)

The space frame of the Volkswagen 1-Litre car (realized as a design study) is made from magnesium, which is even lighter than aluminium. This magnesium frame (picture 2) has been treated with the chrome-free passivation process developed by AHC Oberflächentechnik. After thorough examination of various chrome-free systems of different suppliers, Volkswagen decided to use the AHC process. The space frame is a composite welding and adhesive bonding construction of different magnesium alloys. The coated surface area amounts to a size of approx. 20 m2. Some areas of the magnesium space frame additionally received a clear paint sealing. About two years ago Audi brought a 12cylinder engine onto the market with its A8 L 6.0 quattro. As well as its driving performance, this limousine is notable for its forwardlooking lightweight construction technology. It is the only 12-cylinder model that makes use of the Audi Space Frame ASF® made from aluminium. Also made of aluminium are the wheel supports, the front and rear axle control arms, all the brake callipers, the shock absorber Picture 2: Cockpit of Volkswagen’s 1-Litre-Car: bearings and the wheels. Consequently, lightThe lower part shows the magnesium space weight metals, namely aluminium and magneframe treated with the chrome-free passivation sium substrates, were used for engine parts. A process MAGPASS-COAT . compact construction combining two 6(Photo: Volkswagen) cylinder V engines to a 12-cylinder aggregate in a W shape, together with the lightweight construction, contributes to the low weight of the engine. An essential contribution to peak automotive engineering performance, but one that is often out of sight, is always surface technology. As an example of one such detail, the inlet pipe of the W 12-cylinder engine may be mentioned here (picture 3). It is made of a high-purity sand casting magnesium alloy – a contribution to weight reduction. On the one hand, to combat the marked corrosion tendency of magnesium and its alloys and, on the ®

531 other hand, to satisfy the requirement for a chrome-free coating, the inlet pipe is treated with an aqueous chrome-free passivation solution. The MAGPASS-COAT® layer alone offers only temporary corrosion protection. In the case of the inlet pipe a 200 μm polyester powder coating in titanium silver colour is applied. This layer combination not only has an attractive appearance but stands up to all the strength and corrosion tests demanded by Audi. Another important AHC process for surface treatment of magnesium substrates, namely ® Picture 3: Inlet pipe W 12 made of AZ 91, MAGOXID-COAT , belongs to the group of treated with the the chrome-free passivation plasma chemical processes. Plasma chemical MAGPASS-COAT and a subsequent polyester coatings are carried out in electrolytes which are powder coating saline solutions. Making use of an external (Photo: AHC) power source the workpiece being processed takes on the function of the anode. Anodizing takes place as the plasma is discharged in the electrolyte on the surface of the workpiece to be treated. The substrate surface is partially melted by bursts of oxygen plasma produced in the electrolyte and an adhesive oxide ceramic metal compound forms on the workpiece. 50% of the crystalline oxide ceramic conversion layer grows into the base material whilst 50% grows outwards. Edges, cavities and relief designs are coated uni® formly. The MAGOXID-COAT layer contains a large proportion of highly resistant elements such as spinels, e.g. MgAl2O4. The produced oxide ceramic layers not only provide protection against wear and corrosion but also fulfil requirements regarding hardness, uniform layer deposition, fatigue strength under reversed bending stresses, dimensional accuracy or temperature load capacities [2 – 5]. MAGOXID-COAT® layers provide a good wear resistance similar to that of AHC’s hard anodizing layers, known on the market as HART-COAT® layers. The corrosion and wear resistance are higher than that of comparable conversion layers (DOW and HAE). Correspondingly sealed layers withstand approx. 600 hours in the neutral salt spray test according to DIN 50 021 SS. On an initial thin layer, which is in direct contact with the metal substrate, there is a slightly porous oxide ceramic layer. This layer carries a very porous oxide ceramic layer of equal thickness. The second layer may serve as an adherent base for paint or impregnations such as, for example, PTFE or other lubricants, which leads to improved corrosion and anti-friction properties. All commonly used magnesium alloys can be treated with the MAGOXID-COAT® process. Tests carried out with AZ91HP showed that the MAGOXID-COAT® treatment has only an insignificant negative effect on the fatigue strength of the base material. The MAGOXID-COAT® surface treatment is suitable for many industrial applications where magnesium substrates are used under special conditions as, for example, in mechanical engineering and the automotive sector. From the many varied applications we give here two as examples. ®

532 One possibility for the marking and signing of packaging materials is the application of hot stamping. A manufacturer of pharmaceutical products uses printing plates made of magnesium. Usually, brass or steel printing plates are used for the hot stamping of packaging materials. The printing letters are milled, engraved or ground and may be changed according to application. The manufacture of such printing clichés is comparatively expensive and is only profitable for a large quantity of, for example, one million impressions. For smaller quantities, plating plates made from a magnesium die-casting alloy have become of interest now, typically with a 100 x 50 mm size. The printing letters are manufactured by etching from defined phototechnically treated areas of the magnesium Picture 4: Magnesium printing plates, plate. This method of production is more economic than treated with 20 μm MAGOXIDthe conventional mechanical processing of brass or COAT steel. On the other hand, however, magnesium-based (Photo: AHC) clichés only have a short lifetime if not surface treated. Particularly when used for applications in the pharmaceutical sector the cleaning of the magnesium surface with disinfectants leads to a certain roughness that in turn has a negative influence on the printing image. A surface treatment ® with a 20 μm MAGOXID-COAT layer (picture 4) offers a solution to the problem. It is also possible to produce black layers that contain fade-resistant and chemically inert spinels. It is possible to treat almost any magnesium substrate suitable for industrial use. ® One intended application of MAGOXID-COAT black is, for example, the inner coating of optical components and precision threads, but is also suited for application to heat radiators, in the vacuum technology and the aviation and aerospace industry. A well-known sensor manufacturer has developed a light and ultra-compact optical sensor, destined especially for the automobile industry. This non-contact sensor measures the longitudinal and transversal vehicle dynamics relative to the road surface, thus providing data such as transverse angle, longitudinal and transverse speed as well as longitudinal and transverse acceleration. Due to its compact dimensions (approx. 164 x 52 x 61 mm) and light weight of 500 g this sensor is especially well suited for mounting on the vehicle wheel (also when being steered) for tyre-slip angle measurement. In order to be able to realise this compact and lightweight design a carbon fibre reinforced plastic has been used for the outer housing of the sensor whilst the engineers have decided to use a magnesium alloy for the housing containing the optics, as well as for further components (picture 5). These components require an adequate corrosion protection together with a black surface that reduces light reflection within the optical path. ® These requirements are met by a deep black, 25 μm MAGOXID-COAT layer. Correspondingly sealed layers withstand approx. 400 hours in the neutral salt spray test according to DIN 50 021 SS. Due to its electric insulation properties it effectively avoids ® contact corrosion. The degree of reflection of a black MAGOXID-COAT layer is less than 5%. ®

