Coated Magnesium - Designed for Sustainability?

Coated Magnesium Designed for Sustainability?

Proefschrift ter verkrijging van de graad van doctor aan de Technische Universiteit Delft, op gezag van de Rector Magnificus prof.dr.ir. J.T. Fokkema, voorzitter van het College voor Promoties, in het openbaar te verdedigen op dinsdag 25 november 2008 om 12.30 uur door Christina Elisabeth Maria MESKERS

ingenieur in de Technische Aardwetenschappen geboren te Haarlemmermeer

Dit proefschrift is goedgekeurd door de promotoren: Prof. D.Eng M.A. Reuter (PhD) Prof. Dipl.Ing. U.M.J. Boin Prof. Dr. R. Boom

Samenstelling promotiecommissie: Rector Magnificus Prof. D.Eng M.A. Reuter (PhD) Prof. Dipl.Ing. U.M.J. Boin Prof. Dr. R. Boom Prof. D.Sc. (Tech.) K. Heiskanen Prof. B. Bj¨ orkman Dr. Ir. A. van Schaik Prof. Ir. L. Katgerman Prof. Dr. G.J. Witkamp

voorzitter University of Melbourne/ Ausmelt Ltd., Australia, promotor Technische Universiteit Delft, promotor Technische Universiteit Delft, promotor Helsinki University of Technology, Finland Lule˚ a University of Technology, Sweden MARAS Technische Universiteit Delft Technische Universiteit Delft

Dr. Y. Xiao heeft als projectleider aan de totstandkoming van het proefschrift bijgedragen.

ISBN: 978-9-064643-05-7

©

Copyright 2008, by C.E.M. Meskers. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior written permission of the author. SUPPORT The research reported in this thesis has been financially supported by SenterNovem IOP Surface Technology. Additional support has been provided by a Marie Curie Fellowship of the European Community programme Early Stage Training. Cover design by Ester Abrahams. Printed by Ponsen en Looijen B.V.

Contents Nomenclature

vii

1 Introduction 1.1 The product life cycle . . . . . . . . . . . . . . . . 1.1.1 Product manufacture . . . . . . . . . . . . 1.1.2 Use . . . . . . . . . . . . . . . . . . . . . . 1.1.3 End-of-Life . . . . . . . . . . . . . . . . . 1.1.4 Raw materials production . . . . . . . . . . 1.2 Closing the cycles . . . . . . . . . . . . . . . . . . 1.2.1 Sustainability . . . . . . . . . . . . . . . . 1.2.2 Strategies to achieve sustainability . . . . . 1.2.3 Methods to assess sustainability . . . . . . 1.2.4 Evaluation of coated semi-product recycling 1.3 Outline of the thesis . . . . . . . . . . . . . . . . . 1.4 Publications from this thesis . . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . . .

1 2 2 4 5 7 12 12 13 15 17 19 22

2 Theoretical background 2.1 Kinetic analysis . . . . . . . . . . . 2.2 Exergy analysis . . . . . . . . . . . 2.2.1 Concept . . . . . . . . . . . 2.2.2 Calculation . . . . . . . . . . 2.2.3 Indicator of sustainability . . 2.2.4 Application of exergy analysis

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . recycling .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

25 26 29 29 30 33 34

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

37 38 39 40 42 43 45 45 46 46 48 50

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . to magnesium

3 Characteristics of product 3.1 Introduction . . . . . . . . . . . . . . . . 3.2 Metallic coatings . . . . . . . . . . . . . . 3.3 Conversion coating . . . . . . . . . . . . 3.4 High voltage anodizing coatings . . . . . 3.5 Organic coatings . . . . . . . . . . . . . . 3.6 Examples of coated semi-products . . . . 3.7 Characterisation of coated semi-products . 3.7.1 Methods . . . . . . . . . . . . . . 3.7.2 Metallic coating . . . . . . . . . . 3.7.3 Conversion and anodizing coatings 3.7.4 Organic coatings . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

ii

CONTENTS 3.8

4 EoL 4.1 4.2 4.3 4.4 4.5

Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

treatment of product Introduction . . . . . . . . . Coating removal methods . . Cryogenic de-coating . . . . Chemical de-coating . . . . . Thermal de-coating . . . . . 4.5.1 Experimental setup . 4.5.2 Results . . . . . . . . 4.6 Evaluation of EoL treatments

52

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

55 56 56 58 59 60 61 64 71

5 Kinetic modelling of thermal de-coating 5.1 Introduction . . . . . . . . . . . . . . 5.2 Scission phase . . . . . . . . . . . . . 5.2.1 Activation energy . . . . . . . 5.2.2 Reaction model . . . . . . . . 5.3 Combustion phase . . . . . . . . . . . 5.4 Recyclability parameters . . . . . . . . 5.4.1 Definition . . . . . . . . . . . 5.4.2 Quantification . . . . . . . . . 5.4.3 Contribution to recyclability . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

75 76 76 76 79 83 83 83 85 86

. . . . . . . . . . . . . . interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

89 90 90 92 94 96 96 100 103 103 106 114 115 116 117 119

7 Remelting experiments 7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2 Flux - oxide interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2.1 Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

121 122 122 122

. . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

6 Mg from industrial materials - exergy analysis 6.1 Introduction . . . . . . . . . . . . . . . . . . . . . 6.1.1 Recycling system . . . . . . . . . . . . . . 6.1.2 Preliminary assessment of metal-salt-coating 6.2 Simulation Approach . . . . . . . . . . . . . . . . 6.3 Uncoated semi-product . . . . . . . . . . . . . . . 6.3.1 Exergy analysis . . . . . . . . . . . . . . . 6.3.2 Impact of parameter changes . . . . . . . . 6.4 Semi-product with single component coating . . . . 6.4.1 Metallic . . . . . . . . . . . . . . . . . . . 6.4.2 Non-metallic (oxide) . . . . . . . . . . . . . 6.5 Coated semi-products . . . . . . . . . . . . . . . . 6.5.1 Object 1 . . . . . . . . . . . . . . . . . . . 6.5.2 Object 2 . . . . . . . . . . . . . . . . . . . 6.5.3 Object A - D . . . . . . . . . . . . . . . . 6.6 Discussion and conclusion . . . . . . . . . . . . . .

CONTENTS

iii . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . semi-products . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

123 124 124 125 127 127 127 130 130 130 132

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

135 136 136 137 140 140 142 144

9 Metric for recycling 9.1 Introduction . . . . . . . . . . . . 9.2 Metric for recycling . . . . . . . . 9.3 Coated semi-products in the metric 9.4 A tool for the design process . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

147 148 148 150 152

7.3

7.4

7.5

7.6

7.2.2 Results . . . . . . . . . . . Metal - oxide interaction . . . . . 7.3.1 Approach . . . . . . . . . 7.3.2 Results . . . . . . . . . . . Metal - oxide - salt flux interaction 7.4.1 Approach . . . . . . . . . 7.4.2 Results . . . . . . . . . . . Remelting of coated and de-coated 7.5.1 Approach . . . . . . . . . 7.5.2 Results . . . . . . . . . . . Discussion and conclusion . . . . .

8 Evaluation of remelting process 8.1 Introduction . . . . . . . . . . . 8.2 Remelting losses R . . . . . . . 8.2.1 Single component coating 8.2.2 Coated semi-products . . 8.2.3 Salt slag . . . . . . . . . 8.3 Remelting kinetics . . . . . . . . 8.4 Conclusion . . . . . . . . . . . .

10 Conclusions and Recommendations 155 10.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156 10.2 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157 Bibliography A Theoretical background A.1 De-coating kinetics . . . . . . A.1.1 Derivation of g(α) . . . A.1.2 Model-free methods . . A.2 Exergy . . . . . . . . . . . . . A.2.1 Derivation . . . . . . . A.2.2 Reference environment A.2.3 Exergy types . . . . . . A.3 Thermodynamics of remelting .

159

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

179 179 179 181 183 183 185 185 187

iv

CONTENTS

B Characteristics of coated semi-products 191 B.1 Determination of coating mass and density . . . . . . . . . . . . . . . . . 191 B.2 Composition inorganic components . . . . . . . . . . . . . . . . . . . . . 193 B.3 EPMA images and X-ray mappings . . . . . . . . . . . . . . . . . . . . . 194 C Thermal de-coating experiments C.1 Calibration of TG/DTA equipment C.2 TG/DTA Experiment . . . . . . . C.3 Data processing . . . . . . . . . . C.4 Results . . . . . . . . . . . . . . . C.5 Results iso-thermal experiments . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

205 205 206 210 213 244

D Kinetic Modelling of de-coating D.1 Modelling of scission phase . . D.1.1 Activation Energy . . . D.1.2 Reaction model . . . . D.2 Modelling of combustion phase

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

249 249 249 252 269

. . . .

. . . .

E Remelting simulation

275

F EMF experiments F.1 Equipment . . . . . . . . . . . . . F.1.1 Theoretical background . . F.2 Construction and first experiments F.2.1 Experimental strategy . . . F.2.2 Results . . . . . . . . . . . F.2.3 Discussion . . . . . . . . . F.2.4 Recommendations . . . . . F.3 Continued experiments . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

. . . . . . . .

305 305 306 308 308 310 311 313 314

G Remelting of magnesium ingots G.1 Installation and start-up of melter G.2 Influences on experimental results . G.2.1 Melting . . . . . . . . . . G.2.2 Residue analysis . . . . . . G.3 Experimental results . . . . . . . . G.4 Discussion . . . . . . . . . . . . . G.5 Conclusions . . . . . . . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

315 315 316 316 317 318 321 321

H Remelting experiments

325

I

329

Evaluation of remelting process

J Metric for recycling

331

CONTENTS

v

Summary

333

Samenvatting

335

Curriculum Vitae

337

vi

CONTENTS

Nomenclature a a A e e’ E Ea f(α) g(α) k m m/Z n Q R R

1/min J/mol J/kg J kJ/mol 1/min kg emu mol J 8.314 J/(mol K) MJ

S t T x

activity kinetic model parameter frequency factor specific exergy specific exergy exergy activation energy reaction model integral function temperature dependent factor mass mass to charge ratio mass enthalpy of heating from Tref to T gas constant Remelting losses, impact on remelting = ∆EeNo + ∆Equality + Eresource entropy time temperature mole fraction

Greek letters α β ψ ξ

fraction decomposed heating rate exergy efficiency primary resource intensity

/min -

J/K min K -

viii

NOMENCLATURE

Subscripts c ch destr. e entr. eNo f F i (inter)metall. p sol 0

at extrapolated completion point chemical destructed at extrapolated onset point entrapped entrapment and oxidation at end of mass loss peak at end of de-coating process initial (inter)metallics peak solution at reference conditions

Superscript 0

at reference conditions

Abbreviations DfR DTA DTG EGA ELV EoL LCA MFA WEEE

Design for Recycling Differential Thermal Analysis Differential Thermal Gravimetry Evolved Gas Analysis End-of-Life Vehicle End-of-Life Life Cycle Analysis Mass Flow Analysis Waste Electrical and Electronic Equipment

[Blawert, Morales, Dietzel, Hort, Kainer, Scharf, Ditze and Endres 2005] [Meskers, Xiao, Boom, Boin and Reuter 2007a] [Meskers, Xiao, Boom, Boin and Reuter 2007b] [Meskers, Xiao, Boin, Boom and Reuter 2007] [Meskers, Reuter, Boin and Kvithyld 2008] [Meskers, Reuter and Boin 2008]

Chapter 1

Introduction

2

1.1

CHAPTER 1. INTRODUCTION

The product life cycle

Every product progresses through the four phases of the product life cycle, depicted in figure 1.1. In each phase residues are created. Their amount and composition is largely determined by the complex material combinations created/invented during product design in the manufacture phase. In the End-of-Life (EoL) phase the product is separated in different metal streams, which have to be suitable for raw materials production. Raw materials production requires highquality metal streams to be able to produce metal that can be re-used for manufacturing of the same product again. For this it fully depends on the EoL phase. The efficiency of the EoL treatment and the quality of the obtained metal streams are affected by the material combinations in the (consumer) product. Unfavourable combinations lower the quantity as well as the quality of the metal streams and increase the residues leaving both the EoL and raw materials production phase. At the same time natural resources remain necessary to fulfill the demand for high quality metals and other materials. Choices made during product design thus have a lasting effect on product and material life cycles. Sustainable closed life cycles, in which the quality of materials is preserved and residue creation and resource depletion are minimized, can be attained by selecting the appropriate alternative during product design. To realize this the effect of design choices on the life cycle and more specific the recycling process has to be known. In the remainder of this section the product life cycle will be discussed in more detail based on the life cycle of magnesium semi-products used in consumer products (figure 1.2). Compared to the ideal cycle in figure 1.1 the magnesium cycle is far from closed. Most noticeable is absence of the link between the EoL phase and raw materials production. A closed cycle only exists for class 1 scrap, which goes from the product manufacture phase directly to production from industrial material and returns to the manufacture phase as magnesium alloys. The four problematic streams, including all other types of scrap, are marked by ??? in figure 1.2. Currently these are not or only partially recycled due to economical and/or technical reasons. A closer look at the magnesium cycle, its environmental impact and ’sustainability’ with a focus on recycling is therefore justified.

1.1.1

Product manufacture

Product manufacture starts with designing the product, a process in which creativity and technology meet to create a product that unites all requirements. Sustainability issues have to be an integral part of each stage in the design process as it relates to all elements in the design [Pahl et al. 2007, p.394]. For this a range of tools is available varying from general guidelines to detailed information [Luttropp and Lagerstedt 2006]. Knowledge about ’sustainability’ has to become part of the tacit knowledge the designer has about materials, their production and manufacturing processes [Walker 2006]. Magnesium is one of the construction materials to be considered for consumer products. It is a relatively young metal, discovered in 1755 and isolated in metallic form more than fifty years later by Humphrey Davy. Michael Faraday reduced magnesium chloride by electrolysis in 1808 and obtained magnesium; a method that is still in use for magnesium

1.1. THE PRODUCT LIFE CYCLE

Figure 1.1: Product Life Cycle consisting of Product manufacture, Use, End-of-Life and Raw materials production.

