Mineralogical influence on leaching behaviour of steelmaking slags: a laboratory investigation

DOCTORA L T H E S I S

Department of Chemical Engineering and Geoscience Division of Minerals and Metallurgical Engineering

Mineralogical Influence on Leaching Behaviour of Steelmaking Slags

A Laboratory Investigation

Fredrik Engström

ISSN: 1402-1544 ISBN 978-91-7439-197-8 Luleå University of Technology 2010

Fr edr ik Engström Mineralo gical Influence on Leaching Beha viour of Steelmaking Slags A Labor ator y Inv estigation

ISSN: 1402-1544 ISBN 978-91-7439-XXX-X Se i listan och fyll i siffror där kryssen är

Mineralogical influence on leaching behaviour of steelmaking slags

A Laboratory Investigation

by

Fredrik Engström

Doctoral Thesis

Luleå University of Technology

Department of Chemical Engineering and Geoscience Division of Minerals and Metallurgical Engineering

SE-97187 Luleå

Sweden

Printed by Universitetstryckeriet, Luleå 2010

ISSN: 1402-1544

LIST OF PAPERS

This thesis is based on the following papers:

I. Dirk Durinck, Fredrik Engström, Sander Arnout, Jeroen Heulens, Peter Tom Jones, Bo Björkman, Bart Blanpain and Patrick Wollants:

Hot stage processing of metallurgical slags. Resources, Conservation and Recycling (2008), Vol 52, No 10, p 1121-1131.

II. Mia Tossavainen, Fredrik Engström, Qixing Yang, Nourreddine Menad, Margareta Lidström Larsson and Bo Björkman:

Characteristics of steel slag under different cooling conditions.

Waste Management (2007), Vol 27, No 10, p 1335-1344.

III. Fredrik Engström, Daniel Adolfsson, Qixing Yang, Caisa Samuelsson and Bo Björkman: Crystallization behaviour of some steelmaking slags. Steel Research International (2010), Vol 81, No 5, p 362-371.

IV. Fredrik Engström, Margareta Lidström Larsson, Caisa Samuelsson, Åke Sandström, Ryan Robinson and Bo Björkman: Leaching behaviour of aged steel slags. Submitted (Resources, Conservation and Recycling) 2010.

V. Fredrik Engström, Daniel Adolfsson, Caisa Samuelsson, Åke

Sandström and Bo Björkman: A study of the solubility of pure slag

minerals. Submitted (Minerals Engineering) 2010.

VI. Sina Mostaghel, Fredrik Engström, Caisa Samuelsson and Bo Björkman: Stability of spinels in a high basicity EAF slag.

Proceedings of 6 ^th European Slag Conference, October 20-22, 2010, Madrid, Spain.

F. Engström’s contribution to the papers:

• Practical experiments.

• Responsible for characterization of materials obtained during experimental work using XRD, SEM and leaching.

• Interpretation of leaching data and correlation to mineralogical phases.

• Thermodynamic calculations.

• Responsible for writing Papers II-V and parts of Paper I.

• Supervising and contributing to the discussion of Paper VI.

Related publications not included in this thesis:

VII. Mia Tossavainen, Fredrik Engström, Nourreddine Menad and Qixing Yang: Stability of modified steel slags. Proceedings of 4 ^th European Slag Conference, June 20-21, 2005, Oulu, Finland.

VIII. Qixing Yang, Fredrik Engström, Mia Tossavainen and Daniel Adolfsson: Treatments of AOD slag to enhance recycling and resource conservation. Proceeding of Securing the Future, International Conference on Mining and the Environment, Metals and Energy Recovery, June 27-July 1, 2005, Skellefteå, Sweden.

IX. Qixing Yang, Fredrik Engström, Mia Tossavainen and Mingzhao He:

AOD slag treatment to recover metal and to prevent slag dusting.

Proceeding of the 7 ^th Nordic-Japan Symposium on Science and Technology of Process Metallurgy, Jernkontoret, September 15-16, 2005, Stockholm, Sweden.