533 Today this optical sensor with the black magnesium components has proved to be excellent in practice and contributes considerably to precise, multi-dimensional vehicle measurement engineering. A surface coating well suited to treat magnesium alloys will certainly contribute to an increased application of this most interesting substrate used for structural purposes. However, even the best surface protection is of poor efficiency if the magPicture 5: Optical vehicle sensor: The CFRP outer nesium substrate does not fulfil certain housing contains magnesium components coated with basic requirements. Such as, for example, a deep black MAGOXID-COAT layer from AHC a high-purity quality of the base material, a Oberflächentechnik. magnesium cast that is almost free of (Photo: CORRSYS-DATRON GmbH) pores, or the avoidance of surface impurities with blasting material. A successful surface treatment requires an extensive dialogue between the moulder, the coating company and the user. ®

Literature [1] Kurze, P.: Chromfreie passivierte Magnesiumoberflächen mit nachfolgendem Lackauftrag. 8. Magnesium Abnehmerseminar / 8. Magnesium Automotive Seminar der Europäischen Forschungsgemeinschaft Magnesiumguss e.V. (EFM), 14./15. Juni 2000 in Aalen [2] EP 0 333 048 [3] Olbertz, B.; Haug A.T.: Oberflächenschutz für Magnesiumwerkstoffe, Metalloberfläche 43 (1989) 4, S. 174–178 [4] Kurze P.: Electrochemical Coating of Magnesium Alloys, Tagungsmaterial der DGM Magnesium Alloys and Their Applications, Garmisch-Partenkirchen, April 1992 [5] Kurze P.: Magnesiumlegierungen elektrochemisch beschichten, Metalloberfläche 48 (1994) 2, S. 104–105

534

Approach to Control the Corrosion of Magnesium by Alloying V. Kaese12, P.-T. Tai2, Fr.-W. Bach2, H. Haferkamp3, F. Witte4, H. Windhagen4 1

Volkswagen Group Research, Wolfsburg, formerly Institute for Materials Science and Biomedical Engineering Centre, University of Hanover 2 Institute for Materials Science, University of Hanover 3 Institute for Materials Science, Biomedical Engineering Centre, University of Hanover 4 Hanover Medical School, Hanover

Abstract For Magnesium alloys there are more fields as transport, communications or batteries. One is the use as bioresorbable implant in bone surgery. This saline area has high demands concerning corrosion resistance and is therefore used as main target for alloy development with the leading aspect of corrosion but regarding mechanical properties. A step-by-step building of alloy systems leads to an enhancement of the corrosion resistance by factors. These alloys are composed using common and newly defined principles of corrosion resisting alloying and processing parameters as extruding and heat treatment which aim for the enhancement of corrosion resistance. The final step is the validation of corrosive and mechanical characteristics of Fluorine-alloyed LAE-alloys in comparison to standard alloys as well in laboratory testing as in-vivo-implant in genuine pig [1].

Introduction and Motivation Table 1. Characteristics of Magnesium

Advantage availability density damping neutron holding cross section machining biocompatibility

In between corrosion casting characteristics

Disadvantage ductility creeping electrochemical position

Magnesium pros and cons are dominated by aspects as density and castibility vs. strength, creep and corrosion resistance Table 1. But the characteristics of Magnesium are more complex. The electrochemical border position can be controlled in the field of transportation if the junction is separated with coated bolts or washers [2]. Even the low corrosion resistance in saline environment is not only disadvantageous, it is wanted in the field of sacrificial anodes or batteries. Magnesium batteries are used not only to operate life vest lights or sonar buoys.

535 Experiments as implant material are dating back to 1906 [3]. According Wolf’s Law of Transformation best bone-building is achieved if the fracture is loaded. With increased bone-building the implant has to support less until relief at recovery. Best results will be gained if the implant has comparable mechanical properties as the recovering femur. The motivation to develop implants with sufficient mechanical properties is strong because there is no alloy on stock fitting with the mechanical properties of human femur Figure 1.

Figure 1. Mechanical properties of common implant materials, AZ31 vs. human bone

Besides the mechanical fitting, Magnesium is essential for some bodily functions as the nervous system. Over doses will be spilled out by the urinary tract. These boundary conditions allow further thinking about Magnesium as implant. Fractured collar bones are fixed with Magnesium splints, which lead to a sufficient recovery of the fracture and total dissolving of the implant however coming along with the formation of large gas volumes in vivo [3]. Magnesium dissolves in saline environment as salt spray in the streets or in saline solution of human body. The deductions are that future resorbable Magnesium implants, which could spare the often necessary operational revision, have to be less gas producing in the course of corroding, see [4], and more ductile and stronger. MgY1.6Cd0.25Ag0.3MnCa and MgNd2.46Cd0.12AlMnSi corrode with a hydrogen for3 2 3 2 mation of 0.8 cm H2/(cm ⋅ d) and 1.05 cm H2/(cm ⋅ d). These alloys produce far less 3 hydrogen as the theoretical maximum of 2.25 cm H2/(cm2 ⋅ d). The bone building and recovery is accelerated by factor 1.5 [5]. But the Cd-content is leading to dead-end seen from the aspects of machining and bioresorbable material. Some experiments have been performed as anode vs. a Pt-cathode in thorax of a dog. The corrosion rates of MgMn1.5, MgAl3Zn1 and MgAl6Zn3 vs. the Pt-cathode have been five times higher than necessary for a life of five years [6], see [1]. Besides orthopaedic surgery there are patented developments for heard-thorax solutions as expandable stent [7].