Figure 1.2: Life cycle of magnesium semi-products. Open ends are marked by ???. The dashed arrow indicates the cycle is not closed.

3

4


production today. The light weight and high strength make magnesium very suitable for automotive applications as a Mg engine block weighs 30 kg, while 39 kg Al or 84.6 kg cast iron are needed [Tharumarajah and Koltun 2007]. These masses include the design considerations for the engine block. This is a theoretical calculation because blocks are nowadays watercooled, which is not possible with a magnesium engine block.Pistons were the first car part made of magnesium in 1924, but most famous is its use in the Volkswagen Beetle (25 kg/Beetle) [Harbodt and Brown 2006, Ch. 1]. Since the 1990s the use and production has been steadily increasing from 248000 t in 1998 to 726000 t in 2006 [International Magnesium Association 1995 - 2006]. About half of this is used for diecastings applied in a wide range of products: powertrain and interior car parts [Hydro 2007], cellular phones [Motorola 2007b], laptop housings [HP 2007; Sony 2007; Toshiba 2006], and photo and video cameras [Canon 2007b; Nikon 2007a]. Some examples are shown in figure 1.3. Compared to the 10 - 15 kg magnesium present in a luxury passenger car [Willekens 2004], the mass of magnesium in consumer electronics is much less: about 700 grams for a camera body [Canon 2007a]. The second part of the product manufacture phase includes manufacturing of magnesium semi-products and assembly of the semi-products together with other parts according to the design of the final product. Casting of magnesium starts with remelting of the magnesium alloy ingots in large crucibles. Oxidation and evaporation are prevented by using cover gasses, salt flux or components that form a stable oxide layer on top of the liquid magnesium. Before casting dross (figure 1.4f) is removed, containing up to 90 % metal, the rest is oxide [Westengen 2006]. The magnesium is transferred to the die and a casting is produced. During casting only 50 - 80 % of the metal is used for the part [Hydro 2007], the rest are runners and biscuit attached to the part (figure 1.4c). These are removed directly after casting. The runners and castings with defects (rejects), so-called new class 1 scrap, are returned to the raw materials production phase. Subsequently the casting is machined to obtain close size tolerances, improve surface quality and to add holes and threads. Swarf and fine magnesium material are created (figure 1.4b). Magnesium semi-products can be coated to enhance corrosion resistance, their appearance, or electrical properties. During coating about one percent of the semiproducts is rejected [Brungs and Mertz 2006]. After quality control the semi-products are assembled into final products such as the ones shown in figure 1.3 on page 5 and ready for use.

1.1.2

Use

The use phase of a product covers the actual use of the product in society, its ’life’. How long this life is depends on four factors [Walker 2006; van Nes and Cramer 2006]: Durability Functional obsolescence Technical obsolescence Aesthetical obsolescence


5

Figure 1.3: The magnesium cycle: MgCl2 (upper left) is converted into ingots (lower left). The ingots are used to cast parts (middle), followed by further treatment and assembly to final products (right). The dashed arrow indicates that the cycle is not closed yet: magnesium from EoL products is not converted back to ingots.

Each product is designed to last a certain amount of time. Packaging materials, such as a paper coffee cup doesn’t have to last much longer than the time it takes to finish the coffee. The life of consumer durables is often only a few years [Walker 2006] while cars have a life time between 7 and 20 years [Bertram et al. 2004]. A building on the other hand is made to withstand decades. Product durability can be extended when repair and upgrade are possible. Nevertheless, at some point the product is worn out or unrepairable, and has to be discarded. At the moment it is discarded the product enters the EoL phase. Consumer products entering the EoL phase form a mixture of designs and models, ranging from the latest, most recent design to an old one of many years ago, and everything in between. In the course of time the proportion between older and newer designs changes. Therefore the input to the EoL phase is distributed in nature, time dependent and determined by the life time and design of the products [Reuter et al. 2005, p. 210].

1.1.3

End-of-Life

The amount of products entering the EoL phase is considerable: in Europe 6.5 million tonnes of Electrical and Electronics Equipment waste (WEEE) such as computers, DVD players, fridges, dryers, vacuum cleaners, hand held equipment etc. and 8 - 9 million tonnes of End-of-Life Vehicles (ELV) are created each year [RSA 2006; EC 2007]. Without treatment this would result in a gigantic amount of landfilled materials, which is a misuse of space and material resources.

6


(a) Diecasting rejects (class 1A)

(b) Swarf (class 5)

(c) Runners (class 1B)

(d) Swarf briquette (class 5)

(e) Coated rejects (class 2)

(f) Dross (class 6A)

Figure 1.4: Magnesium residues and scrap created during product manufacture. In brackets the scrap class is indicated.


7

Most End-of-Life products are collected via central drop off points or via drop off at store where the product was bought. It is followed by physical treatment: a series of processes which will convert the EoL product back into as many useful materials as possible. Scarcity of resources, a high monetary value of metals in the EoL product, or the potential hazard of a material are the main drivers for recycling [Reuter et al. 2005, p. 16]. The magnesium semi-products in the EoL products are liberated from the other materials by breaking the EoL product in smaller pieces during shredding. This is a crucial step as liberation of materials only takes place during shredding or manual dismantling. Magnesium parts that are physically joined by bolts or screws to other material are liberated. Also when there is a difference in plastic behaviour between magnesium and other materials liberation during shredding can take place. For example a ceramic part can be separated from a metal part because when shredded the brittle ceramics breaks into pieces while the more ductile metal is merely deformed. Coatings cannot be removed, because of the small joint and chemical bond between the coating and the magnesium substrate. Thus the shredder efficiency is mainly influenced by the materials and joining techniques selected during product design [Castro et al. 2005; van Schaik and Reuter 2007b; Reuter and van Schaik 2008]. Afterwards the materials are separated by a series of automatic or manual sorting operations [Reuter et al. 2005, pp. 376-388]. A magnesium metal stream can be obtained. Depending on the materials used in the product design and the way they are connected, it can consist of a mixture of different magnesium alloys, aluminium alloys and contaminants such as oil, dirt, coatings, and inserts made of stainless steel, aluminium and copper [Westengen 2006]. These originate from incorrectly sorted or unliberated particles, as a trade-off between quality (grade) and quantity (recovery) of the stream is made during separation [Kelly and Spottiswood 1989]. The magnesium stream has insufficient quality to be remelted to high-purity standard alloys in the raw materials production phase and leaves the magnesium cycle as collected scrap (figure 1.2).

1.1.4

Raw materials production

From natural resources Magnesium from natural resources is produced using dolomite (42%), brines (36%) or seawater (18%) as resource [Kammer 2000]. Seventy-five percent of the world primary magnesium is produced in China [Kramer 2006], where the Pidgeon process is the dominant production method. It is a simple, low cost process with large environmental impact [Aghion and Golub 2006; Brooks et al. 2006]. The other method for magnesium production is electrolysis of MgCl2 using a molten salt as electrolyte [Habashi 1997; Aghion and Golub 2006]. Characteristics of both processes are given in table 1.1. The magnesium metal obtained from the reduction process is refined and alloyed according to standard alloy specifications for die-cast (table 1.2) and extrusion alloys. Before the ingots are cast dross is removed from the liquid metal surface. It is rich in metal, poor in oxide and salt so it is recycled as indicated in figure 1.2.

8


Table 1.1: Characteristics of primary magnesium production methods [Habashi 1997; Ramakrishnan and Koltun 2004; Brooks et al. 2006; Aghion and Golub 2006]. Electrolytic process → Mg(l) + Cl2(g)

Reaction

MgCl2(s

Feed (kg/kg Mg)

4 MgCl2 or 3.5 MgCO3 + 1 Cl2 + 0.5 C 700 - 800 1 · 105 (atmospheric) 18 - 28 hydro power, gas

Temperature ( ) Pressure (Pa) Energy (kWh/kg Mg) Energy Source

or l)

Pidgeon process 2MgO(s) + 2CaO(s) + FeSi(s) → 2Mg(g) + Ca2 SiO4(s) + Fe(s) 10-15 (Ca, Mg)CO3 + 1.1 FeSi + 10 C 1200 10 - 20 (vacuum) 45 - 80 coal, gas

From production scrap and EoL products Remelting of EoL materials conserves material and energy resources. It converts the residues created in the life cycle back to magnesium, utilizing material resources more efficiently. The energy consumption of remelting is only 3 kWh/kg magnesium, this is less than 5% of the energy consumption of primary production [Hydro 2007; Kammer 2000]. In particular when compared to the Pidgeon process large amounts of energy are saved during recycling. Processing and manufacturing residues and EoL scrap are classified based on their oxide content, salt content and type and degree of contamination because during remelting: any MgO already present in the feed material or created during remelting cannot be reduced back to magnesium, hence valuable magnesium metal is lost. copper, nickel or silicon impurities cannot be removed with conventional refining methods so high purity requirements for alloys might be difficult to reach [Ditze and Scharf 2005]. a salt slag is created, which has to be treated in a responsible manner [Boin 2001; Kammer 2000]. Several classifications are in use for example from Antrekowitsch and Hanko [2002]; MEL [2007]; Boin [2001]. All agree on the composition of class 1, differences are found in the classification of the other scrap types. The classification by Boin [2001] is used in this thesis (table 1.3) and examples of the classes are shown in figure 1.4 on page 6. Industrial raw materials can be remelted with or without salt flux. The latter is only possible with class 1 scrap. It contains hardly any oxides, so a salt flux is not necessary and a cover gas is used for melt protection. High-purity alloys are obtained, so the quality of the metal in the scrap is preserved. Again dross and sludge (class 6) are created, which are recycled using a salt flux. [Hanko and Macher 2003]. Remelting with salt flux is suitable for all the other scrap classes (including class 1). Currently mainly class 1 and 6A are remelted; the other scrap classes are not recycled for technological and economical reasons. Class 2 - 7 scrap is a relatively small amount (table 1.4), although it will increase when the magnesium usage in consumer products

Al

3.7 3.7 4.5 5.6 5.6 8.5 8.5 8.5 4.6 5.6 1.9 1.9

-

4.8 4.8 5.3 6.4 6.4 9.5 9.5 9.5 5.5 6.6 2.5 2.5

Cu

0.04 0.015 0.008 0.25 0.008 0.08 0.25 0.025 0.008 0.008 0.008 0.008

Cu

Fe

... 0.0035 0.004 ... 0.004 ... ... 0.004 0.004 0.004 0.004 0.0035

Fe

9990A 0.003 ... 0.04 9995A 0.01 ... 0.003 9998A 0.004 0.0005 0.002 Maximum values unless range is given.

AS41A AS41B AM50A AM60A AM60B AZ91A AZ91B AZ91D AJ52AD AJ62AD AS21A AS21BD

Al

... ... 0.001

Pb

0.22 - 0.48 0.35 - 0.6 0.28 - 0.50 0.15 - 0.50 0.26 - 0.50 0.15 - 0.40 0.15 - 0.40 0.17 - 0.40 0.26 - 0.5 0.26 - 0.5 0.2 - 0.6 0.05 - 0.15

Mn

0.004 0.004 0.002

Mn

0.01 0.001 0.001 0.01 0.001 0.01 0.01 0.001 0.001 0.001 0.001 0.001

Ni

0.001 0.001 0.0005

Ni

... ... ... ... ... ... ... ... ... ... ... 0.06 - 0.25

Rare Earth

0.005 0.005 0.003

Si

0.60 - 1.4 0.60 - 1.4 0.08 0.20 0.08 0.20 0.20 0.08 0.08 0.08 0.7 - 1.2 0.7 - 1.2

Si

... ... ...

Na

... ... ... ... ... ... ... ... 1.8 - 2.3 2.1 - 2.8 ... ...

Sr

... ... ...

Sn

0.10 0.10 0.20 0.20 0.20 0.45 - 0.9 0.45 - 0.9 0.45 - 0.9 0.20 0.20 0.20 0.25

Zn

... 0.01 0.001

Ti

... 0.01 0.01 ... 0.01 ... ... 0.01 0.01 0.01 0.01 0.01

Other metallic

0.01 0.005 0.005

Other impurities

0.30 ... ... 0.30 ... 0.30 0.30 ... ... ... ... ...

Other impurities

Table 1.2: Chemical requirements for alloys used for die castings according to the ASTM B-93 standard [ASTM 2005] and for magnesium used for remelting according to ASTM B-92 [ASTM 2001].

1.1. THE PRODUCT LIFE CYCLE 9

10


Table 1.3: Classification of magnesium scrap based on type of contaminants present [Boin 2001]. Associated contaminations

1A Compact, clean 1B Thinwalled, clean 2 New + Fe, Al inserts 3 Collected scrap 4 Unclean, oily scrap 5A Clean swarf/chips 5B Oily swarf/chips 6A Dross from casting 6B Sludge 7 Salt slag

None

Oil

Paint/ coating

Oxide, sand

Flux, salt

Metal: Fe, SS, Al

Metal: Si, Cu, Mn, Ni

X X X -

X X X -

X X -

X X X X X X

X

X X X -

X X X

increases. The scrap is molten in open or covered crucibles between 700 to 750 and a salt flux, a mixture of KCl, NaCl, MgCl2 and 0 - 15 wt.% CaF2 , is added to remove solid particles. Low-density salt flux (low CaF2 -content) protects the metal surface against oxidation and is a ”packaging material” for nonmetallic component particles, such as oxides. To make sure the salt flux floats on the magnesium a density difference of 0.15 - 0.20 g/cm3 compared to the metal is required. This salt slag (class 7), high in oxide-content and low in salt-content, is removed from the surface of the metal bath before refining. High density salt flux, are used for refining the metal [Javaid et al. 2006; Avedesian and Baker 1999]. The CaF2 -content is high so it can strip the oxide layer from the magnesium droplets and the droplets can coalesce. Moreover a density difference of 0.5 - 0.8 g/cm3 between salt and metal is created so the salt will sink to the bottom of the crucible. During its descend to the bottom the salt takes up oxides and other solid particles but leaves dissolved impurities behind, cleaning the melt along the way. After the metal has been transferred to a holding furnace, the salt slag is removed [Kammer 2000]. The amount of salt flux to be added depends on the scrap class. Class 1 scrap requires about 1% of the scrap mass, while more contamined scrap needs more than 5% to capture all particles [Javaid et al. 2006; Kammer 2000]. Inserts, present in class 2 and 3 scrap (rejects and collected scrap), lower the quality of the metal during remelting. In particular copper and nickel dissolve very well in magnesium and dilution is often the only solution to meet alloy specifications [Tathgar, Engh and Bakke 2000]. Aluminium, manganese, zinc and silicon originate from inserts as well as from incorrectly sorted magnesium or aluminium alloys, which is typical for collected scrap. These elements are regarded impurities when the desired alloy does not contain them. The produced magnesium alloy can then not be recycled back to die-castings, instead it is re-used in applications such as sacrificial anodes, steel desulphurization [Oeters et al.