X. Qixing Yang, Lotta Nedar, Fredrik Engström and Mingzhao He:

Treatments of AOD slag to produce aggregates for road construction. Proceeding of AISTech 2006, May 1-4, 2006, Cleveland, USA, Vol. 1, p 573-583.

XI. Qixing Yang, Björn Haase, Fredrik Engström and Anita Wedholm:

Stabilization of EAF slag for use as construction material.

Proceedings of REWAS 2008, Global Symposium on Recycling,

Waste Treatment. Minerals, Metals & Materials Society, October 12- 15, 2008, Cancun, Mexico, p 49-54.

XII. Fredrik Engström, Margareta Lidström Larsson, Caisa Samuelsson and Bo Björkman: Ageing investigation of steel slags from electric arc furnace processes. Proceedings of REWAS 2008, Global Symposium on Recycling, Waste Treatment. Minerals, Metals &

Materials Society, October 12-15, 2008, Cancun, Mexico, p 353- 358.

XIII. Qixing Yang, Fredrik Engström, Bo Björkman and Daniel Adolfsson: Modification study of a steel slag to prevent the slag disintegration after metal recovery and to enhance slag utilization.

Proceedings of the VIII international conference on molten slags, fluxes and salts, January 18-21, 2009, Santiago, Chile, p 33-41.

XIV. Fredrik Engström, Caisa Samuelsson and Bo Björkman:

Mineralogical influence of different cooling conditions on leaching behaviour of steelmaking slags. Proceedings of the 1 ^st International Slag Valorisation Symposium, 6-7 April, 2009, Leuven, Belgium, p 67-80.

XV. Charlotte Andersson, Bo Björkman, Fredrik Engström, Sina

Mostaghel and Caisa Samuelsson: The need for fundamental

measurements for a sustainable extraction of metals. Seetharaman

– Seminar 2010.

XVI. Daniel Adolfsson, Fredrik Engström, Ryan Robinson and Bo Björkman: Cementitious phases in ladle slag. Accepted for publication in Steel Research International, November 2010.

XVII. Daniel Adolfsson, Fredrik Engström, Ryan Robinson and Bo Björkman: Hydraulic properties of ladle slag. Submitted to Cement and Concrete Research, September 2010.

XVIII. Daniel Adolfsson, Ryan Robinson, Fredrik Engström and Bo Björkman: Hydraulic properties of mayenite. Submitted to Cement and Concrete Research, September 2010.

XIX. Chandra Sekhar Gahan, Jan-Eric Sundkvist, Fredrik Engström and Åke Sandström: Utilization of steel slags as neutralizing agents in biooxidation of a refractory gold concentrate and their influence on the subsequent cyanidation. Submitted to Resources, Conservation and Recycling, September 2010.

XX. Chandra Sekhar Gahan, Jan-Eric Sundkvist, Fredrik Engström and

Åke Sandström: Comparative assessment of Industrial oxidic by-

products as neutralising agents in biooxidation and their influence

on gold recovery in subsequent cyanidation. Proceedings of the 11 ^th

International Seminar on Mineral Processing Technology MPT-

2010, December 15-17, 2010, Jamshedpur, India.

ABSTRACT

The Swedish steelmaking industry produces large amounts of by-products.

In 2008, the total amount of slag produced reached approximately 1,300,000 metric tons, of which 20% was deposited. Due to its strength, durability and chemistry, steel slag is of interest in the field of construction, since it has similar or better qualities than ordinary ballast stone, which makes it a competitive construction material. However, some steel slags face an array of quality concerns that might hinder their use. These concerns generally involve the following physical and chemical properties:

• Volume expansion

• Disintegration

• Leaching of metals

By controlling and modifying process parameters during slag handling in

liquid state, the physical and chemical properties of steel slags can be

adequately modified to obtain a high-quality product for external

application. The present work was undertaken as a research project within

the Minerals and Metals Recycling Research Centre, MiMeR. The major

goal of this work has been to investigate how different treatment methods

including hot stage processing, cooling rates, ageing time and chemical

composition influence the final properties of the slag. Analysis techniques

used in this investigation include: thermodynamic calculations using

Factsage ^TM , X-ray diffraction analyses (XRD), scanning electron

microscopy (SEM), leaching tests (EN12457-2/3) and thermo-gravimetric

analyses (TG).