536

Theoretical Approach and Methods Discussed AZ31 or AZ63 corrode to fast and release to much hydrogen. The corrosion rate for in vivo corroding metal has to be controlled by alloying and metallurgical processing, not by coating, which could cause stroke. There is data to evaluate the influence of alloying elements but only partly sorted by principles. The metallurgical processing can be focused e.g. on controlling the impurities or to homogenise precipitation in the matrix. The influence of the discussed elements. Figure 2 can be summarised in principles for alloying components Table 2, see [1]. H Li

Be

B

C

Na Mg

Al

Si

Ca Sr

Cr Y

Zr

Mn

Fe

Co

Mo

Ni

Cu

Zn

Ag

Cd

Pr

Nd

Gd

Dy

O

F

S

Cl

Ga

La

Ce

N

Er

Sn

Sb

Pb

Bi

Yb

Figure 2. Periodic system of elements whose influence on Magnesium-corrosion is described

Some principles are written down now Table 3, as example #14 is chosen to be discussed here. Magnesium does corrode in NaCl-solution because the Mg(OH)2-layer is not stable to NaCl, the forming MgCl2 is soluble and not protecting. Looking for some alloying component which could form insoluble layer components, we choose Fluorine. It is commonly known as component of conversion not alloying, however Fluorine is not easy to alloy and MgF2 is not thermodynamically stable to NaF. The first obstacle can be bypassed by using a Fluorine salt with a second component known to enhance the corrosion resistance: AlF3. The alloying of AlF3 in AZ31 is leading to a reduction of the corrosion rate by factor 2 in artificial seawater. The second obstacle, looking for thermodynamically stable vs. NaF and water insoluble layers has to be gotten around by calculation. The enthalpy of reaction of NaF and Fluorine salts composed with known alloying components is compared, e.g. leading to LiF and CaF, both nearly insoluble in water solution Figure 3. Lithium is chosen, because it does fit better with the principles than Ca, which even causes hot cracking. Fluorine is successfully inserted in the hdp-LAE system by alloying [9].

537 Table 2. State of the art principles of corrosion influencing alloying of Magnesium, see [1]

1.

Avoiding elements with noble electrochemical potential or lower hydrogen overvoltage Avoiding water solvable inclusions as MgCl2, Mg3N2 Precipitating impurities as Al-Fe-compounds by Mn Grain refining of Mg-Al-alloys by adding carbon to reduce porosity Rapid solidification to enrich the matrix which corrosion resistance enhancing elements Dehydrating and compacting the natural MgO-Mg(OH)2-layer, e.g. by Al or Ce Binding the anions by opposite charge Building of amorphous, compacted and relaxed layers, e.g. by Cr, even to shift potential Alkalising the layer to shift pH-value over 9.5 by Li to stabilise the natural layer

2. 3. 4. 5. 6. 7. 8. 9.

Table 3. Now defined principles to enhance the Magnesium corrosion resistance by alloying

10. 11. 12. 13. 14. 15. 16.

Maximised solution of corrosion enhancing component in the matrix Geometrical fitting of the lattice of layer components Binding Mg-cations by double charged oxidised groups Thermodynamic calculation of multiphase micro structure with aspect of corrosion Thermodynamic calculation of layer stable vs. NaCl-containing solutions Absorption of neutrons (for using Magnesium in atomic power plants) Biocompatibility of components

Enthalpie of reaction [kJ/mol]

-300 SnF2 PbF2

-500

ZnF2

-700 2/3 AlF3

MgF2

-900

2 LiF

-1100

CaF2

-1300 0

100

200

300

400

500

Temperature [°C]

Figure 3. Enthalpy of reaction of NaF and Fluorine salts, data see [8]

600

700

800

538 Results Gastaschen hydrogen bubbles

Detail LAE442

Detail

LAE442(AlF3)2 at%, ϕ = 3,7

Figure 4. Extruded LAE442 (top) and LAE442(AlF3)2 at, X-ray after 2 weeks of genuine pig femur, width of picture: 120 mm

The effectiveness of Fluorine in LAE442 as cast is reached with 2at% added to the Magnesium-content. The corrosion rate of as cast material in potentiostatic dipping test in artificial seawater acc. to ASTM B 1141–51at room temperature is coming down from 0.2 mm/a to 0.15 mm/a to 0.1 mm/a to 0.1 mm for 0, 1, 2 and 4 at% (AlF3). LAE442 plus 2at% AlF3 is tested in extruded condition to enhance mechanical properties in comparison with LAE442 in vivo in genuine pig femur and shows for the reference LAE442 bubbles of hydrogen and signs of disintegration. LAE442 plus 2 at% (AlF3) shows original geometry and no visible hydrogen bubbles Figure 4.

Conclusions and Outlook 16 principles of alloying Magnesium to enhance corrosion resistance are proposed, one example is presented and jointed with the task to design an alloy with high corrosion resistance for Magnesium implants, one of the oldest fields of Magnesium applications. It is shown that with thermodynamic calculation the corrosion resistance by Fluorine addition can by enhanced further in synthetic seawater and in vivo in genuine pig. The reference LAE442 is to expensive for common use, but it has perspective in the field of implants but alloying with Fluorine containing salts will reach other alloy systems.

539

Acknowledgement The research has been supported by the German Ministry of Education and Research BMBF as one winning project named RemOs of the Program KOMED in 1999 and the Deutsche Forschungsgemeinschaft, Program SFB390. Special thanks to P. Lindner and P. Juchmann, Salzgitter, to A. Pisch, Grenoble, and R. Schmid-Fetzer, Clausthal, for thermodynamical advise plus calculating and the work group of RemOs.

References [1] V. Kaese, Beitrag zum korrosionsschützenden Legieren von Magnesiumwerkstoffen, VDI-Verlag, Düsseldorf, 2002. [2] H. Schreckenberger, Korrosion und Korrosionsschutz von Magnesiumwerkstoffen für den Automobilbau – Problematik der Kontaktkorrosion, VDI-Verlag, Düsseldorf, 2001. [3] A. Lambotte, Bulletins et Mémoires de la Société Nationale de Chirurgie Vol. 58 (1932)2, p. 1325-1334. [4] M.G. Seelig, Archives of Surgery Vol. 8 (1924) p. 669-680. [5] G.B. Stroganov et al., German Patent DE 1953241, 1971. [6] G. Fontenier, R. Freschard, M. Mourot, Medical and Biological Engineering Vol. 13 (1975)5, p. 683-689. [7] H. Haferkamp, V. Kaese, M. Niemeyer, M. Rodman, Tai Phan-tan, B. Heublein, R. Rohde, Magnesium 2000, Editors E. Aghion, D. Eliezer, Dead Sea, Israel, p. 159-164. th [8] O. Kubaschewski, C.B. Alcock, P.J. Spencer, Materials Thermochemistry. 6 ed. Oxford, Pergamon Press, 1993. [9] P. Juchmann, Beitrag zur technologischen Eigenschaftserweiterung von MagnesiumWerkstoffen durch Lithium, VDI-Verlag Düsseldorf, 1999.