11

Table 1.4: Approximate amount of scrap per class produced in Europe in 2003 [Ditze and Scharf 2005], and worldwide in 2000 [Javaid et al. 2006].

1A Compact, clean, new 1B Thinwalled, clean, new 2 New with Fe, Al inserts 3 Collected scrap 4 Unclean, oily scrap 5A Clean swarf and chips 5B Oily swarf and chips 6A Dross 6B Sludge 7 Salt slag

Europe (t/y)

World (t/y)

n.a. n.a. n.a. 8 000 n.a. n.a. 1 000 2 500 400 4 000

22 693 19 926 n.a. n.a. n.a. n.a. 2 768 2 768 n.a. n.a.

2008] or alloying of aluminium [Westengen 2006] for which the purity requirements are less stringent. In this manner it leaves the magnesium life cycle and becomes unavailable for re-use/recycling. Coatings and paint layers, oil and grease present on the scrap contaminate the melt. De-coating prior to remelting is necessary to avoid contamination as well as emissions of toxic gaseous compounds [Antrekowitsch and Hanko 2002; Westengen 2006]. Additional salt flux is needed to remove the increased amount of solid particles, creating more salt slag. The silicon used in lubricants ends up in the metal phase [Westengen 2006], so extra refining or dilution might be necessary to meet alloy specifications. The high surface to mass ratio of swarf and chips (class 5) leads to excessive metal loss due to oxidation during remelting. The metal recovery is lower, while extra salt flux is necessary to capture the created oxide particles. Briquetting is done to reduce the surface area of the swarf (figure 1.4d, page 6). Clean chips are remelted or used for steel desulphurization, while remelting of oily chips is technically feasible although the costs might exceed the metal value in the scrap [Westengen 2006]. Both sludge (class 6B) and salt slag (class 7) are currently not further treated. These materials contain salt flux, intermetallic and oxide particles, and entrapped magnesium particles, making it difficult to obtain a usable alloy [Westengen 2006]. Because of its reactive nature it is classified as hazardous waste, so it has to go to special waste storage places. An adequate treatment for this material is necessary since its amount will increase when more magnesium is used and remelted. Central treatment comprising of crushing and sieving to obtain a metal, oxide and salt fraction, as done with aluminium salt slag, is not straightforward. Several complicating factors exist [Boin 2001]. First each remelter uses its own salt composition. When the salt slags are treated together a new salt flux mixture will be created, which cannot be sold back to the remelter. Second, the magnesium metal that is recovered is a mixture of all alloy types. Remelting of this material creates an alloy that does not fit any alloy specification, because of its heavy metal content. Hence a new

12


application/destination has to be found for this metal. Finally as the annual amount of salt slag is small, about 4000 tonnes per year in Europe, treatment is not economically feasible because of unfavourable economies of scale. For the moment dross and salt slag are leaving the life cycle (figure 1.2).

1.2

Closing the cycles

Coated semi-products are present in two of the four residues that leave the magnesium life cycle, namely in rejects and collected scrap. When these semi-products are recycled the amount and composition of the dross and salt slag residue is affected as well. To recycle coated magnesium in a sustainable manner, the impact of the coating on the remelting process has to be known. During product design selection of the coating takes place, fixing the coating characteristics and properties from this moment on for the rest of its life cycle. The choices made in the manufacturing phase are therefore directly linked to the recyclability of coated semiproducts as well as the sustainability of the magnesium cycle as a whole. Thus a system perspective is necessary for the evaluation of the ’sustainability’ of coated magnesium recycling. First the concept of sustainability is clarified together with the strategies that can be followed to attain sustainable life cycles. Progress towards sustainability can be monitored using a range of indicators based on measurable process and product parameters. Their suitability for the evaluation of coated magnesium and its recycling is discussed. Last the approach used in this thesis for the evaluation of the impact of coatings on remelting and for the evaluation of the sustainability of coated magnesium is explained.

1.2.1

Sustainability

The concept of sustainability and sustainable development are firmly rooted in the view of native Americans, Aboriginals and Inuit people. That the interaction of mankind with nature could have negative side-effects was in the Western world at large not recognized until the late 19th century. It led to the establishment of Yellowstone Park (USA) in 1872: the first National Park in the world [NPS 2007]. Many others followed. During the 1960s and 1970s the concern of the general public for environmental issues increased. The Green movement developed as a reaction on increasing environmental destruction, air pollution and the energy crisis. It resulted in the foundation of organisations like the Club of Rome (1968), the Environmental Protection Agency EPA (1970), Greenpeace (1971) and the United Nations Environment Programme UNEP (1972) as well as implementation of environmental protection measures. The work of the Brundtland Commission brought sustainability and sustainable development to the general public. The Commission defined ’sustainability’ as [WCED 1987, p. 8]: To meet the needs of the present without compromising the ability of future generations to meet their own needs.

1.2. CLOSING THE CYCLES

13

This is quite a philosophical definition that describes a process of change, which involves environmental stewardship, social equity and justice and economic issues. No numerical values and specific targets are mentioned in the definition so it can fulfill its role as ’future landing place’ without becoming outdated in a changing world. Sustainability is thus a fluid, dynamic goal. A more specific description that is easier to use in practice are the Four System Conditions developed by The Natural Step organization [Robèrt et al. 2002]. The first three conditions address the use and depletion of natural resources, the creation and treatment of waste, the preservation of biodiversity and the preservation of soil and water quality. The last condition relates to social and economic equity. While sustainability describes the goal, sustainable development describes how this goal can be reached. Actions such as recycling, reduction of resource use or phasing out harmful substances are examples of this. Strategic investments, social principles such as transparency and dialogue, and political means, can be used to create the appropriate circumstances for action [Robèrt et al. 2002]. To achieve sustainable development a good relationship between society and technology is necessary [Aoe 2007] and involvement of ethics is crucial [Walker 2006].

1.2.2

Strategies to achieve sustainability

Strategies such as Industrial Ecology, Cleaner Production, Zero emission, Eco-efficiency and Design for Environment/Recycling/Sustainability can be used as guidance for the actions that should be taken. They have to be combined with the appropriate environmental indicators to measure the progress towards sustainability. Despite the developed strategies, indicators and good intentions Korhonen [2007] notes that the real and concrete progress in sustainable development has been amazingly slow during the past 30 years.

Industrial Ecology established in 1989 can be defined as the study of the flows of materials and energy in industrial and consumer activities, of the effects of these flows on the environment, and of the influences of economic, political, regulatory, and social factors on the flow, use, and transformation of resources [Lifset and Graedel 2002, p. 4]. More concretely it looks at the effect of product design and manufacture on the environment by examining material and energy flows and using ecosystems as models for industrial activity. Industrial ecology has a strong emphasis on closure of the materials cycles, which manifests itself in the form of a life cycle perspective, materials and energy flow analysis and system modelling. In this strategy technology is regarded as a vehicle to avoid and solve environmental problems. Main instruments or actions are de-materialization (using less materials for a product), resource efficiency improvement (using materials more efficiently) and design for environment (adjusting product design to lower the environmental impact). Furthermore social and economical aspects such as consumption pattern and organizational arrangements are studied to bring about changes in companies and society.

14


Cleaner Production is a preventive strategy that starts from an environmental viewpoint. It involves two pathways: efficiency improvement and materials substitution [Jackson 2002]. Processes based on these principles conserve resources, eliminate toxic and dangerous materials and wastes and reduce the quantity of emissions at the source. Products made with Cleaner Production principles in mind have a reduced impact on environment, health and safety during their life cycle [UNEP 2001a]. Eco-efficiency encompasses the delivery of competitively priced goods and services that satisfy human needs and bring quality of life, while reducing ecological impacts and resource intensity throughout the life cycle, to a level at least in line with the earth’s carrying capacity [UNEP 2001b]. It is the most used approach in industry, as it starts from an economical viewpoint. The strategy is criticized for its large focus on efficiency. As efficiency is determined by the larger system it is a part of it cannot be used as a goal in itself [McDonough and Braungart 2002, p.65]. As a result eco-efficiency often only minimizes the material flows instead of changing them [Braungart et al. 2007]. Moreover, it is impossible to know what the most efficient technology is or will be [Korhonen 2007]. Furthermore the costs to the environment and human health are usually not reflected in the retail price of products, hence unsustainable goods can still be more competitively priced than sustainable goods [Lakhani 2007]. Cradle-to-cradle design or eco-effectiveness has as goal to generate cyclical, metabolisms that maintain or upgrade the quality and productivity of all material resources through many cycles of use [McDonough and Braungart 2002]. Mineral or synthetic materials have to remain in a closed-loop system of manufacture, recovery, and re-use while bio-degradable materials are returned to the environment to feed biological processes [Braungart et al. 2007]. In this manner no waste will be created. Eco-effectiveness is achieved by cradle-to-cradle design of products and processes incorporating cost, performance and aesthetics as well as ecological intelligence, justice and fun using among others substitution, re-design and efficiency as tools. The cradle-to-cradle philosophy has already been adopted by companies such as Ford and Nike [Roston 2002] and is currently embraced in the Netherlands with such enthusiasm it could be called a hype [Martens and Amelung 2007]. This culminated in the development of a cradle-to-cradle education program at the master-level, which should start in September 2008 [van de Sandt 2008]. In this sense cradle-to-cradle might be regarded as the strategy which is best marketed to the general public. Cradle-to-cradle is also criticised, because of the uncertainty about achieving sustainability. The strategy is more associated with making money, continuation of consumption behaviour and technological changes, while social, economical and cultural changes are neglected [Martens and Amelung 2007]. The question if the consumption behaviour should be changed, is not asked [Keuning 2008]. Also the limitations of the uptake of nutrients in the biosphere is underexposed. Furthermore the goal of cradle-to-cradle and its tools are already included in strategies that are worked out better such as Industrial Ecology and Cleaner Production [Zeilmaker 2008].


15

Design for Recycling (DfR) means recycling considerations are incorporated into product and process design from the start so environmental impact is minimized. It is a product oriented approach and implicitly a life cycle perspective and systems approach are taken as impacts throughout the product life cycle are considered [Lifset and Graedel 2002]. DfR requires a fundamental description of product recycling systems which can be combined with that (description) of the design of the product [van Schaik and Reuter 2004a]. In practice this means that advanced recycling simulation and optimization models need to be linked to design (CAD) software as done by van Schaik and Reuter [2004a]; Reuter et al. [2006]; van Schaik and Reuter [2007b]. When based on recycling technology DfR will increase both the recycling rate of the product and the quality of the material streams produced from the product during recycling [van Schaik and Reuter 2007a].

1.2.3

Methods to assess sustainability

Progress towards sustainability has to be monitored using appropriate environmental indicators or metrics [Wall and Gong 2001; Dewulf and Van Langenhove 2005]. An indicator should be an easy to understand, unambiguous, measurable quantity, which reflects the status of large systems and is based on fundamental principles [OECD 1998]. Indicators show if the objectives set by the chosen strategy and actions are coming closer. A range of complementary indicators is available, each with a different focus and level of detail so it fits to the actions to be monitored [Robèrt et al. 2002]. Regular re-evaluation of the system will give the progress over time. Mass Flow Analysis describes the flow of mass in processes, companies or countries based on the First Law of thermodynamics. All streams are expressed in units of mass, thus the changes in quality of the flows cannot be directly assessed. Selection of the most preferable process based on a mass flow analysis is difficult, because MFA does not link the different scrap types or wastes produced to environmental impact [Behrens et al. 2007]. MFA of magnesium scrap remelting have been made by Ditze et al. [Ditze and Scharf 2003, 2005]. The balances provide information about the distribution of impurities over the product and waste streams, and a overview of the flows entering and leaving. Other indicators based on mass flow analysis are among others the Ecological Rucksack and MIPS. The Rucksack of a product compares the total material input during its life cycle to the weight of the product itself and is a rough estimate of the resource intensity. A more detailed approach is Material Input Per unit Service (MIPS), which describes the material consumption from cradle to cradle per unit service or function delivered by a product [Schmidt-Bleek 1993]. Life Cycle Analysis identifies and quantifies the resource usage and waste creation during the life cycle of products, and assesses the impacts of those on the environment ¨ [Finnveden and Ostlund 1997]. After defining the product system and purpose of the LCA, first the Inventory analysis is compiled. It quantifies all input and output flows and links the contribution of each flow to environmental effects. The Impact assessment converts the inventory analysis to environmental impacts by calculating the value of the