The results from this study show that it is possible to control/change the properties of the final product by additions to the liquid slag, thereby changing the chemical composition, as well as by varying the rate and method of cooling. The mineralogical composition, the size of the crystals and the composition of some solid solutions are affected by the cooling rate. The solubility of elements such as chromium and molybdenum varies, probably due to their presence in different minerals. The reactivity of the investigated slag samples increases as the cooling rate increases.

When steel slags are aged, the leaching properties of the materials are changed. The total leachability and the pH decrease for all the investigated samples. All elements except magnesium decrease in leachability. As the slags are aged CaCO 3 is formed on the slag surfaces.

The degree of carbonation differs between different slags, due to the presence of different calcium-rich minerals in the slag. In order to form CaCO 3, the calcium-containing mineral must be dissolved. This means that the solubility of the calcium-containing mineral will affect the outcome of the carbonation. The rate of dissolution for six typical slag minerals was investigated in order to distinguish the difference in solubility between the different minerals. Acidic to alkaline pHs (4, 7 and 10) were selected to investigate the solubility of the minerals under conditions comparable to those prevailing in newly produced slags and the potential future pH values obtained under acid conditions. It can be concluded that all six minerals behave differently when dissolving and that the rate of dissolution is generally slower at higher pH. At pH 10, the solubility of merwinite, akermanite and gehlenite is considered slow. The dissolution of -Ca 2 SiO 4

is not affected in the same way as the other minerals when the pH is

changed.

ACKNOWLEDGEMENTS

First and foremost, I wish to thank my thesis advisors, Professor Bo Björkman, Associate Professor Caisa Samuelsson and Dr Qixing Yang for all your help and guidance during the course of my work. Dr Margareta Lidström Larsson, för att jag fick lära känna dig samt för den korta men lärorika tiden vi fick tillsammans inom området slagg, du är saknad.

I would also like to thank my colleagues and friends for putting up with me all these years. Danny P, Ryro, Ulla, Anaitich, Raggsockan, Challe and Secharo, you have all ‘now left the building’, so to speak. Åke B, Sempan, Biggish, Anders, Samuel, LILLKORVEN, U-tuff och sist men inte minst Muchtagellerna; without you, this work would have been considerably more difficult. Thanks also to Associate Professor Nourreddine Menad, now working at BRGM in France, for all the fruitful and entertaining discussions.

To member companies in MiMeR, Jernkontoret – The Swedish Steel Producers Association (TO55 – Steel production residues), VINNOVA – The Swedish Governmental Agency for Innovation Systems, Mistra – The foundation for Strategic Environmental Research and CAMM – Centre for Advanced Mining & Metallurgy, I extend my sincere thanks for invaluable financial support and commitment. Without your help, this study would never have been possible.

Finally, I wish to thank my family. Linda, thanks for all your help and wonderful support. Thank you Moa and Emil, you mean everything to me.

Johan, Mum and Dad, Thank you for helping to make me the person I am.