540

Magnesium Alloys Protection by Microplasmic Processing Belozerov Valeriy, Isakov Sergey, Makhatilova Anna, Subbotina Valeriya National Technical University “Kharkov Polytechnic Institute”, Kharkov; Isakova Tatiana – Innovation Center “MAGIC Solutions” ltd., Kharkov

Technology description Magnesium alloys have high reactivity and at maintenance in atmospheric conditions can corrode with a noticeable velocity. The base modes of Magnesium alloys protection from corrosion are both oxide layers formation and protective coatings deposition. However, the application of traditional modes of coating deposition are not always reduced to required properties reaching. Last years new methods of coating deposition and surface hardening with use of high concentrated power sources intensively explicate: electric discharges, lasers etc. One of such methods is the microplasmic processing (microarc oxidation), which permit to get functional coatings on a detail surface. The both deformable AM60B (Mg - 6% Al - 0,4% Mn) and cast AZ91 (Mg - 9% Al 0,7% Zn - 0,3% Mn) Magnesium alloys were processing by MPP. The process was carried out in anodic-cathodic condition in an alkaline electrolyte with adding of inorganic compounds. The power supply of a condenser type was utilized. It is necessary to mark, that it was impossible to organize microarc discharges on Magnesium alloys surface in an alkaline electrolyte. It is bound that on a surface of an alloy the dielectric barrier layer is not formed because of lack of interaction of Magnesium with alkali. Use of water solution of a silicate (NaSiO3) has allowed to transfer process in arc discharges conditions. The power of such discharges is very high, that there are both melting of metal and formation of craters. The steady microplasmic process was implemented in composite electrolytes containing the following compounds NaOH, KOH, NaAlO2, Na5P3O10, NaSiO3 etc. As a result of such processing the surface layers of Magnesium alloys will be converted to a ceramic coating consisting of crystalline oxides and spinel. The X-ray phase analysis has shown, that the base phases of a coating are MgO, MgAl2O4, MgSiO3, the quantitative relation between which is defined by both electrolyte structure and electrolysis parameters (processing duration, current density and other). Thus, the organization of process in microarc discharges conditions requires the relevant selection of an electrolyte structure and electrolysis parameters, that allows to form coatings by thickness 200-300 microns at overgrowth velocity 1-2 microns per minute. The coating hardness makes HV 500-600 that is at 8-10 time higher then the hardness of a base (HV 60).

541 Two structures of an electrolyte distinguished by a specific resistance and acidity (pH) were utilized and also the following parameters were varied: – Current density from 5 up to 20 A/dm2; – Duration of processing from 4 about 120 minutes; – Electrolyte structure.

Test procedures and results Surface roughness There are no difference between surface roughness of AM60B and that of AZ91 alloys. Surface roughness increases with MPP coat thickening. Surface roughness varied from 9 to 20 microns (Rz) with coating thickness from 9 to 40 microns. Rz means 10-point average peak height difference of surface roughness. Surface roughness measured by roughness measuring equipment. SEM observation of the samples shows the good surface roughness. Adhesive test The microplasmic processing ensures a high adhesion of a coating with a base. At the first stage the treated surface was scratched cross-hatching 10 x 10 lines of 1mm interval on the surface by a sharp knife and then were peeled by adhesive tape and checked whether at least one cross-hatching area (1x1mm) is peeled or not. All tested coatings were not peeled from substrates. At the second stage cross-hatching samples were dipped into water for 240 hr and then were peel by adhesive tape and checked whether at least one cross-hatching area (1 x 1mm) is peeled or not. All tested coatings were not peeled from substrates. Complex corrosion test X-figurative cross-hatching samples were immersed in complex corrosion cycle bath for a certain long time (2 months). The corrosion cycle includes salt spraying (1 hour) and drying step (1 hour) and so on. After that the corrosion width were measured at the crosshatching lines. The following conditions were used: – the salt solution: 5mass% NaCl (95mass% water) – temperature of the exposed zone: 50 degrees in C – the pH range from 6.5 to 7.2 The criteria are as follows: Fair/ F: Corrosion width x ≤ 0.5 mm Marginal/ M: Corrosion width 0.5 mm < x ≤ 1 mm Poor/ P: Corrosion width 1 mm < x ≤ 5 mm Very Poor/ VP: Corrosion width x > 5 mm The results have shown Table 1, that in case of a cast alloy AZ91 (Mg - 9% Al - 0,7% Zn - 0,3% Mn) at particular electrolysis conditions the corrosion sign was not revealed (corrosion width x ≤ 0,5 mm), whereas under other requirements of an electrolysis the corrosion width reached 1 - 5 mm.

542 On a deformable alloy AM60B (Mg - 6% Al - 0,4% Mn) was impossible to ensure a high corrosion stability of a coating Table 2. Table 1. Alloy AZ91

№ of a sample

AZ2 AZ4 AZ5 AZ6 AZ7 AZ8 AZ9 AZ12 AZ13 AZ14 AZ21 AZ15 AZ18 AZ22 AZ16 AZ24

Mode of MPO Type of an electrolyte

Current density, A/decimeter2

Time, mines

1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2

10 20 20 10 10 20 20 40 5 20 20 20 20 20 20 20

30 20 20 15 15 10 10 9 120 15 15 8 8 8 4 4

Thickness, micron

Criteria

23-27 29-32 28-33 14-15 14-16 13-16 15-17 22-26 39-44 24-26 20-24 16-19 13-17 8-12 12-14 15-17