16


environmental indicator. Thus the impact ’climate change’ is quantified by calculating the global warming impact in kg CO2 -eq. and the impact ’extraction of resources’ has resource depletion rate as indicator. A single impact score can be obtained by multiplying all impacts by weighing factors and summation of the outcome. The use of weighing factors is seen as a source of manipulation and as the major drawback of the LCA method [Gong and Wall 2001]. Life Cycle Analyses have been made for magnesium car components consisting of 70% primary and 30% recycled magnesium [Ramakrishnan and Koltun 2004; Koltun et al. 2005; Hakamada et al. 2007; Tharumarajah and Koltun 2007]. The studies have been limited to the impact of the component on climate change and energy usage. Magnesium production from raw materials causes only 22% of the total impact. The Pidgeon process has the highest impact, using 354 MJ primary energy per kg Mg and emitting 42 kg CO2 eq. per kg Mg [Ramakrishnan and Koltun 2004]. Electrolytic production of magnesium reduces the greenhouse gas emission to 19 kg CO2 -eq. per kg Mg, while remelting of magnesium emits only 1.7 - 3.6 kg CO2 -eq. per kg ingot and consumes 11.4 MJ/kg magnesium of primary energy [Tharumarajah and Koltun 2007; Hakamada et al. 2007]. The use phase accounts for the remaining 78 % of the environmental impact. The above-mentioned LCAs did not address other environmental impacts such as resource depletion, quality loss and creation of the residues, which are indicated in figure 1.2 on page 3. Furthermore only uncoated parts have been studied. Exergy analysis describes the quantity and quality of material and energy flows, as both types of flows have a definable and calculable exergy content. Detailed knowledge about the composition of all flows is required, making it particularly suitable for a fundamental assessment of industrial processes. Exergy analysis is founded on the First and Second Law of Thermodynamics. Mass and energy within a system or process are conserved (First Law), unlike exergy which is partially destructed (Second Law). The amount of destructed exergy and the exergy of waste materials can be related to environmental impact, hence exergy can be used as a measure of sustainability [Rosen and Dincer 1997; Ayres et al. 1998; Wall and Gong 2001; Bastianoni et al. 2007]. Despite this, exergy analysis is rarely used to evaluate the sustainability of recycling systems. Dewulf et al. [2000] developed indicators to quantify the sustainability of technological processes. The indicators use exergy to describe resource renewability, emission toxicity, material re-use and recoverability and process efficiency. Ignatenko et al. [2007] assessed the performance of the car recycling system. Waste production, heat loss, quality loss and energy conversion loss are quantified with exergy. Performance of the recycling system is linked to product design using the exergetic efficiency and primary resource intensity of the system as indicators. Design for Recycling models link product design, recycling technology and environmental impact models to each other to provide a technological basis for Design for Recycling. The advanced recycling optimization models developed by van Schaik and Reuter [2004b]; Reuter et al. [2006]; Ignatenko et al. [2007]; Ignatenko [2008] for the physical, metallurgi-


17

cal and thermal treatment of End-of-Life Vehicles predict the input for and performance of recycling scenarios and the influence of design on the recycling rate of a product [van Schaik and Reuter 2004a]. The models deal with impact of the distributed nature of EoL products and the change in product design over time on the recycling system and recycling rate. Considered in the models are [Reuter et al. 2006]: Material quality and calorific value as a function of 1) material, 2) material combinations and connections made during product design, 3) particle size and 4) degree of liberation i.e. material combinations in particles after shredding. Economic value of intermediate streams. Separation physics and thermodynamics of unit operations. Losses and emissions. Flexible flow sheet layout to accommodate changing product design. Distributed and dynamic properties of present and future (car) design. The models employ exergy analysis as a tool to quantify resource and recovery efficiency so the sustainability of the recycling system’s performance can be estimated [Ignatenko et al. 2007]. In this way evaluation and optimization of the entire (recycling) system on a technical and material quality basis with regard to material and energy recovery is enabled [Ignatenko 2008]. Furthermore the models can indicate which technological solutions need to be developed to achieve the desired recycling rates and if this is possible from a technological and economic point of view [Reuter et al. 2006]. The advanced recycling model is presently used to calculate recycling rates of car designs by the European car industry [Reuter and van Schaik 2008]. The actual connection between the product design and the recycling optimization model as well as between the environmental impact models and the recycling model is made using fuzzy rule models developed by van Schaik and Reuter [2007b], which are presently integrated in CAD software [van Schaik and Reuter 2007a] and with life cycle assessment tools for environmental impact evaluation [van Schaik and Reuter 2007b] by the European car industry. In the Design for Recycling models, the product design made by the designer in CAD is inserted in the fuzzy rule liberation model that predicts the characteristics of the materials obtained after shredding the product. This serves as the input to the advanced recycling optimization model, which calculates among others the recycling rate of the product and the quality of the materials after recycling for a selected recycling process/route. Using the outcomes of the recycling model the fuzzy rule models gives feedback to the designer about the recyclability of the product. He or she can directly see the consequences of the chosen material combinations, connections and joints on the recycling rate (recyclability) [Reuter and van Schaik 2008]. Also the fuzzy rule models can provide the necessary fundamental information about the behaviour of products during recycling to LCA tools improving environmental evaluation. The outcomes of LCA can be used in the design process as well.

1.2.4

Evaluation of coated semi-product recycling

In the magnesium life cycle several problematic residues are created:

18


Dross and fine material (product manufacture). Rejects and swarf (product manufacture). Collected scrap (End-of-Life). Dross and slag (raw materials production). These residues inhibit closed loop recycling of magnesium semi-products, posing a burden on society and environment. Sources of magnesium metal are left unused, as the residues are disposed or dispersed through use in other industries. At the same time primary magnesium production continues to consume a disproportional amount of energy and material resources compared to remelting of magnesium residues. Increased demand for magnesium semi-products in consumer goods will increase the amount of residues leaving the cycle, as all phases (and residues) in the life cycle are interconnected. Legislation for example the European Union’s End-of-Life Vehicle (ELV) and the Waste Electronic and Electrical Equipment (WEEE) directive [EU 2000, 2003] serves as an extra incentive for closure of the magnesium cycle in a sustainable manner within a limited time span. For example, the implementation of the ELV directive has led to the development of advanced recycling optimization models and Design for Recycling models for cars [van Schaik and Reuter 2004a, 2007a,b]. In other words: minimization of the amount of the created residues while simultaneously preserving the quality of magnesium metal is necessary. Failure to do so results in magnesium losing its (appeal) attractiveness as a metal for structural applications affecting industries throughout the life cycle. Solutions contributing to closure of the life cycle are urgently needed from an environmental as well as an socio-economical point of view. Changes in the magnesium cycle itself are necessary to achieve sustainable closed life cycles and involves adjustment of product design and recycling processes. To evaluate and optimize the recycling system for coated magnesium semi-products, detailed understanding of the parts of the system as well as their interconnectivity is a prerequisite [Reuter et al. 2005]. Hence a fundamental study of the influence of coated semi-product characteristics (i.e. product design) on the magnesium recycling/remelting process and its products is necessary. The outcomes will fit in the DfR approach as the study involves the impact of design choices on recycling, as well as prediction of this impact and feedback to the design phase. The principles of Industrial Ecology will provide the framework underlying this study, because of its holistic approach and emphasis on prevention and the use of technology as means for problem solving. The Industrial Ecology toolbox has to be extended with metrics based on first principles, since the overall performance of the system is determined by physics, thermodynamics and material quality. These three have to be addressed simultaneously [Ignatenko et al. 2007]. Exergy analysis is used to quantify the impact of coatings present on the semi-product on the remelting process. Exergy is based on thermodynamic principles so quantity and quality changes of the remelting product (metal) and residues (salt slag) are taken into account. Furthermore exergy can deal well with the level of detail required for the evaluation. MFA and LCA are unsuitable for this purpose. Kinetic analysis is applied to assess the speed of the interactions between coating and metal during remelting. Lab-scale remelting experiments and simulations of the remelting

1.3. OUTLINE OF THE THESIS

19

process provide the necessary information for kinetic and exergy analysis. The suitability of thermal de-coating, a common EoL treatment to improve recycling, for coated magnesium is investigated experimentally. The de-coating process is described and the ease of de-coating is quantified with kinetic analysis. Based on the results expressions for the (three) parameters describing remelting and decoating are derived, which are linked to coating characteristics. The detailed knowledge about coated magnesium and its behaviour during remelting fits to the existing Design for Recycling model, which includes an unit operation for magnesium recycling [van Schaik and Reuter 2004b; Reuter et al. 2006; Ignatenko et al. 2007; van Schaik and Reuter 2007b; Ignatenko 2008]. The current study addresses a very specific material, material connection method and unit operation in recycling optimization model of the DfR model at a very detailed level. It provides data which can be used in the advanced recycling optimization model for a better description of the recycling of coated magnesium. In this way it could aid in showing the impact of coated magnesium used in current and future product designs on the product recycling rate to designers. In this manner product design is connected to the recyclability of coated magnesium and to the sustainability of the whole magnesium and product life cycle. The three parameters, remelting thermodynamics, remelting kinetics and de-coating kinetics are visually represented in figure 1.5. Each axis represents a parameter, and a coated semiproduct has its own place in this 3-dimensional space. In this metric recyclability and sustainability are expressed based on fundamental principles. A direct quantitative link between product and process design on the one side and recyclability/sustainability on the other has been established. In this manner a ranking of coated products is made possible, so design choices unfavourable for recycling can be avoided and Design for Recycling is facilitated.

1.3

Outline of the thesis

Based on observations in the previous section the main research question can be formulated as: The quantification, prediction and minimization of the effect of coatings on the recycling system of magnesium scrap on the basis of thermodynamics and kinetic analysis. In order to assess and optimize the recyclability a fundamental metric will be proposed, which can easily be integrated in computer aided design tools for consumer product manufacture. Chapter 2 The necessary background information about kinetic analysis for thermal de-coating and exergy analysis for the evaluation of the remelting process is provided. Kinetic and exergy analysis are the tools used for evaluation of the recyclability of coated magnesium.

20


Figure 1.5: The three parameters of coated magnesium recycling combined into a 3-dimensional metric. The sphere represents the location of the coated semi-product [Meskers, Reuter, Boin and Kvithyld 2008].

1.3. OUTLINE OF THE THESIS

21

Chapter 3 A literature survey has been conducted to determine the coatings that can be applied on magnesium. Using the outcomes of the survey and experimental methods the structure and composition of the coating on six different coated semi-products is determined. These objects are used for the de-coating and remelting experiments. The outcomes of the experiments will be related to the coating properties determined in this chapter. Chapter 4 describes the EoL treatment of coated semi-products. Emphasis is on thermal de-coating of the four painted magnesium products introduced in chapter 3. First the experimental set-up and special considerations for thermal de-coating of magnesium are discussed. The de-coating process in argon and air atmosphere is described using the mass loss (DTG) curves of the objects. The chapter concludes with an evaluation of the suitability of the removal methods for coated magnesium. Chapter 5 Using the DTG curves obtained in chapter 4, kinetic analysis is used to quantitatively describe the observations and characterize the behaviour of each coating. Based on kinetic analysis an expression for the ease of de-coating is developed. This expression links de-coating behaviour to coating and product design. The coated semiproducts are ranked on ease of de-coating. Chapter 6 The impact of coatings and coating components on the remelting process is predicted using thermodynamical modeling and evaluated using exergy analysis. Their effect on the metal and salt slag is described. Single coating layers as well as the coatings found on the magnesium semi-products have been used in the simulations. The losses during remelting due to oxidation, intermetallic formation, composition changes are quantified in terms of exergy. A comparison of the remelting losses of the coating components is possible. Chapter 7 The predictions obtained in chapter 6 are verified by laboratory experiments. Flux - oxide and magnesium - oxide equilibria have been investigated. The interaction between magnesium, flux and oxide are discussed in more detail for three different oxides. Furthermore coated and de-coated semi-products are remelted. Chapter 8 The impact of coatings on the remelting process is characterized in a quantitative manner by a single parameter: R, which represents the remelting losses. For this the exergy analysis and the remelting losses from chapter 6 are used. The different coatings and the coated semi-products can be grouped according to their impact on the remelting process. A distinction between components with high and low recyclability can be made. This provides a direct relation between coating characteristics/design and the remelting process. Reaction kinetics, their influence on the remelting process and the recyclability are briefly discussed.

22


Chapter 9 The three parameters describing the recycling of coated magnesium are combined to a metric for recycling. It describes the coated semi-products taking into account pre-treatment and remelting using parameters based on first principles. This metric can be used as a tool for product designers, coating designers/manufactures and remelters to assess the recyclability of coated magnesium semi-products. Furthermore the metric indicates improvements by comparing the actual recyclability to desired recyclability. Chapter 10 Based on the findings in the previous chapters, conclusions are drawn and recommendations are made.