Fredrik Engström, December 2010, Luleå, Sweden

CONTENTS

1 INTRODUCTION ... 1

1.1 BACKGROUND... 1

1.2 STEELMAKING SLAGS ... 9

1.3 UTILIZATION OF STEELMAKING SLAGS ... 12

1.4 CRYSTALLIZATION THEORY ... 15

1.5 AIM AND SCOPE ... 19

2 MATERIALS AND EXPERIMENTAL PROCEDURE ... 21

2.1 MATERIAL ... 21

2.2 EXPERIMENTAL PROCEDURE ... 21

3 RESULTS AND DISCUSSION ... 27

3.1 LITERATURE REVIEW - HOT STAGE MODIFICATION, PAPER I ... 27

3.2 A MINERALOGICAL INTERPRETATION OF THE SOLIDIFICATION /SOLUBILITY OF STEEL SLAGS. PAPERS II & III ... 28

3.3 AGEING INVESTIGATION OF EAF SLAGS, PAPER IV ... 43

3.4 A STUDY OF THE SOLUBILITY OF PURE SLAG MINERALS, PAPER V ... 49

3.5 STABILITY OF SPINELS IN HIGH-BASICITY EAF SLAG, PAPER VI. ... 51

4 CONCLUDING DISCUSSION ... 55

5 CONCLUSIONS ... 63

6 FUTURE WORK ... 65

7 REFERENCES ... 67

1 INTRODUCTION

1.1 BACKGROUND

Large amounts of by-products are produced by the Swedish steelmaking industry each year. In 2008, the total amount of slag produced reached approximately 1.3 million tonnes, corresponding to almost 0.3% of the total steel slag production worldwide the same year [1,2]. 35%, mostly blast furnace slag, was sold as external products and approximately 20% was used for landfilling. The remaining 45% was used internally at the steel plants. The landfilled amount is high in comparison to Europe overall. In 2006, only 7% of the slags produced was dumped, while approximately 80% was used in external applications [1]. There are several reasons for the low utilization of steel slags in Sweden compared to other European countries: high availability of good stone material, high availability of land for landfilling, and a high share of scrap-based and high-alloy steelmaking.

In Sweden, a number of goals have been formulated in order to obtain a so-called good building environment, promoting sustainable management of land, water and other resources [3]. Among these criteria are:

• By 2010, extraction of natural gravel in the country will not exceed 12 million tonnes per year.

• The total quantity of waste generated will not increase and maximum use will be made of its resource potential while minimizing health and environmental effects and associated risks.

To date, these criteria “goals” have not been reached. Due to its strength,

durability and chemistry, steel slag could be considered in the field of

construction, since the material provides similar and sometimes even better

properties than granite and flint gravel [4]. Using slag in construction would contribute to a reduction in the amount of landfilled material and, at the same time, preserve natural resources, fulfilling the environmental goal.

However, the possibility of using slag is limited due to the lack of rules and guidelines regarding testing, assessment and use of slag in Sweden. The technical and environmental obstacles for not using some slags in construction include:

• Volumetric expansion

• Disintegration

• Leaching of metals

According to Monaco and Lu [5], the volumetric expansion is considered to be associated with the presence of free lime (CaO) and free periclase (MgO) in the solidified slag. Free lime and periclase react with moisture, resulting in an expansion due to the formation of Mg(OH) 2 and Ca(OH) 2 [6].

The understanding of these phenomena has led to the development of several treatment techniques, including steam and water treatment, in order to enhance the hydration [7].

It is well established that, upon cooling, pure dicalcium silicate undergoes

several phase transformations from the high-temperature -Ca 2 SiO 4 to the

low-temperature -Ca 2 SiO 4 polymorph, Figure 1. As the athermal,

martensitic-like transformation of the monoclinic -polymorph to the

orthorhombic -polymorph is accompanied by a volume expansion of about

12%, high internal stresses are built up in the slag during this

transformation, finally causing the disintegration of the slag [8,9].

Figure 1: Phase transformation of pure Ca 2 SiO 4 [10].