M M P M F F M M M P M P M M M M

Thickness, micron

Criteria

23-27 31-33 32-36 14-16 17-20 18-20 23-27 35-36 21-24 16-18 8-10 13-16

P P P P VP P P P P P VP P

Table 2. Alloy AM60B

№ of a sam- Mode of MPO ple Current density, Type of an A/decimeter2 electrolyte

Time, mines

AM2 AM4 AM5 AM6 AM8 AM11 AM13 AM15 AM30 AM28 AM18 AM19

30 20 20 15 10 10 9 90 15 8 4 4

1 1 1 1 1 1 1 1 2 2 2 2

10 20 20 10 20 20 40 5 20 20 20 20

543 The corrosion trials have shown, that the protective properties depend on coating phase structure, which defines its specific volume. In that case, when the ratio of coating specific volume to the base one is more then 1, the formed coating is under compression stresses operation, that reduces in improving its continuousness and increase of a corrosion stability. If the ratio is less then1, the coating is under strain stresses operation, that reduces its corrosion stability. To get high anticorrosive properties of a coating it is necessary to increase a content of spinel MgAl2O4 that is achieved by Aluminum content magnification in an alloy or in an electrolyte. Thus, MPP technology can successfully be used for a raise of a corrosion stability and hardness of magnesium alloys.

544

Class A Surface Quality for Magnesium Diecastings Using Ceramic Precursor-based Coatings R. Gadow, D. Scherer, C. C. Stahr University of Stuttgart, Institute for Manufacturing Technologies of Ceramic Components and Composites – IFKB, Stuttgart, Germany

1

Introduction

Weight reduction is today one of the priority objectives for the automotive industry. More and more safety and comfort systems are integrated to modern vehicles. This means an increase of weight and therefore an increase of fuel consumption. In order compensate this and to reduce environmental pollution automotive manufactures are forced to implement light weight design in every component of a new car. Magnesium (Mg) provides a further weight reduction compared to other light metal alloys, concerning to its low specific weight which is still 30% under Aluminum. A major part of magnesium is manufactured by high pressure diecasting process. Major drawbacks of Mg are the poor surface properties concerning corrosion and friction as well as the fire risk during machining operations. To make Mg-parts useable for the automobile body a well designed coating is needed which covers corrosion protection and optical requirements in “class A quality”.

2

Painting systems for automotive bodies

Paintings for automotive bodies need not only to protect the metallic substrate before corrosion. They have additionally to match high decorative requirements. To fulfill these requirements a multilayer system is needed. Although many studies have been made to prevent Mg alloys from corrosion, no system is well established yet especially not for the automobile body. The more promising methods are hindered by fairly high production costs. Normally the first step to improve corrosion resistance of a light metal is a conversion coating. This improves paint adhesion and prevents the spread of corrosion under the paint film [1]. The commonly used chromium containing baths are no longer of interest for new coating systems due to their critical environmental and health hazard. Two environmental friendly chromium free conversion coatings are commercially available, a fluorozirconate and a phosphate-permanganate treatment. Details of this conversation coatings are given elsewhere [2], [3], [4]. Second step is a primer which has the function to protect against spread of corrosion and mechanical damage as well as to reduce the surface roughness [5]. In the third step a filler is applied. The filler improves mechanical resistance against chipping as well as it increases the resistance against moisture. The filler may be grinded or polished to improve optical appearance. At last the top coat is applied which can be a one layer solid color coat or a metallic two layer coat.

545

3

Painting systems for magnesium parts

On the one hand there are the optical requirements like excellent gloss and smoothness for visible outer parts in automotive applications, on the other hand Mg diecastings have very poor surface properties. Mg can only be very costly grinded due to its reactivity. Therefore a new coating is needed which can smooth the surface without mechanical posttreatment. Of course the paint system must protect reliably against corrosion. To minimize investment costs the new system needs to be integrated into available standard coating lines. The first step is again a chromium-free conversion coating, which is explained in chap. 2. As second layer a silicon-modified organic resin, so called precursor, is applied. These resins excel advantageous curing conditions, good adhesion and good mechanical properties. The silicon part improves hydrophobicity and weather resistance, UV stability and temperature resistance. Also the silicon-modified resins have excellent spreading and leveling behavior [6]. Last but not least they can be applied through various standard processes like dip-coating, air-spraying, electrostatic spraying and brushing. In that they can be manufactured in available coating lines. To finish the coating system and achieve the final optical appearance a standard one or two layer topcoat is applied. Fig. 1 and Fig. 2 show schematically a standard coating on Mg substrate compared with a precursorbased coating. Clearcoat (20-30 µm)

Clearcoat (20-30 µm)

Basecoat (10 µm)

Basecoat (10 µm)

Filler (50-120 µm) Wet paint or EPS

Precursor (30-70µm)

Primer

Conversion coating

Chromate Magnesium

source: Norsk Hydro

Figure 1. typical coating used for magnesium wheels

4

AM 50 substrate

Figure 2. coating system for magnesium diecastings in Class A quality

Corrosion Test

The corrosion behavior of the precursor-based coating system was tested by salt spray according to ASTM B117 (240 h) and cyclic testing according to VDA 621-415 (10 cycles). Three different AM50A samples were pretreated with different conversion coatings: chromate, fluorzirconate and phosphate-permanganate. The following precursor layer was applied by pneumatic air spraying and cured at 160 °C for 20 min. To complete the coating system a commercially available two layer top coat was applied: a metallic basecoat and a clearcoat. The samples were scribed to the metal before the corrosion testing. The results of salt spray and cyclic testing are shown in figure 3 and 4.

546

phosphate permanganate pretreated

fluorozirconate pretreated

chromate pretreated

Figure 3. Coating systems after 240 h ASTM B117 salt spray testing with scribe

cyclic test VDA 621-415

phosphate permanganate pretreated

cyclic test VDA 621-415

fluorozirconate pretreated

cyclic test VDA 621-415

chromate pretreated

Figure 4. Coating systems after 10 cycles according to VDA 621-415

No matter which conversion coating was used, all coating systems showed excellent corrosion resistance in both tests. None or only minimal scribe creepage can be detected [7].

547

5

Conclusion

Precursor-based coating systems open the way for magnesium in the automobile body. Due to the possibility to integrate the new coating in available standard coating lines it can economically be deposited. Posttreatment of the diecastings is minimized and a class A surface finish is achieved. Sufficient corrosion resistance can be achieved even if a nonchromate pretreatment is used.