1.4

Publications from this thesis

Journal publications: Meskers, C.E.M.; Reuter, M.A.; Boin, U. and Kvithyld, A. (2008), A fundamental metric for recycling applied to coated magnesium. Metallurgical and Materials Transactions B, vol. 39, no. 3, pp. 500-517. Meskers, C.E.M.; Xiao, Y.; Boom, R.; Boin, U. and Reuter, M.A. (2007), Evaluation of the recycling of coated magnesium using exergy analysis. Minerals Engineering, vol. 20, no. 9, pp. 913-925. International conferences: Meskers, C.E.M.; Reuter, M.A. and Boin, U., (2008) Design for recycling - a key to sustainable magnesium application. In: REWAS 2008 - global symposium on Recycling, Waste Treatment and Clean Technology. 12 - 15 October, Cancun, Mexico. Meskers, C.E.M.; Xiao, Y.; Boin, U.; Boom, R. and Reuter, M.A. (2007) Assessment of the resource efficiency of remelting & refining of coated magnesium. In: Hilty, L.M.; Edelmann, X. and Ruf, A. (Eds.), R’07 World Congress - Recovery of Materials and Energy for Resource Efficiency, pp. 1-6. 3 - 5 September, Davos, Switzerland. ISBN: 978-3-905594-49-2 Meskers, C.E.M.; Boin, U.; Boom, R.; Xiao, Y. and Reuter, M.A. (2007) Predicting the influence of coatings on recyclability of magnesium using exergy analysis In: Beals, R.; Pekguleryuz, M. and Neelameggham, N. (Eds.), Magnesium Technology 2007, pp. 293-298. 25 February - 1 March 2007, Orlando, USA. ISBN: 978-087339-663-9 Meskers, C.E.M.; Xiao, Y.; Boom, R.; Boin, U. and Reuter, M.A. (2006) Evaluation of the recycling of coated magnesium using exergy analysis. Extended abstracts of Material, Minerals & Metal Ecology (MMME06), 14 - 15 November 2006, Cape Town, South Africa. (CD-Rom) Meskers, C.E.M.; Kvithyld, A.; Reuter, M.A.; Engh, T.A. (2006) Thermal decoating of magnesium - a first step towards recycling of coated magnesium. In: Luo, A.; Neelameggham, N. and Beals, R. (Eds.), Magnesium Technology 2006, pp. 33-38. 12 - 16 March 2006, San Antonio, USA. ISBN: 978-0-87339-620-2

1.4. PUBLICATIONS FROM THIS THESIS

23

Smaller conferences: Meskers, C.E.M.; Xiao, Y.; Boom, R.; Reuter, M.A. and Boin, U. (2006) Recycling van coated magnesium - uitdagingen en huidig onderzoek. Mag IV - Van theorie naar praktijk, 13 June 2006, Lelystad, the Netherlands. Meskers, C.E.M.; Boin, U. and Reuter, M.A. (2005) Recycling of coated magnesium - thermal de-coating of magnesium die-castings. TPA dag IOP Oppervlaktetechnologie, 21 June 2005, Maarssen, the Netherlands. Meskers, C.E.M.; Reuter, M.A. and Boin, U. (2004) Recycling of coated magnesium. Mag III - Magnesium in beweging, 4 November 2004, Houten, the Netherlands (poster presentation).

24


Chapter 2

Theoretical background

The appendices A and E referred to in this chapter can be found on the CD in the back of this book.

26

2.1

CHAPTER 2. THEORETICAL BACKGROUND

Kinetic analysis

Kinetic analysis can be used to describe the influence of temperature on the reaction rate of any reaction or process. In this thesis kinetic analysis is primarily used to describe the mass loss during thermal removal of organic coatings on magnesium semi-products. The measured change of coating mass over time forms the input from which the activation energy, the frequency factor and the reaction model of the de-coating reaction(s) can be determined. Using these kinetic parameters predictions about the behaviour of the coating during de-coating can be made. A solid-state decomposition reaction can be described as [Galwey and Brown 1998]: dα = k(T ) f (α) dt

(2.1)

which is an expression for the reaction rate. In this case that is the rate of decomposition of the organic coating. f(α) is a function describing the reaction model (table 2.1) and k(T) is the reaction rate constant. α represents the fraction of the coating that is decomposed. It is defined by equation 2.2 in which mi and mf indicate initial and final mass. α=

mi − m(t) mi − mf

0≤α≤1

(2.2)

The temperature dependent factor k is mostly described by the Arrhenius equation: k(T ) = A e −Ea /(RT )

(2.3)

Integration of equation 2.1 over the duration of the reaction describes the reaction itself. Thus the integral function g(α) is obtained by integrating α from zero to α and t from zero to t and inserting the Arrhenius equation for k. After integrating, the integral reaction model of a decomposition reaction during an isothermal heating program becomes g(α) = A e −Ea /(RT ) t

(2.4)

When instead of an isothermal temperature program a constant heating rate program is used, equation 2.1 has to be adjusted. The change in α is now related to the change in temperature instead of to the change of time during the reaction. To achieve this dα/dt is replaced by β(dα/dT) where β is the heating rate. dα A −EA /(RT ) = e f (α) dT β

(2.5)

The intermediate steps of this substitution can be found in appendix A.1. After integration an expression for g(α) is obtained: g(α) =

A β

Z 0

T

e −Ea /(RT ) dT

(2.6)

2.1. KINETIC ANALYSIS

27

Table 2.1: Solid-state rate expressions for different reaction models [Galwey and Brown 1998]. Model Nucleation models: Power law (P) Avarami-Erofe’ev (A2) Avarami-Erofe’ev (A3) Avarami-Erofe’ev (An) Geometrical contraction models: Contracting area (R2) Contracting volume (R3) Diffusion models: 1D diffusion (D1) Reaction-order models: Zero-order (F0) First-order (F1) Second-order(F2)

Differential form f(α)

Integral form g(α)

nα(n−1)/n 2(1 − α)[− ln(1 − α)]1/2 3(1 − α)[− ln(1 − α)]1/3 n(1 − α)[− ln(1 − α)]1/n

α1/n [− ln(1 − α)]1/2 [− ln(1 − α)]1/3 [− ln(1 − α)]1/n

2(1 − α)1/2 3(1 − α)2/3

1 − (1 − α)1/2 1 − (1 − α)1/3

1/2 α

α2

1 (1 − α) (1 − α)2

α − ln(1 − α) (1 − α)1 − 1

The integral can be rewritten by replacing Ea /(RT) with y [Khawam and Flanagan 2005a]. The exponential integral in equation 2.7 has no analytic solution, instead it is approximated to obtain a solution. Here an approximation from Jahnke and Lösch [1966] is used (eqn. A.14. Z AEa ∞ e −y dy (2.7) g(α) = βR y y2 The kinetic triplet consisting of activation energy E, frequency factor A and reaction model f(α) or g(α), can be used for characterisation of de-composition reactions. The sinlge use of the activation energy of the reaction is not sufficient for characterisation [Maciejewski 2000]. Determination of the triplet can be done using a model-fitting or isoconversional model approach. Model fitting involves two steps. Step one is determination of the reaction model that best fits the α vs. T or t curve, followed by calculation of Ea and A using the Arrhenius equation [Khawam and Flanagan 2005b]. The model fitting approach has three drawbacks. The kinetic triplet is assumed constant and can be determined from a single experimental curve. Further, only a single reaction rate is determined [Khawam and Flanagan 2005a]. Instead of model fitting isoconversional methods should be used. These methods do not assume a reaction model when determining the activation energy at progressive values of α [Brown et al. 2000; Ozawa 2001; Khawam and Flanagan 2005a; Vyazovkin and Sbirrazzuoli 2006]. The methods can detect the dependency of Ea on α, which indicates that a multistep reaction mechanism occurs [Brown et al. 2000]. In that case each step contributes to the effective activation energy and the overall reaction rate will vary with temperature and α [Vyazovkin 2000]. The relation between activation energy and fraction converted can aid in exploring the effect of polymer composition on the degradation kinetics [Vyazovkin

28


and Sbirrazzuoli 2006]. Depending on the temperature program used different equations are available to determine the values of the kinetic triplet. When isothermal experiments are carried out, the standard isoconversional method is based on equation 2.4 which is rewritten as [Vyazovkin 2000]: − ln tα, i = ln

A g(α)

−

Eα RTi

(2.8)

Knowledge about g(α) or ln A is not required to determine Eα , because the slope of a plot of − ln tα versus 1/Ti enables calculation of the activation energy for each α. For constant heating rate temperature programs the integral in equation 2.6 has to be approximated so the dependency of the activation energy on α can be determined. The methods from Ozawa, Wall and Flynn use the Doyle approximation of the exponential integral [Khawam and Flanagan 2005a]. In this work the Coats-Redfern approximation (eqn. A.17 is employed [Coats and Redfern 1964], which gives after inserting it in equation 2.6: ln

Eα βi = const. − 2 Tα,i RTα,i

(2.9)

g(α) is included in the constant (first right hand term), so no reaction model has to be presumed to obtain the activation energy from a plot of ln(βi /T2α,i ) vs. 1/Tα,i . To be able to use isoconversional methods for the kinetic analysis of constant heating rate experiments the experimental data from three different heating rates has to be available. Another way to determine the activation energy is through the Kissinger equation (eqn. 2.10) [Kissinger 1957]. Tp,i is defined as the temperature at which the reaction rate is maximum. β Ea ln (2.10) = const. − 2 Tp RTp The activation energy is determined from the slope of a plot of ln β/T2p versus 1/Tp . Although it appears to be an isoconversional method, it is not. In the Kissinger equation one of the assumptions that the activation energy does not change during the reaction. It is only determined at Tp instead of at each α so it is independent of the fraction converted. Further it is assumed the reaction model is constant. Vyazovkin and Sbirrazzuoli [2006] note that the fraction decomposed at the peak temperature changes with heating rate, which introduces uncertainties in the value of Ea . The other two parameters of the kinetic triplet can now be determined. An expression for g(α) is found by fitting known reaction models like the ones in table 2.1 to the experimental data. For mass loss under isothermal conditions equation 2.4 is used, while for constant heating rate conditions equation A.12 is used with a suitable approximation for the exponential integral (eqn. A.14). The previously calculated Ea is inserted, expressions for g(α) (table 2.1) are attempted and simultaneously a value for A is obtained.

2.2. EXERGY ANALYSIS

2.2

29

Exergy analysis

Exergy is used to measure the mass and quality losses which occur during the production of magnesium from industrial materials like coated semi-products (figure 1.2). Exergy analysis quantifies the created residues and can assist in reducing them so the magnesium life cycle can be closed. Furthermore exergy can point out the deviations between the desired standard alloy composition given in table 1.2 and the obtained alloy after remelting. In this section exergy (analysis), its suitability as an indicator for ’sustainability’ and its application to magnesium recycling are clarified.

2.2.1

Concept

Exergy (E) is as essential to the functioning of the Earth, as electricity is to a light bulb [Ayres et al. 2007]. The sunlight reaching the earth is a flow of energy. All of this energy can be used for work, hence its energy equals its exergy. Fundamental processes in nature like geological processes, biological processes,wind, tide and rain, are driven by sunlight. Photosynthesis converts CO2 and water to nutrients and biomass, using exergy from sunlight and incorporating it in nutrients and biomass. The food and feed (biomass) that animals and humans eat, thus contain exergy. Exergy is also embodied in all the materials and fossil fuels we use, as they are created from biological and geological processes, which in turn are driven by solar energy. This connection between matter and energy was already brought forward by Isaac Newton as he wrote [Newton 1730]: Are not gross Bodies and Light convertible into one another, and may not Bodies receive much of their Activity from the Particles of Light which enter their Composition? Albert Einstein proved this relation in 1905 by his famous formula on mass-energy equivalence (eqn 2.11) in which E is the maximum amount of work obtainable, m is mass and c is the velocity of light. So mass and energy are ”the same thing” [Wall and Gong 2001]. E = mc2

(2.11)

Unlike energy and mass, exergy is usually not conserved. Only in an ideal, reversible process the exergy of the resources necessary for the process is completely converted to products. Most real processes are non-ideal and irreversible. The exergy put into the process in the form of energy and mass is only partially recovered in the product(s) and waste(s). The remainder of the incoming exergy is destructed and ”converted” in entropy. This feature makes exergy useful for the evaluation of processes and systems because it can be used to identify where losses and exergy destruction occur. The word exergy was introduced by Z. Rant in 1954 to replace ”technical available work” [Szargut 2005]. Exergy is based on the 1st and 2nd Law of Thermodynamics (appendix A.2.1. It is defined as [Rosen and Dincer 1997]:

30

CHAPTER 2. THEORETICAL BACKGROUND The maximum amount of work that can be obtained from a stream of matter, heat or work, as it comes to equilibrium with a reference environment, and it is a measure of the potential of a stream to cause change, as a consequence of not being completely stable relative to the reference environment.

An easier to understand description of the exergy content of materials is given by Gong and Wall [2001]. They regard the exergy content of a natural resource as a general measure of its potential usefulness. For example MgCl2 used for magnesium production has an exergy content of 1661.76 kJ/kg, while the more useful magnesium metal has an exergy content of 25765 kJ/kg. Sometimes this usefulness can lead to harmful consequences. Thus exergy embodied in wastes would be one measure of its potential for causing harm to the environment.

2.2.2

Calculation

An exergy balance of a process or system can be written as [Wall and Gong 2001]: Ein = Eout + Edestr. Eraw

materials

=⇒

+ Eheat = Eusable

(2.12)

products

+ Eunusable

products

+ Edestr.