The polymorphic transformation of  to -Ca 2 SiO 4 is known to occur in

some EAF (Electric Arc Furnace) and AOD slags (Argon Oxygen

Decarburization) from stainless steel production depending on composition

and the cooling rate [11]. In addition to dust problems at the slag yard, the

usage of the disintegrating slag in external applications is limited due to the

high amount of fines. One route to preventing slag disintegration is to inhibit

the - to -transformation of Ca 2 SiO 4 or completely avoid the presence of

Ca 2 SiO 4 . The first option was elaborated in 1986 by the development of a

borate-based stabiliser for stainless steel slags [12]. Seki and co-workers

[12] corroborated that by adding borates to a high-temperature slag,

Ca 2 SiO 4 grains in the slag could be stabilized. The addition of only 0.2 wt%

of B 2 O 3 was sufficient to prevent the disintegration of a slag with 51 wt%

CaO, 33 wt% SiO 2 and 11 wt% MgO. Phosphorus also exhibits a stabilizing effect of Ca 2 SiO 4 . Experiments conducted in laboratory scale have shown promising results; however, larger amounts are required compared to borate additions [13]. Instead of using chemical additions, the - to - transformation of Ca 2 SiO 4 can also be avoided by rapid cooling [14].

According to the laboratory experiments, the required slag cooling rate is about 5°C/s [15,16]. This stabilization method was further developed by showing, in laboratory scale, that a granulation process transforms a disintegrating slag into a slag product suitable for construction applications [17]. The second option to avert slag disintegration is by avoiding the presence of Ca 2 SiO 4 by modifying the slag composition. Adding a relatively large amount of a SiO 2 source seems to be the best way to avoid Ca 2 SiO 4

precipitation. In laboratory experiments, AOD slag was stabilized with 12 wt% of waste glass, containing 70–75 wt% of SiO 2 [18]. The cost of such an operation would be far lower than that of the borate additions. However, the limited heat content and heat conductivity of slags make it difficult to apply this method of dissolving large amounts of SiO 2 in an industrial environment.

Apart from volume stability and disintegration, leaching of potentially

hazardous compounds during reuse is another key issue in slag

valorisation. The leaching from steel slags is generally characterized as a

surface reaction, followed by a solid-solid diffusion process, in order to

retain equilibrium in the materials [19]. It is therefore reasonable to believe

that the rate of leaching decreases with time as the diffusion from the bulk

of the solid slag to the surface is slow. Minimization of the surface area

be assumed to decrease the leachability. One way of introducing such a layer is by letting the slag react with CO 2 (g), forming calcium carbonates, CaCO 3 . Research has shown that carbonation of alkaline solid material can lead to an improvement of their environmental qualities [20,21]. The mechanism behind this formation depends on several factors such as temperature, particle size, porosity and CO 2 diffusivity [22-24]. According to Huijgen et al. [22], the diffusion of calcium through the solid slag matrix, towards the surface, appears to be the rate-determining step in carbonation, implying that the solubility of all calcium-containing mineral in the slag will affect the outcome of the carbonation of the slag surface.

However, when it comes to leaching, the exact mechanisms still remain unclear. Therefore, a lot of effort is being put into a mineralogical interpretation of the leaching. High resolution techniques such as XANES (X-ray absorption near-edge structure) and WDS (wavelength dispersive spectroscopy) together with simulations of the leaching behaviour are today used in order to characterize the slag in detail, in order to identify all possible sources of release. The possibility to simulate the leaching behaviour from aged steel slags, using geochemical modelling has been investigated [25,26]. These types of simulations sometimes fail to fully describe these complex systems, mostly because available data (solubility/thermodynamic/sorption) for the minerals occurring in the slag systems often are incomplete or missing.

Chromium is the metal in the slag that is given the most attention when it

comes to leaching. Several types of slag contain a significant fraction of

oxidized chromium [27]. This is especially the case for ferrochromium and

stainless steel slags. Nevertheless, chromium oxides are also encountered

in regular steel slags and several types of non-ferrous slags. When these

slags are recycled, chromium comes in contact with the environment.

Although Cr ³⁺ is an essential nutrient for mammals, elevated levels are hazardous and may cause, for example, skin rashes. Cr ⁶⁺ is carcinogenic;

hence, its presence in ground water should be prevented [28,29].