6

References

[1] K. Yasuhara, “Surface pretreatment of aluminum for automotive applications” in Automotive Paints and Coatings (Ed.: G. Fettis), VCH Verlagsgesellschaft, Weinheim, 1995, pp. 9-27 [2] J. I. Skar, M. Walter, D. Albright, “Non-chromate conversion coating for magnesium die castings” SAE technical paper series 970324, 1997 [3] H. Gehmecker, “Chrome free pre-treatment of magnesium parts” in Proceedings of IMA 51, 1994 May 17-19, Berlin, Germany, pp. 32-34 [4] D. Hawke, D.L.Albright, “A phosphate-permanganate conversion coating for magnesium” in Metal Finishing, Elsevier Science, 1995, pp. 34-38 [5] Z. Vachlas, “Primers for the automotive industry”, in Automotive paints and coatings (ED.: G. Fettis), VCH Verlagsgesellschaft, Weinheim, 1995, pp. 28-71, [6] R. Gadow, F.J. Gammel, F. Lehnert, D. Scherer, German Patent application DE 10006270.9 [7] R. Gadow, F.J. Gammel, F. Lehnert, D. Scherer, J. I. Skar, „Coating systems for magnesium diecastings in Class A surface quality“ in Magnesium Alloys and their Applications (Ed.: K. U. Kainer), WILEY-VCH Verlag GmbH, Weinheim, 2000, pp. 492-498

548

Wear and Corrosion Resistant Ceramic-based Multilayer Coatings with dry Lubrication Ability R. Gadow, D. Scherer, C. C. Stahr University of Stuttgart, Institute for Manufacturing Technologies of Ceramic Components and Composites – IFKB, Stuttgart, Germany

1

Introduction

Magnesium alloys offer new ways in light weight engineering due to their low specific weight and their high strength to weight ratio. The use of magnesium alloys in many engineering applications is mainly limited by their unsatisfying surface properties and the poor corrosion resistance. Therefore an effective and appropriate coating technology has to be adapted to enable a successful implementation of magnesium alloys under tribological load and in corrosive environment. In most cases a one layer coating is not able to meet all demands. To match the different requirements multilayer coatings have to be used. The coating is combined of at least one hard coating which protects against wear and an additional coating ensuring functionality and improved dry lubrication ability. This can either be a polymer or a thin film.

2

Coating technology and experimental setup

Several approaches to increase the wear and corrosion resistance were done before. Lacquers for example may protect against corrosive attack, but only as long as they are not hurt. They also may provide dry lubrication if filled with dry lubricants like PTFE, but abrasion resistance is normally low. Thin film technology offers coatings with e.g. dry lubrication ability, which are hard and abrasion resistant. But normally they are brittle and due to the low Young’s modulus of light metals they crack if load is increased too much. Thermally sprayed hard coatings resist reliably against wear, but due to process-related coating porosity corrosive environment may attack through the coating. Thermally sprayed coatings may provide dry lubrication ability in some cases, but normally the coefficient of friction is high. The solution must be combined multilayer coatings, consisting of a hard thermally sprayed layer and an additional lacquer or thin film layer [1]. For the hard coating ceramics, metals and cermets can be used. The decision what kind of material is to be used depends mainly on the application. To show the tribological behavior of combined coatings a TiO2 primary coating was applied to a magnesium substrate by Atmospheric Plasma Spraying (APS). For details of thermal spray technologies see [2]. It was covered on the one hand with an a-C:H thin film. On the other hand a dry lubricating varnish containing PTFE and MoS2 was applied by air spray. The lacquer was applied to the TiO2-surface once as sprayed and second on a grinded surface. If a thermally sprayed

549 coating is combined with a polymer coating the process-related roughness is useful to keep the dry lubricant filled lacquer on the surface. Even if the peaks are worn dry lubricants remain on the ground. Thin films are only applied to a grinded surface. The hard coating supports against cracking under high load, peaks would be a potential source of failure. The tribological results were made by a pin-on-disc test with a normal load of 10 N, constant humidity of r.h. = 40% and temperature of 25 °C. The counterpart was a hardened steel 100Cr6 ball with a diameter of 5 mm. In a second investigation basic corrosion tests were done with thermally coated Mgsubstrates. In order to evaluate the behavior of different materials a metal (Al/Si), a cermet (WC-CoCr) and two oxide-ceramics were applied to AZ91 Mg-plates. The coatings were done by APS-spraying except for the WC-CoCr which was done by High Velocity Oxygen Fuel (HVOF) Spraying. To expose only the coated side of the Mg-plate a tube was glued to the surface. The tube was filled with solution of NaCl (3%). After 24 h the tube was removed and the surfaces was optically examined.

3

Results and discussion

3.1

Tribological results

The hardness of the TiO2 coating was measured in the cross section and was HV 0,5 = 962. Roughness as sprayed was Ra = 3.7 μm. The grinded surface roughness was Ra = 0.05 μm. Figure 1 shows the coefficient of friction versus the number of oscillations for the different coating systems. As a reference the data from a TiO2 coating without functional second coating is added. 1

TiO2 not coated

Coefficient of friction µ

0.8 lubricant lacquer grinded

0.6

0.4

0.2

DLC

lubricant lacquer as sprayed

0 0

20000

40000

60000

Number of oscillations Figure 1. Tribological behavior of coated Mg substrates

80000

1000

550 Figure 2 shows the wear rate for the different friction pairings. The wear of the coating was measured through tactile surface scanning and a software-based loss volume calculation. The wear on the ball was calculated after measuring the spherical loss. 0.3 surface wear volumetric wear [mm³]

0.25

wear on 100Cr6 ball

0.2 0.15 0.1 0.05 0 DLC

lubricant lacquer; lubricant lacquer; TiO2 not coated TiO2 polished TiO2 as sprayed

dry lubricant coating Figure 2. Volumetric wear of substrate and counterpart in pin-on-disc test

The coefficient of friction for the uncoated TiO2 is not acceptable high, but anyway it is tolerably constant. Therefore no galling effects are expected. Wear on the coating surface is high and on the ball it is even the highest. For the combination grinded TiO2/lubricant lacquer a significant improvement of the friction coefficient is visible, but only in the beginning. After a short period the lacquer is worn and again TiO2 slides against 100Cr6. This results in high wear on the coating surface, but wear on the ball is reduced. The two systems grinded TiO2/thin film and TiO2 as sprayed/lubricant lacquer show excellent behavior. The coefficient of friction is constantly low. Wear is reduced to a minimum [3]. 3.2

Corrosion test results

Figure 3 to 6 show the results of the corrosion test after the removal of the tube.