The amount of destructed exergy can be calculated using the Gouy-Stodola Law: Edestr. = T0 · Σ∆S

[J]

(2.13)

When the exergy content of all streams entering and leaving the system is known, the amount of destructed exergy can be determined without using the Gouy-Stodola Law. The exergy content of a stream of heat is equal to: Eheat = Q ·

(T − T0 ) T

[J]

(2.14)

Here Q represents the enthalpy associated with heating the raw materials to temperature T. No exergy flow is associated with heat rejected to the environment [Szargut 2005], for example during the cooling of ingots, hence it is part of Edestr. . The exergy content E of a stream of matter at reference environment conditions is equal to its chemical exergy content [Szargut 2005]: X X Estream = nphase · ech, phase + nsol · ech, sol [J] (2.15) i

i

Where: ech,phase =

X

xi · e0i

[J/mol]

(2.16)

i

for components with ai = 1. For liquid or solid solutions with interaction between the components so that ai 6= 1: X ech, sol = xi · (e0i + RT0 ln ai ) [J/mol] (2.17) i


31

Depending on the thermodynamic data available for the investigated system ech, sol can be calculated using the activity of the components, or ideal behaviour can be assumed and ai is replaced by xi . The difference between ideal behaviour (Raoult’s Law) and non-ideal behaviour is illustrated with the magnesium-tin system. In a liquid magnesium-tin alloy (solution), the activity of magnesium strongly deviates from ideal Raoult behaviour as shown in figure 2.1 on page 32. aLMg (on the vertical axis) is smaller than xLMg (on the horizontal axis) , even when only a very small amount of tin is present. On the contrary the activity of tin is hardly influenced by a small amount of magnesium in the alloy. At 90 at.% Sn its activity is equal to 0.9, so it still follows Raoult’s Law. Deviations from ideal behaviour like in the Mg-Sn system will affect the outcome of equation 2.17. When a solution consisting of 90 at.% Mg and 10 at.% Sn is considered ideal then: ech,

sol

= xSn · (e0Sn +RT0 ln xSn ) + xMg · (e0Mg + RT0 ln xMg ) = 0.1 · (551.9 · 103 + 8.314 · 298.15 · ln(0.1)) + 0.9 · (626.1 · 103 + 8.314 · 298.15 · ln(0.9)) = 617874.18 J/mol

In reality the solution is non-ideal so the activities of tin and magnesium from figure 2.1 have to be used when calculating ech, sol . These are much lower than the mole fraction. The exergy content of the non-ideal solution is calculated as: ech,

sol

= xSn · (e0Sn +RT0 ln aSn ) + xMg · (e0Mg + RT0 ln aMg ) = 0.1 · (551.9 · 103 + 8.314 · 298.15 · ln(0.01)) + 0.9 · (626.1 · 103 + 8.314 · 298.15 · ln(0.8)) = 617040.64 J/mol

The difference between the two calculations is 833.54 J. The non-ideal solution has a lower exergy content because the entropy of the solution is higher. Activities of the components in solution are determined experimentally and can be found in reference works. Only if they are not available ideal behaviour of the solution is assumed. In order to calculate the exergy of a material or stream a reference environment has to be selected. It defines the pressure, temperature and composition (chemical potential) of the surroundings in which the stream is located. The reference environment determines the standard specific exergy content e0 of a component or element. The natural environment of the Earth is not suitable as reference environment. Earth is not in an equilibrium state as temperature and pressure vary with time and location. Instead of Earth a general applicable reference model is used, in which the behaviour of the natural environment is balanced with the theoretical requirements [Rosen and Dincer 1997]. For comparison of processes or assessment of environmental impact, general validity of the reference model is a must. Several reference models have been proposed. Most commonly used is the reference substance model developed by Jan Szargut, which will be used throughout this research as well. It is based on the temperature (298K), pressure (1 atm) and common components

32


Figure 2.1: Thermodynamic activities for liquid Mg-Sn alloys at 1073 K [Predel 1997]. The activity of magnesium (aLMg ) and tin (aLSn ) strongly deviate from Raoult’s Law, which states that the activity of a component is equal to its mole fraction in the alloy.


33

of the earth. The specific exergy e0 of all elements and a selection of components in the reference state is found in Szargut [2005]. The specific exergy of other components can be calculated using equation A.45 in appendix A.2. More information about the reference environment can be found there too. An exergy analysis consists of several steps: Selection of system to investigate, definition of the system and its boundaries and identification of all mass and energy flows entering and leaving the system. Collection of data: chemical composition, mass, temperature and pressure of all material flows and information about the energy flows entering and leaving the system, using a Mass Flow Analysis or energy balance. Selection of the reference environment used in the exergy calculation. Calculation of the exergy of each flow using equations 2.14 - 2.17. Determination of Edestr. using equation 2.12. Evaluation of the system and optimisation where necessary or possible.

2.2.3

Indicator of sustainability

The exergy of a material represents its potential for change. Depending on the type of material this potential is regarded as positive or negative. Useful materials have a positive potential for change. They are more concentrated, more structured, low entropy ¨ materials, which have a higher exergy than their surroundings [Finnveden and Ostlund 1997]. Ores are more valuable than bedrock, while metals are even more valuable and useful than ores. Exergy reflects this, so it an be used as a measure for potential usefulness [Gong and Wall 2001]. For example MgCl2 used for magnesium production has a much lower exergy than magnesium metal (table E.30). The exergy difference between the two indicates the minimum amount of work necessary to convert magnesium chloride into magnesium. The work has to be supplied in the form of energy and additional materials or reagents, leading to primary resource depletion [Rosen and Dincer 1997]. The exergy difference also indicates the loss of usefulness or value when magnesium reacts back to its chloride or oxide form during its life cycle. Not-usable streams, such as off gasses, cooling water, and waste products, have a negative potential for change. They can cause undesired and uncontrolled reactions when released into the environment. The exergy of the unusable streams (Eunusable ) can be a measure of its potential to cause harm [Gong and Wall 2001]. The term exergy losses is often used for the sum of Edestr. and Eunusable (eqn. 2.18).  internal: equipment loss + structural loss = E destr. exergy losses = (2.18) external: exergy in waste = Eunusable Internal losses are a measure of the irreversibility of the process. A part of the losses originate from technical imperfections in the equipment used for the process. The structural loss cannot be reduced unless fundamental changes in the principle of the process are made. To compensate the internal losses more exergy has to be put in the process, which

34


may have an environmental impact, so Edestr. indirectly affects the environment [Wall and Gong 2001]. External losses are related to technical progress and process efficiency. The losses can be regarded as the consequence of the absence of utilization of the by-products from the process [Szargut 2005, p. 3]. Reduction of Eunusable can be done by improving the process efficiency and by finding applications for the unusable materials so they become products. Primary resource depletion is linked to the input side of the exergy analysis. Primary raw materials and fossil fuels entering the system/process lead to depletion of the reserves. Depletion is often expressed in mass or energy units, although it does not involve mass or energy. Both cannot be destroyed or depleted, only transformed, as stated by the first law of thermodynamics (eqn. A.25). Instead the usefulness of the resource is consumed [Gong and Wall 2001], as the metal disperses in products and fuel is combusted. Therefore exergy could be used as a metric for depletion of material and energy resources. Resource degradation or quality loss can be quantified using exergy. Quality loss often involves changes in the chemical composition of the input and output material which leads to a change in specific chemical exergy. The difference in ech can quantify the quality loss. This is a form of exergy destruction, which is related to the creation of entropy (more chaos) by the Gouy-Stola Law (eqn. 2.13). Resource depletion and resource degradation are closely related. Quality losses can be counteracted by refining or dilution, which requires exergy input in the form of resources. If these are primary/natural resources, resource depletion is indirectly an environmental impact of resource degradation. Using recycled metals or scrap instead appears to avoid depletion. However, it can be regarded as depletion of the total stock of high quality metal available in the metal cycle. Although this is not an direct impact on the natural environment, it is still unsustainable as primary productionthen remains necessary to produce ’new’ high-quality metal. It can be concluded that the ’sustainability’ of a process or system can be assessed using exergy as an indicator/metric. It quantifies how efficiently material and energy flows are utilised while taking into account the quantity and quality of these flows. The ’cost’ of quality loss and waste creation is reflected by internal and external exergy losses. Resource depletion is accounted for by the exergy present in the additional primary resources used during the process. As exergy is based on thermodynamics, the effect of changes in production and manufacturing processes on the complete system can be easily assessed and evaluated, allowing for analysis from a life cycle perspective.

2.2.4

Application of exergy analysis to magnesium recycling

The sustainability of the magnesium remelting process is related to the quality of the produced metal and the creation of residues during remelting. Going into more detail, its sustainability is determined by the loss of magnesium metal due to oxidation, entrapment and intermetallic formation, the composition of the obtained metal after remelting, the refining and alloying materials necessary to meet alloy specifications and the amount of salt slag created. Exergy and exergy analysis are employed to quantify these process characteristics in a single unit, Joules, enabling the comparison of remelting of different


35

types of coated semi-products. For this purpose three types of exergy loss are defined, which are directly linked to the phenomena observed during and after remelting: entrapment and oxidation losses (∆EeNo ), quality loss (∆Equality ) and material resource loss (Eresource ), which make up the remelting thermodynamics parameter of the metric (figure 1.5) for evaluation of the recyclability of coated magnesium. The overall efficiency of the remelting process is expressed by the exergy efficiency, while primary resource intensity indicates how efficiently primary resources are used within the process. The salt slag amount and composition is represented by Eunusable , which results from the exergy balance of the process.

Exergy efficiency ψ (eqn. 2.19) quantifies how much of the exergy put into the system is leaving the system as usable products [Ignatenko et al. 2007]. The exergy efficiency will never equal unity for real processes. Invariably internal exergy losses occur and some exergy is destructed. Maximum efficiency is obtained when only usable materials and energy leave the system, so the external exergy losses are zero, and when the internal losses are minimized. In this manner the exergy efficiency reflects the technological and environmental performance of the system [Ignatenko et al. 2007]. ψ =

Eusable Ein

0 0

and

W < U + P0 V − T0 S −

(A.38) X

µi0 ni

(A.39)

It can be concluded that the amount of work obtained during an irreversible process is P smaller than the work obtained of a reversible process. Hence U + P0 V − T0 S − µi0 ni is the maximum amount of work that can be delivered. Exergy (E) is defined as the maximum work obtained to return a material from its specified state to materials common in the environment in a reversible way, while heat is exchanged

A.2. EXERGY only with the environment [Szargut 1989], it follows that: X E = U + P0 V − T0 S − µi0 ni

185

(A.40)

The difference between E, the maximum useful work obtained and W, the actual useful work is the amount of exergy destructed (Edestr. ). It represents the irreversibilities in the system. Equation 2.13 is called the Gouy-Stodola Law. X Edestr. = T0 · ∆S [J] (A.41)

A.2.2

Reference environment

In order to calculate exergy using equation A.40 the pressure, temperature and composition (chemical potential) of the surroundings have to be defined. This reference environment also has to act like a thermal reservoir, as this is one of the requirements used for the derivation of exergy in the previous section. The following requirements have to be met by the reference environment [Rosen and Dincer 1997]: It has a stable equilibrium state. Its exergy is zero. It is non-reactive so no reactions between its chemical components take place. It is an infinite system. It is a sink and source of heat and materials. The temperature, pressure and chemical potential are constant for each of the chemical components. The natural environment of the Earth does not meet these requirements. It is not at equilibrium and temperature and pressure vary with time and location. Instead of Earth a general applicable reference model is used, in which the behaviour of the natural environment is balanced with the theoretical requirements [Rosen and Dincer 1997]. For comparison of processes or assesment of environmental impact, general validity of the model is a must. Several reference models have been proposed. Most commonly used is the reference substance model developed by Jan Szargut, which will be used throughout this research as well. The reference temperature (298 K) and pressure (1 atm) are based on the Earth’s average. For every chemical element a reference substance is selected, which is either from the atmosphere, hydrosphere or lithosphere (figure A.2). Reference substances are common substances in the earth’s environment, which represent the products of the interaction between the environment and the chemical element [Szargut 1989]. The concentration of the reference species is based on the average composition of air, sea or crust and is taken as the zero level for the exergy calculation. Each chemical element has a standard exergy e0ch in kJ/mol, which can be found in Szargut [2005].

A.2.3

Exergy types

The total exergy of a system or stream can be further divided into four types: chemical, physical, kinetic and potential exergy.

186

APPENDIX A. THEORETICAL BACKGROUND

Figure A.2: Origin of reference substances used to calculate exergy of the elements [Rivero and Garfias 2006].

Chemical exergy is the difference between the chemical composition of the stream under consideration and the reference environment. Dolomite (CaMgCO3 ), for instance, contains exergy because it has a higher proportion of magnesium and calcium per kg than the Earth’s crust. Each material stream can contain a phase, which is a mixture of i components where the component’s activity ai = 1, and a solution, with interaction between the components so that ai 6= 1. The chemical exergy (Ech ) of a stream is written as (eqn. A.42): Ech =

X

nphase · ech,

i

phase

+

X

nsolution · ech,

solution

(A.42)

i

Where: ech,

phase

=

X

xi · e0i

(A.43)

i

for phases. For solutions: ech,

solution

=

X

xi · (e0i + RT0 ln ai )

(A.44)

i

Depending on the thermodynamic data available for the investigated system ech, solution can be calculated using the activity of the components, or ideal behaviour can be assumed and ai is replaced by xi . The standard specific exergy content e0i of a component or element depends on the chosen reference environment. It is found in Szargut [2005] or calculated using equation A.45, inserting e0 of the elements in the component and the Gibbs energy of formation. ∆G0fmn

A.3. THERMODYNAMICS OF REMELTING

187

can be found in reference books or thermodynamic databases. The calculated e0 can be found in table E.30 on page 303. X e0component = ∆G0f mn + nelement i · e0i (A.45) The exergy content of organic substances or resins can be calculated using the group contribution method (eqn. A.46), which is based on the contribution of each hydrocarbon group i to the exergy of the organic molecule j [Szargut 2005]. X X ej = ni · ei and Eresin = mresin xj · ej (A.46) i

j

where ni is the number of times hydrocarbon group i occurs in the molecule j and ei is the specific exergy of group i. The value of ei of each group is found in Szargut [2005]. Physical exergy reflects the deviations of pressure and temperature of a stream with respect to the reference environment. The contribution of a difference in temperature is calculated using [Szargut 2005]: eph,

T

T = (∆hP0 − T0 ∆sP0 )

(A.47)

T0

The contribution of pressure differences to the physical exergy only has to be calculated when the pressure of the stream differs from the pressure in the reference environment. The pressure in the magnesium recycling system is atmospheric so the physical exergy due to pressure differences can be omitted. The exergy of a stream of heat is equal to: Eheat = Q ·

(T − T0 ) T

(A.48)

Here Q represents the enthalpy associated with heating the raw materials from the reference temperature T0 to temperature T. No exergy flow is associated with heat rejected to the environment [Szargut 2005]. Hence the heat released during the cooling of products is accounted for in Edestr. . Kinetic and potential exergy are related to motion and elevation of the stream or system. Water flowing down a water fall has kinetic and potential exergy. Both are not relevant for the assessment of recycling system of coated magnesium.