551

Figure 3. Al2O3/TiO2-coating after corrosion test

Figure 4. TiO2-coating after corrosion test

Figure 5. Al/Si-coating after corrosion test

Figure 6. WC-CoCr-coating after corrosion test

The most severe corrosive attack is seen for the Al2O3/TiO2 and the TiO2 coatings. Both coatings show severe cracks and are delaminated from the Mg substrate. This is due to the high porosity of these APS sprayed coatings, especially in form of open porosity. Hydrogen is formed and pressure is build up to a certain degree [4]. As a result delamination of the ceramic coatings occurs. Large areas of the Al2O3/TiO2 and the TiO2 coatings delaminate simultaneously. For the AlSi coating instead only minor corrosive attack is seen at the end of the test changing the optical appearance of the area of the Al/Si coating which was exposed to the NaCl solution. This is due to the lower porosity of this coating and the quasi ductility of the Al/Si coating preventing a brittle failure, as in the case of the ceramic coatings. For the HVOF sprayed WC/Co/Cr coating severe local corrosive attack can be observed at areas of open porosity due to the high differences in the electrochemical potential. But due to the higher adhesive strength and the higher ductility of the HVOF sprayed cermet coating as well as due to the lower porosity a delamination of larger areas of the coating is prevented. Combined coatings were not tested at this time, but it is expected that corrosive attack will not occur as long as the polymer film is intact and a total electrochemical isolation is achieved.

552

Conclusion Combined coatings based on ceramic hard coatings are a powerful instrument to protect light metal and in this study particularly magnesium against wear. Functional polymer or thin film coatings enable dry lubrication ability. Ceramic hard coatings do not always protect against corrosion itself. Additional treatments and coatings are needed. The right combination of hard and functional coatings enables the application of light metal components in many fields. Automotive, aerospace or consumer products are only three of them. The costs of a combined coating are tolerably low particularly for the combination of a hard ceramic coating with a lubricant lacquer. The combination with thin films may be economical for small high end components.

References [1] R. Gadow, A. Killinger, D. Scherer, Ceramic Polymer Composite Coatings, Mat.wiss. u. Werkstofftechnik, 29, 1998, pp. 292-299 [2] L. Pawlowski, The science and engineering of thermal spray coatings, John Wiley & Sons Ltd. Chichester, 1995 [3] R. Gadow, D. Scherer, Novel processing for combined coatings with dry lubrication ability, Ceramic Transactions, Vol. 129, Innovative Processing and Synthesis of Ceramics, Glasses and Composites V, eds. N.P. Bansal et. al., The American Ceramic Society, Westerville, Ohio, 2002, pp. 125-136 [4] C. Friedrich, R. Gadow, D. Scherer, Functional Ceramic and Metallurgical Coatings on Magnesium Components, Proceedings of the International Thermal Spray Conference & Exhibition, 28-30 May 2001, Singapore

553

New Generation of Keronite Process for Corrosion Protection of Magnesium S. Hutchins, P. Shashkov, V. Samsonov and A. Shatrov Keronite Ltd, Granta Park, Great Abington, Cambridge CB1 6GP, UK

1

Introduction

Magnesium’s favourable strength-to-weight ratio makes it attractive to automotive and other industries. The cost of both raw magnesium and finished-goods is reducing as production increases. However, the lack of an acceptable-cost, chrome-free corrosion pretreatment has hampered its widespread adoption. This paper introduces a new generation [1] of plasma electrolytic oxidation (PEO) coating process that addresses the combined requirements of corrosion protection, scratch resistance, paintability (including A-class surfaces), acceptable cost, series production and large parts (up to 2 m2) capability. A further benefit of smoother coatings is to enable wear applications without the need to polish. This new PEO application category is mentioned here for completeness only and is not the main subject of this paper.

2

Process Description

The Keronite process is an environmentally-friendly, bath-based treatment for light alloys where a pulsed voltage is applied to the substrate component and the resulting sparks create a fused ceramic layer on the surface of the metal. The latest advancements in the Keronite process focus on optimisation of waveform and frequency of electrical pulses together with the introduction of acoustic vibration of a certain frequency range and air micro-bubbles into the electrolyte. Amplitude, duration and shape of both positive and negative pulses were optimised to achieve controllable high deposition rates of 2-10 μm/min. Acoustic vibration was introduced into the bath by pumping electrolyte through a specially designed hydrodynamic device working in the audible frequency range. Overlapping of electrical and acoustic frequency ranges had a synergistic effect on the process. The hydrodynamic generator is also responsible for oxygenating the electrolyte. The resulting Keronite process produces hard, smooth and compact layers on Magnesium.

3

Coating Characterisation

3.1

Morphology

Traditional PEO coatings are porous and have roughness increasing with thickness. This gives a good key for topcoats but limits hardness, corrosion performance of unsealed

554 specimens, and achievement of an A-class paint finish. Samples of AZ91D alloy were coated with 10 μm of Keronite and viewed by Electron Microscope. The surface has a lenticular structure with fine pores (Figure 1), smaller than with anodised Mg, also appears smoother and denser [2]. These characteristics will help roughness and hardness issues whilst allowing enough porosity for paint adhesion. X-ray analysis shows that the main constituent is spinel (MgAl2O3).

Figure 1. SEM of surface of 10 μm Keronite layer on AZ91D

3.2

Hardness

The in-plane hardness of a 60 μm Keronite layer was tested at Cambridge University [3] as a function of distance through the coating using a Berkovitch nanoindenter with 10 mN load (Figure 2). The average hardness of 7.1 GPa (approx 700 HV) compares favourably with that of bare magnesium (approx 100 HV) and is similar to that of hardened tool steel. 16

10 mN load, Berkovitch indenter

14

Hardness: 7.1±2.9 GPa

Hardness /GPa

12

10

8

6

4

2

Top of Coating

Substrate Interface 0

0

10

20

30

40

50

Approximate through thickness scale/μm

Figure 2. Hardness profile of 60 μm Keronite coating on AZ91D

60

70

555

4

Corrosion Properties

4.1

Description of Samples

Plates of cast AZ91D were coated with nominal 10 μm of Keronite. The authors found that 10 μm provides a balance of corrosion protection and scratch resistance with acceptable cost and coating time (2 minutes). The coating is designed as a pre-treatment for other sealers or top-coats. Therefore, the samples were post-treated with different coatings, including automotive-standard e-coat and powdercoat [4], as shown in Table 1. Automakers seek protection against galvanic corrosion and corrosion after the coating has been scratched. Sample plates were prepared accordingly. Table 1. Description of AZ91D samples prepared for testing.