A.3

Thermodynamics of remelting

Generally mixtures can be classified as ideal or non-ideal solutions. In an ideal solution there is no interaction between the components, hence the activity a is equal to the mole fraction x. In non-ideal solutions components do interact. The deviation from the ideal

188


solution behaviour can be expressed using excess functions. A non-ideal solution has a tendency to separate when a/x is larger than 1 and the enthalpy of mixing is positive. Tendency to compound formation in the solution occurs when a/x is smaller than 1 and increases with temperature, while the mixing enthalpy is negative [Gaskell 2003]. The mixing energy of a non-ideal solution is given by: ideal Gmix = G0ref + Gxs mix + Gmix

(A.49)

The second term, describing the deviation from the ideal mixture is often complex. Thermodynamic modeling is done using the Calphad method [Saunders and Miodownik 1998]. Phenomena in multicomponent systems can be well predicted using accurate thermodynamic characterization of the experimentally determined binary and ternary systems included in the multicomponent system. The prediction is obtained by minimizing the Gibbs energy of the total system, including the mixing energy (eqn. A.49). Different models are used for thermodynamic characterization. Random substitution model: used for liquids and solids without ordering, often these are metals. Sublattice model: used to describe solution and compound phases such as interstitial phases (eg Laves phase) and intermetallic phases. Modified Quasi-Chemical Model. This model enables modeling of complex Gibbs energy variations with composition such as occur in liquid salt mixtures. The liquid orders around specific compositions associated with chemical or physical phenomena [Pelton et al. 2000; Pelton and Chartrand 2001].

The Modified Quasi-Chemical Model is the model used for modeling liquid salt mixtures. The relevant mixtures and the model will be discussed briefly. The atoms or molecules of a binary solution of A and B, are distributed over sites on a quasi-lattice. Interaction between A and B takes place by pair formation according to: A − A + B − B = 2A − B

(A.50)

in which A-B is called a first neighbour pair and this type of ordering is called first nearest neighbour ordering. The degree of ordering depends on the composition of the solution. The ordering in the solutions can cause large deviations from the ideal solution behaviour. Liquid salt mixtures containing Na, K, Ca, Mg and Ba chloride are described and modeled by Chartrand and Pelton [2001d] and Chartrand and Pelton [2000]. Molten salts with MgCl2 have extensive short range ordering, which is treated by introducing the concept of second nearest neighbour ordering. It is represented by equation A.51, where A and B can be Ba, Ca, K, Mg or Na. (A − Cl − A) + (B − Cl − B) = 2(A − Cl − B)

(A.51)

The amount of experimental data for the Na,K,Mg//Cl system is large, hence the reliability of the model is high. This model describes the fluoride-free salt flux system as used during

A.3. THERMODYNAMICS OF REMELTING

189

remelting of magnesium. The presence of BaCl2 in the mixture also causes short range ordering, but in a lesser extend than for MgCl2 . Less experimental data is available for the barium chloride containing systems. Mixing enthalpies for binary systems had to be estimated and no experimental data was available for quaternary systems with barium chloride [Chartrand and Pelton 2000]. Fluoride salt mixtures containing MgF2 exhibit extensive short range ordering. The fluoride systems such as Na, K, Mg, Ca//F are difficult to verify experimentally [Chartrand and Pelton 2001d]. The limited availability of experimental data means more interpolation and assumptions in the models. The salt flux with CaF2 addition is represented by the Na, K, Mg, Ca//Cl, F system. In this system the deviation from ideal behaviour is very large. Besides second nearest neighbour ordering, (eqn. A.51) there is also first nearest neighbour ordering: A − F + B − Cl = A − Cl + B − F

(A.52)

where A and B can be Ca, K, Mg, Na. The model results for multicomponent liquids, such as salt flux with CaF2 , are based on the binary and ternary systems. Experimental data obtained for this system is limited to liquidus temperatures only. In addition the accuracy of the data is low, and many compositions are determined by interpolation. Hence the error can be considerable. These are thus true predictions [Chartrand and Pelton 2001b]. Although the available models are a useful tool for predicting what can happen during remelting of (coated) magnesium with salt flux, experimental verification of these complex systems is necessary.

190


Appendix B

Characteristics of coated semi-products B.1

Determination of coating mass and density

Object 1 and 3: The coating density can be calculated/estimated based on the thickness of each layer determined from the microscope images (figures B.1 - B.3) and the density of the coating component found in Lide [2006]. Object 2: The coating density is calculated based on the thickness of the coating layer from the EPMA images, namely 10 µm. It is assumed the coating consists only of Mg3 (PO4 )2 and AlPO4 , and the composition is constant throughout the layer. The amounts of aluminium and magnesium in the coating are given by the spot analysis of the phosphor-rich layer in table B.4. Object A - D: Pieces were placed in containers with TetraHydroFuran (THF) to remove the coating. The THF with solids and coating in it was filtered. Liquid and solids were separated by filtration using Whatman filter 542 that retains particles larger than 3 µm [Whatman 2007]. Subsequently the filter with solids and some organic material was ashed in a chamber furnace at 600 . Afterwards the amount of coating, organic materials (resin and organic pigments) and inorganic pigment and filler particles per cm2 scrap was calculated (table B.1). The ratio between pigment and resin varies between 0.2 and 0.4 and is different for each coating. Using the coating thickness determined from the EPMA images the density of the organic coating can be calculated. The densities obtained are lower than the specific gravity of standard epoxy (1.15) and polyester (1.12 - 1.46) [Shackelford 2001]. The numbers are closest to the specific gravity of polyethylene, which is unlikely. In the calculation it has been assumed that the coating thickness is independent of the location on the semiproduct. This assumption can cause variations in the coating density as well. Blanchard et al. [2005] reports that the difference in thickness between a flat piece and a curved piece can be a factor two. The density of the coating on object D is very high for an organic coating. It is more in the order of magnitude of a metallic coating. In the calculation the coating thickness based

192

APPENDIX B. CHARACTERISTICS OF COATED SEMI-PRODUCTS

on the EPMA image (30 µm) is used. Replacing the coating thickness in the calculation with a value based on the optical microscope images (table B.5) gives a more reasonable value. Using 175 µm gives a density of 1139 mg/cm3 . This value is more in line with the specific gravities found in Shackelford [2001] and indicate the pieces de-coated with THF came from a curved part of object D.

Table B.1: Coating mass and density of object A - D determined by removing the coating with THF.

Object Object Object Object

A B C D

Coating mass mg/cm2

Pigment mass mg/cm2

Resin mass mg/cm2

Coating density mg/cm3

8.7 ± 0.6 85.7 7.4 ± 0.05 19.9 ± 5.6

2.4 ± 0.5 22.0 1.3 ± 0.02 5.3 ± 1.2

6.3 ± 1.0 63.7 6.1 ± 0.1 14.7 ± 4.4

866.8 ± 55.2 927.0 819.4 ± 5.2 1139

Table B.2: Calculated coating mass and density of object 1 - 3. Density from Lide [2006]; Shackelford [2001]. Layer thickness µm

Density g/cm3

Mass mg/cm2

3 15 15 2 35

7.14 8.96 8.90 7.15 8.67

2.14 13.4 13.4 1.43 30.36 1.98 0.23 2.21 ??

Object 1

Zinc Copper Nickel Chromium Total

Object 2

Mg3 (PO4 )2 AlPO4 Total

10

2.17 2.56 2.21

MgO.Al2 O3

12 - 20

3.58

Object 3

B.2. COMPOSITION INORGANIC COMPONENTS

B.2

193

Composition inorganic components

Table B.3: XRF analysis of residue on surface after de-coating. Object A Wt.% Std. Error Na2 O 0.174 MgO 1.76 Al2 O3 1.93 SiO2 2.61 P2 O5 0.368 SO3 0.252 Cl 0.0155 K2 O 0.0145 CaO 40.36 TiO2 43.25 V2 O5 0.159 Cr2 O3 0.110 MnO 0.0416 Fe2 O3 3.60 Co3 O4 0.125 NiO 0.0059 CuO 0.0080 ZnO 0.0644 Br -.SrO 0.0096 Nb2 O5 -.SnO2 0.0411 BaO 0.100 F 5.00 ZrO2 -.PbO -.-

0.019 0.15 0.15 0.18 0.004 0.028 0.0017 0.0016 0.54 0.55 0.023 0.012 0.0028 0.21 0.014 0.0016 0.0016 0.0032 -.0.0012 -.0.0058 0.036 0.12 -.-.-

LOI

-.-

-.-

Object B wt.% Std. Error 0.273 0.426 11.98 4.68 0.188 2.19 0.0137 -.0.0577 72.74 0.268 0.141 0.0068 0.687 -.-.0.0115 0.0212 0.0145 0.0809 0.0050 0.115 6.05 -.-.-.-.-

0.030 0.047 0.36 0.23 0.003 0.16 0.0015 -.0.0062 0.49 0.039 0.016 0.0024 0.076 -.-.0.0016 0.0015 0.0007 0.0090 0.0011 0.006 0.13 -.-.-.-.-

Object C Wt.% Std. Error 0.147 2.18 3.01 0.775 0.195 0.157 0.0210 0.0070 24.83 63.70 0.217 0.335 0.0353 4.23 -.0.0063 0.0177 0.0489 -.-.0.0056 0.0390 . 1mm Metal in dross = metaldross · 100% metal > 1mm Metal in sludge = · 100% sludge

Yield =

Appendix H

Remelting experiments

Table H.1: Materials used for experiments.

MgCl2 NaCl KCl CaF2 Fe2 O3 SnO2 Cr2 O3 CaO TiO2 SiO2 Al2 O3 BaSO4

Supplier

Name

Art.No.

Alfa Aesar GmbH Merck KGaA Mallinckrodt Baker B.V. Acros Organics Merck KGaA BDH Chemicals Ltd Riedel-de Haen AG Mallinckrodt Baker B.V.

Magnesium chloride, anhydrous Sodium chloride Potassium chloride Calcium fluoride, anhydrous Iron(III)-oxide Tin(IV)-oxide Chrome (III)-oxide Calcium oxide

Sigma-Aldrich Chemie GmbH Sigma-Aldrich Chemie GmbH Sigma-Aldrich Chemie GmbH

Quartz Aluminium Oxide Barium sulfate

012315 1.06404.1000 0209 21.825.1000 3924 103-506 312233 0078 CZ0701 83340 342718 243353

Table H.2: Particle size in µm of oxides used for oxide-flux interaction experiments.

D1 D5 D10 D25 D50 D75 D90

Al2 O3

BaSO4

Cr2 O3

Fe2 O3

SiO2

SnO2

TiO2

30.84 53.88 60.20 72.37 87.45 104.85 124.18

0.50 0.66 0.79 1.09 1.60 2.37 3.61

1.11 2.75 3.90 6.28 9.97 15.15 21.48

0.20 0.67 1.31 4.43 17.09 36.61 54.29

0.74 1.45 2.41 7.39 22.55 44.65 68.01

0.21 0.28 0.34 0.49 0.77 1.22 1.83

0.22 0.28 0.33 0.46 0.69 1.01 1.40

326

APPENDIX H. REMELTING EXPERIMENTS

Table H.3: Masses for metal-oxide experiments.

TiO2 SiO2 Cr2 O3 Fe2 O3 BaSO4 SnO2 CaO Al2 O3 SnO2 Al2 O3 Fe2 O3 TiO2 Cr2 O3 SiO2 BaSO4 CaO

In Oxide (g)

Metal (g)

Out Oxide (g)

Metal (g)

0.0603 0.0367 0.0926 0.0823 0.0835 0.1089 0.0615 0.0720 0.3593 0.2494 0.3593 0.4536 0.3422 0.3411 0.8306 0.3284

2.3037 2.2784 2.3854 2.4406 2.7521 2.4546 2.7822 2.8035 2.6030 2.6591 2.4347 3.1747 2.4420 3.0819 3.0163 3.0765

0.0744 0.1067 0.1242 0 0.1644 0.0855 0.0058 0.6630 0.2887 0.2671 0.3976 0.5016 0.7286 0.8606 0.4152

2.2500 2.3854 2.4153 2.7672* 2.4141 2.7699 2.8717 2.3032 2.5953 2.4779 3.2817 2.2871 2.6832 2.9327** 2.9641***

* Mass loss 0.0473 g. ** Mass loss 0.2349 g. *** Mass loss 0.045 g.

Table H.4: Masses for metal - flux - oxide experiments. Cr2 O3

BaSO4

SnO2

Cr2 O3

BaSO4

SnO2

Oxide Mg Mg9Al Flux Temperature

(g) (g) (g) (g) ( )

2.124 13.020 7.498 720

5.99 15.658 10.377 720

2.642 11.511 6.504 720

3.477 25.045 2.883 700

3.069 23.801 6.559 700

5.154 19.530 2.486 700

Metal Slag

(g) (g)

13.087 9.59

16.544 14.909

11.254 9.319

25.764 5.588

24.28 8.54

22.527 4.789

327

Table H.5: Remelting of objects.

Coated Object 1 Object 2 Object 3 Object A Object B Object C Object D De-coated Object A Object B Object C Object D

In Scrap (g)

Flux (g)

Out Metal (g)

Slag (g)

18.521 30.135 14.194 15.499 10.563 9.470 13.29

17.935 30.135 22.949 23.364 15.616 14.99 19.54

27.74 36.50 7.51 19.64 16.08 10.63 14.40

6.18 23.60 28.62 9.93 8.41 5.31 15.09

7.578 18.89 17.52 14.05

21.509 31.60 26.992 21.00

16.47 17.89 24.24 15.45

14.28 ? 16.16 13.70

328

APPENDIX H. REMELTING EXPERIMENTS

Appendix I

Evaluation of remelting process

Table I.1: R of simulation of remelting magnesium with a single coating component (chapter 6.4 page 103) . Target alloy composition is 9998A magnesium, reached by dilution with primary magnesium metal.

Uncoated Nickel Copper CaO Cr2 O3 MgSnO3 SiO2 TiO2 BaSO4

R 0.1 µm layer MJ

R 1 µm layer MJ

R 5 µm layer MJ

R 10 µm layer MJ

5584.18 276316.95 276615.79 5456.34 5551.88 5565.97 5491.5 37564.13 5482.02

2783092 2783935 6511.51 5555.42 75605.02 62412.43 40822.31 6211.63

13727495 13940629 11787.49 5571.04 338506.74 329460.77 40563.09 9282.63

27879537 27958086 28324.29 5590.59 775724.25 642675.76 40238.89 13553.15

Table I.2: R of remelting Mg9Al with coating of objects 1 and 2, and de-coating residue from objects A - D (described in chapter 6.5 page 114). Impurity limits are from 9998A magnesium or AZ91D alloy and can be reached by dilution with virgin metal or by refining with MgCl2 and virgin metal.

Object Object Object Object Object Object

1 2 A B C D

R Dilution to 9998A limits MJ

R Dilution to AZ91D limits MJ

R Refining to 9998 limits MJ

5481863.9 5338.83 8485.88 75727.62 6428.79 30310.08

1053771.9 1557.61 3265.88 37548.62 2224.79 14163.08

– 191.03 234.97 868.87 223.72 24131.08

330

APPENDIX I. EVALUATION OF REMELTING PROCESS

Appendix J

Metric for recycling

Table J.1: Recyclablity parameters describing the thermal de-coating of magnesium semiproducts. ln A is the average ln A of all experiments done at the selected β. Argon

Air

Object

Peak

β

/min

E/(RTp ) -

A A A B B B C C C C C C D D D D D D D D D

1 1 1 1 1 1 1 1 1 2 2 2 1 1 1 2 2 2 3 3 3

5 10 20 5 10 20 5 10 20 5 10 20 5 10 20 5 10 20 5 10 20

35.73 35.15 34.43 34.76 35.07 43.57 43.00 42.24 22.61 21.38 21.55 31.91 31.38 30.63 30.72 30.39 29.88

± ± ± ±

1.23 1.17 2.24

± ± ± ±

2.03 3.22 3.19

± ± ± ± ± ± ± ± ±

4.18 3.92 1.74 1.76 40.26 39.35

ln A 1/min

E/(RTp ) -

ln A 1/min

34.36 ± 0.02 34.43 ± 34.43 ± 0.01 35.21 ± 35.13 ± 0.03 43.74 ± 43.40 ± 43.18 ± 0.08 19.63 ± 0.15 19.69 ± 19.92 ± 0.14 32.29 ± 0.64 31.72 ± 31.44 ± 0.09 46.15 ± 0.34 46.52 ± 46.5 ± -

28.43 ± 2.43 27.69 ± 2.29 27.21 ± 2.27 30.47 ± 1.70 29.78 ± 1.58 29.17 ± 1.61 22.30 ± 2.02 21.71 ± 1.99 52.48 ± 4.61 34.82 ± 7.15 34.34 ± 7.05 34.22 ± 6.98 27.91 ± 6.13 27.2 ± 6.16 27.03 ± 5.83 28.07 ± 4.52 28.26 ± 5.01 30.17 ± 5.33 -

27.16 27.21 27.11 29.08 29.09 29.28 20.13 19.92 19.98 35.34 34.21 34.41 25.54 25.47 26.66 27.72 28.02 26.66 -

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

0.05 0.05 0.02 0.06 0.15 0.06 0.20 0.11 1.28 0.06 0.57 0.23 0.08 1.80 0.30 0.06 1.51

332

APPENDIX J. METRIC FOR RECYCLING

Summary Coated magnesium - designed for sustainability? Design of consumer products, such as cars, electronics etc. requires the selection and combination of various materials. One of these materials is magnesium. to improve its properties and appearance it can be coated. After manufacturing and use the products reach the End-of-Life phase, in which the products are collected and separated in the different materials. The metals are recycled so they are suitable for making new products. A sustainable closed life cycle as described above, in which material quality is preserved and waste creation and resource depletion are minimized, can be attained by selecting the appropriate material combinations during design of the product. Sadly the life cycle for coated magnesium is far from closed. At the moment it stops at the End-of-Life phase. Coated magnesium is separated from other materials, but not further recycled by remelting it into new magnesium. On the other hand, uncoated magnesium is recycled. To recycle magnesium in a sustainable manner the impact of the coating (chosen during design) on the recycling process has to be known. For this detailed quantitative understanding of the parts of the recycling system as well as their interconnectivity is a pre-requisite. The current study evaluates the recyclability of coated magnesium from three perspectives: de-coating kinetics, remelting thermodynamics and remelting kinetics. These are linked to product design. In this way a connection between product and process design and recyclability / ’sustainability’ is established. A good recyclability of coated magnesium means: 1. Its organic coating can be completely removed by thermal de-coating. 2. Its inorganic coating (components) have little impact on the remelting process: little metal losses due to oxidation, entrapment and intermetallic formation, little change of the alloy composition and quality, little usage of primary material resources to compensate quality changes.

The first criterion relates to de-coating kinetics, while the second relates to remelting thermodynamics and kinetics. Necessary information has been obtained by characterizing seven different coated magnesium semi-products, subjecting them to a thermal de-coating process and recording

334

SUMMARY

the mass and temperature change during de-coating. Using kinetic theory two parameters have been selected to describe de-coating. One describes the speed or velocity of de-coating (ln A). The other combines the threshold to overcome before de-coating starts with the temperature at which maximum mass loss occurs in the form of a ratio (Ea /(RTp )). The remelting process has been simulated using a thermodynamic modeling package to obtain the composition of metal and slag after remelting magnesium with inorganic coating (components). In these simulations an infinite reaction time has been used. Exergy has been used to quantify the impact on remelting as it can express both mass and quality changes in a single unit (Joules). The parameter R introduced in this thesis quantifies the impact as the sum of the entrapment and oxidation losses (∆EeNo ), the quality losses (∆Equality ), and the material resource loss (∆Eresource ) based on exergy. Replacement of the infinite reaction time with a shorter reaction time will reveal the influence of kinetics on the remelting process. Combination of R with the remelting kinetics will give the actual impact of the coating on the remelting process. Combining the three parameters in a 3-dimensional metric visualizes the recyclability of the current design and the desired recyclability, indicated by the positions in the metric. In this way necessary improvements are easily visualized and explained. The metric and its parameters provide a detailed, fundamental understanding of the processes in the recycling system. Furthermore they reflect the life cycle approach in several ways. First, the combination of coating plus magnesium as present in a product is considered. Second, the entire recycling process is taken into account. Last, the recycling and recyclability of the product are linked to its design. During the product design phase this metric for recycling can be used as a tool to explain the effect of design choices on recyclability, contributing to Design for Recycling (DfR). The current study addresses a very specific material (magnesium), connection method (coating) and unit operations (thermal de-coating, remelting) at a very detailed level. The resulting metric and its parameters can be used in DfR models, in particular in the Recycling Optimization model of the DfR models developed by Reuter and van Schaik (2008). In the models optimal recycling solutions for products are derived as function of the materials in them and optimized based on economics, recycling rate and exergy, among others. Utilisation of the metric for recycling and the DfR models will bring a sustainable, closed magnesium life cycle closer so in future people can say: coated magnesium - designed for ’sustainability’ !

Christina E.M. Meskers

Samenvatting Coated magnesium - designed for sustainability? Tijdens het ontwerp van consumentenproducten zoals auto’s, electronica etc. worden verschillende materialen geselecteerd en gecombineerd. Een van deze materialen is magnesium. Om de eigenschappen en het uiterlijk van magnesium te verbeteren kan een coating worden aangebracht. Na fabricage en gebruik bereiken de producten de Endof-Life fase, waarin de producten worden verzameld en gescheiden in de verschillende materialen. De metalen worden gerecycleerd zodat ze weer geschikt zijn voor het maken van nieuwe producten. Een duurzame gesloten levenscyclus zoals hierboven beschreven, waarin de kwaliteit van materialen wordt behouden en de vorming van afval en het uitputten van grondstoffen wordt geminimaliseerd, kan gerealiseerd worden door passende materiaalcombinaties te selecteren tijdens het ontwerp van het product. Helaas is de levenscyclus van magnesium met een coating verre van gesloten: het stopt bij de End-of-Life fase. Magnesium met een coating wordt gescheiden van de andere materialen, maar niet verder gerecycleerd door het te hersmelten tot nieuw magnesium. Magnesium zonder coating daarentegen wordt wel gerecycleerd. Om magnesium op een duurzame manier te recycleren is het nodig het effect van de coating (die gekozen is tijdens het ontwerp) op het recycling proces te weten. Hiervoor is een gedetailleerd, kwantitatief begrip van de onderdelen van het recyclagesysteem alsmede van hun onderlinge interactie noodzakelijk. Deze studie evalueert de recycleerbaarheid van magnesium met een coating vanuit drie perspectieven: kinetiek van ontlakken, thermodynamica van hersmelten en kinetiek van hersmelten. Deze worden gekoppeld aan het ontwerp van het product. Op deze manier wordt een verband gelegd tussen producten procesontwerp enerzijds en recycleerbaarheid en duurzaamheid anderzijds. Magnesium met een coating is goed recycleerbaar als: 1. De organische coating volledig verwijderd kan worden d.m.v. thermisch ontlakken. 2. De anorganische coating (componenten) weinig invloed hebben op het hersmeltproces: weinig metaalverliezen door oxidatie, door inclusie in de slak en door vorming van intermetallische verbindingen, weinig verandering van de samenstelling en kwaliteit van de legering, weinig verbruik van primaire grondstoffen om de kwaliteitsverandering te compenseren.

336

SAMENVATTING

De eerste eis heeft betrekking op de kinetiek van het ontlakken, de tweede op de thermodynamica en kinetiek van het hersmelten. Zeven verschillende magnesium halffabrikaten met een coating zijn gekarakteriseerd, ze zijn thermisch ontlakt en de gewichts- en temperatuurverandering tijdens het ontlakken is vast gelegd. Met behulp van kinetiek zijn twee parameters geselecteerd om het ontlakken te beschrijven. Een beschrijft de snelheid van ontlakken (lnA), de andere combineert de drempel die overkomen moet worden voordat het ontlakken begint met de temperatuur waarbij het maximale massaverlies optreedt (Ea /(RTp )). Het hersmeltproces is gesimuleerd met behulp van een thermodynamisch modelleerpakket om de samenstelling van metaal en slak na het hersmelten van magnesium met anorganische coating (componenten) te verkrijgen. Bij deze simulaties is uitgegaan van een oneindig lange reactietijd. Exergy (E) is gebruikt om de invloed op het hersmelten te kwantificeren omdat het massaen kwaliteitsveranderingen in dezelfde eenheid (Joules) uitdrukt. De parameter R die in dit proefschrift wordt ge¨ıntroduceerd kwantificeert de invloed als de som van de oxidatie- en inclusieverliezen (∆EeNo ), de kwaliteitsverliezen (∆Equality ), en het verbruik van primaire grondstoffen (∆Eresource ) op basis van Exergy. Door de oneindig lange reactietijd door een kortere reactietijd te vervangen wordt de invloed van kinetiek op het hersmeltproces bekend. Combinatie van R met de kinetiek van het hersmelten geeft de werkelijke invloed van de coating op het hersmeltproces. De drie parameters zijn samengebracht in een 3-dimensionale metriek. De recycleerbaarheid van het huidige product ontwerp en de gewenste recycleerbaarheid worden weergegeven door de posities in de metriek. Op deze manier kunnen de noodzakelijke verbeteringen gemakkelijk zichtbaar en begrijpbaar gemaakt worden. De metriek en zijn parameters dragen bij aan een gedetailleerd, fundamenteel begrip van de onderdelen in het recyclagesysteem. Verder weerspiegelen ze de holistische benadering van de levenscyclus omdat 1) de combinatie van magnesium met coating zoals toegepast in producten is onderzocht, 2) het gehele recyclageproces is gebruikt en 3) het recycleren en de recycleerbaarheid van het product is verbonden met het ontwerp. Tijdens de productontwerp fase kan deze metriek als hulpmiddel gebruikt worden om het effect van materiaalkeuzes op de recycleerbaarheid uit te leggen, en draagt zo bij aan Ontwerp voor Recyclage (Design for Recycling - DfR). De huidige studie beschouwt een specifiek materiaal (magnesium), verbindingsmethode (coating) en process (thermisch ontlakken, hersmelten) op een zeer gedetailleerd niveau. De ontwikkelde metriek en zijn parameters kunnen toegepast worden in DfR modellen, in het bijzonder in het Recycling Optimization (Recyclage Optimalisatie) model ontwikkeld door Reuter en van Schaik (2008). In de modellen wordt de optimale recyclageroute voor producten bepaald als functie van de materialen in het product, en geoptimaliseerd op basis van onder andere rendabiliteit, recycleerpercentage en Exergy. Gebruik van de metriek voor recyclage en voor de DfR modellen brengt een duurzame gesloten magnesium levenscyclus een stap dichterbij zodat men in de toekomst kan zeggen: ”coated magnesium - designed for sustainability!” Christina E.M. Meskers

Curriculum Vitae Christina Meskers attended secondary school at the ”Katholieke Scholengemeenschap Hoofddorp” in Hoofddorp. In September 1996 she started her university studies in Resources Engineering at the Delft University of Technology, Faculty of Applied Earth Sciences. For her M.Sc. thesis, she carried out research in the field of slag chemistry, investigating the relation between the mineralogy of iron ores used for pellets and slag formation during pellet induration. On 9 March 2004 she received her M.Sc. degree. On February 1st , 2004 Christina started her PhD. research on quantitatively evaluating the ’sustainability’ of coated magnesium used in consumer products. The research was supervised by prof. M.A. Reuter and prof. U. Boin of Delft University of Technology. Prof. R. Boom joined the project in a later stage. Project leader was Dr. Y. Xiao. It is part of SenterNovem IOP Surface Technology’s Magnesium program. A Marie Curie Fellowship of the European Community programme Early Stage Training made a seven month research visit to NTNU, Trondheim (Norway) possible. An additional travel grant of the IOP Surface Technology facilitated a five month research visit to the University of Melbourne, Australia. Furthermore, Christina presented her research at several international conferences and in journal publications. In 2007 Christina received the Young Leader award from the Extraction and Processing division of The Minerals, Metals and Materials Society (TMS).