Sample Label

Description

A.1 – A.4 B.1 – B.4 C.1 – C.4 D.1 – D.4 E.1 – E.4 F.1 – F.4 G.1 – G.4

10 μm Keronite as-coated 10 μm Keronite plus sodium silicate sealant (Na2SiO3) 10 μm Keronite plus sol-gel sealant (NTC GmbH, Basecoat U-Sil100) 10 μm Keronite plus e-coat1 (McDermid Electrolac High Build XD4434) 10 μm Keronite plus e-coat2 (BASF GV82/9438) 10 μm Keronite plus powder-coat (HB Fuller P4M5229 polyester) 10 μm Keronite plus e-coat2 plus powder coat

H.1 – H.2

Bare Magnesium

4.2

Corrosion and corrosion after scratch

One sample of each type (A.1, B.1, etc) was scribed through to the substrate and subjected to 750 h of salt spray according to ASTM B117. The corrosion was rated according to ASTM D1654 (Figure 3, 4 Table 2). All coatings show ratings of 9 or 10 after 750 h, in both scribed and unscribed regions. The presence of e-coat or powder coat (D.1-G.1) ensures a 10 rating in both categories. These results exceed those after 240 h of Beals et al. [5]. 10

9

8

Rating (ASTM D1654)

7

6

Keronite K + silicate sealant K+ sol-gel

5

K + e-coat1 K + e-coat2

4

K+ powder K + e-coat2 + powder

3

Mg

2

1

0 0h

24h

96h

192h

336h

500h

750h

Time in salt spray (h)

Figure 3. Corrosion resistance of Keronite plus additional coatings on AZ91D according to ASTM B117

556 10

9

8

7 ASTM 1654 Rating

Keronite K + silicate sealant

6

K+ sol-gel K + e-coat1

5

K + e-coat2 K+ powder 4

K + e-coat2 + powder Mg

3

2

1

0 0h

24h

96h

192h

336h

500h

750h

Time in Salt Spray

Figure 4. Corrosion creep from scratch using ASTM B117 test and ASTM D1654 corrosion creep rating.

4.3

Galvanic corrosion

One sample of each coating (A.2, B.2, etc) was drilled before coating and fitted with a zincplated M6 steel bolt after coating. The bolt was fixed using a nylon nut and the samples subjected to 750 h salt spray. The intention with these samples is to continue the test until 2000 h and rate using ASTM D1654 with the bolt removed. Nonetheless, inspection of the samples at 750 h was very encouraging on samples with e-coat or powder coat (D.2–G.2). 4.4

Phospahte Bath Compatibility

A wish of some automotive Mg users is to paint components in a conventional steel paint line. The early stages of such a process usually include zinc phosphate then e-coat. Magnesium is corroded by phosphating and cannot be painted on such a line. Samples A.3, B.3 etc were coated in full, dipped in zinc phosphate (ABPHOS 700Z) at Plastic Coatings Ltd, then tested by ASTM B117-97 with only the as-coated Keronite showing any sign of corrosion after 750 h (Figure 5, Table 2). Other workers [6] claim success passing Keronite-coated AZ91D through a steel paint process, including zinc phosphate, with no subsequent corrosion problems.

5

Mechanical Performance

5.1

Scratch Resistance

Samples A.4–G.4 were tested at the University of Hull [7] for scratch resistance using a Rockwell-type conical diamond indenter (tip Ø = 0.4 mm) in an IonCoat Multipath Scratch Tester to produce 4 parallel scratches with linearly increasing normal load Ln, up to 100 N. Table 2 shows the load (Lc2) that detached the composite coating from the substrate. As-coated Keronite detached at Lc2 = 14 ± 2N (approximately 3 times higher than anodised Mg [6]). The two sealants increased the resistance approximately 30% and e-coat1 by nearly 3 times. For coatings (F.4, G.4) including powder coat, the indenter did not detach the Keronite, even with maximum loading.

557 10

9

8

ASTM 1654 Rating

7

6

5 Keronite

4

K + silicate sealant K+ sol-gel

3

K + e-coat1 K + e-coat2

2

K+ powder

1

0 0h

24h

96h

192h

336h

500h

750h

Time in Salt Spray

Figure 5. ASTMB117 corrosion performance after immersion in zinc phosphate bath

5.2

Paint Adhesion

Samples A.4-G.4 were sprayed on one half with Newlon N17 water-based paint, and on the other with Trimite B401AE 262/3 acrylic solvent-based paint. These were then subjected to and rated by paint adhesion test ASTM D3359-97. The results (Table 2) show that the as-coated Keronite provides an excellent key for the paint. On samples B.4-G.4, any delamination occurred between the top two layers of coating, leaving the Keronite/polymer interface intact. Table 2. Summary of corrosion and mechanical performance of Keronite on AZ91D

750 hour Salt Spray Corrosion Resultsb

A B C D E F G H

Scratch Resistance

Paint Adhesion Ratingf

Coatinga

Subjective Creep After Breakthrough Waterc d c Rating from Scratch Phosphate force LC2 (N) based

Solventbased

K K + silicate seal K + sol-gel seal K + e-coat1 K + e-coat2 K + powder coat K + e-coat2 + powder coat Bare magnesium

10 9 10 10 10 10 10

9 9 9 10 10 9 10

0

0

6 10 10 10 10 10 10

14 ± 2 18 ± 1 18 ± 1 45 ± 2 N/T >100e >100e

4 4 4 4 4 4 4

5 5 5 5 5 5 5

N/T

N/T

N/T

N/T

Notes (N/T = not tested) a Keronite coating used was 10 μm for all samples. K = Keronite. b Tested according to ASTM B117-97 c Rating to ASTM D1654-92. 10 = no visible corrosion, 9 =