One way to decrease Cr release from the slag is by incorporating the chromium into stable mineral phases during cooling of the slag. The leaching levels are believed to be limited substantially when chromium is contained in a spinel phase—AB 2 O 4 , with A being a bivalent cation and B a trivalent cation [30]. Based on this, Kühn and Mudersbach [31] related chromium leaching to the concentration of spinel-forming compounds in the slag. They performed a large number of extraction tests on industrial stainless steel electric arc furnace slags and derived an empirical formula, factor sp, to relate the overall slag composition to the chromium leaching:

Factor sp = 0.2MgO + 1.0Al 2 O 3 +nFeOx 0.5Cr 2 O 3

‘n’ is a number between 1 and 4, depending on the oxidation state of the slag. When factor sp is below 5, a high chromium leaching is observed.

When factor sp is above 5, chromium leaching is low.

In addition to enrichment of chromium in spinel phases, microprobe

measurements performed on synthetic manufactured steel slags have

shown that chromium can be enriched in numerous phases, e.g. bredigite

(Ca 7 MgSi 4 O 16 ), merwinite (Ca 3 MgSi 2 O 8 ) and wollastonite (CaSiO 3 ) [32]. As

in the case of carbonation, this means that the solubility of the different

chromium-containing slag minerals must be known before the leaching of

metals, e.g. chromium, can be explained.

Apart from what already has been discussed, very little is reported in the literature regarding the influence of cooling on the properties of slag, especially for steelmaking slags. As the solidification behaviour is highly dependent on the exact composition, results are not easily transferable to other slags. Ground granulated blast furnace slags (GGBS) are known to possess improved hydration reactivity compared to slowly cooled blast furnace slag, due to the formation of glass [33]. The formation of a glassy material depends on both the chemical composition and the cooling conditions. According to Daugherty et al. [34], glass was easier to produce, as the acidity of the slag increased for a series of synthetic slag compositions that was quenched and annealed. Ionescu et al. [35], [36]

have shown how water quenching of steel slag results in products with a high content of glassy material. Silicate melts have high viscosity due to long molecule chains, and rearrangement into crystals only takes place slowly. If the cooling is rapid, the slag passes from a liquid state to a solid without development of a crystalline structure [37]. Glasses, such as granulated slags, can be regarded as super-cooled liquids having a very high viscosity.

Besides glass formation, controlling cooling conditions can be a means of affecting mineral transformation and, consequently, the solubility of elements like chromium. According to Lee and Nassarella [38], Cr ⁶⁺ is usually formed at lower temperatures. They suggest that by cooling the slag rapidly, the formation of Cr ⁶⁺ will be limited due to the slower kinetics.

Monaco and Lu [39] have reported variations in the composition of the

wustite-type solid solution as well as a variation in crystal size when cooling

differently. Despite extensive research efforts on these topics, a number of

questions remain. The leaching behaviour is still the primary cause for

difficulties with valorisation.

1.2 STEELMAKING SLAGS

Slag is formed through a chemical reaction of a flux and the gangue of an ore, the ash from a fuel, or with the impurities oxidized during the production and refining of a metal. In most iron- and steelmaking processes, the slag is in intimate contact with the liquid metal and chemical reactions readily occur between the slag and the metal. The main purpose of slag in processes such as basic oxygen furnace (BOF) and electric arc furnace (EAF) is to extract unwanted elements from the steel bath, help preventing metal oxidation and limit the heat losses from the steel [40].

1.2.1 BOF slag

BOF slag is produced as a by-product when iron is converted into steel in the BOF converter through the injection of oxygen. The slag is produced by the addition of fluxes, such as limestone and dolomite, during the process.

Unwanted elements such as carbon, silicon and phosphor are either oxidized to gases or chemically bonded into the slag. The most common minerals found in the BOF slag are listed in Table 1. All minerals may not necessarily be present in all slags of a given type.

Table 1: Most abundant minerals usually found in BOF slag.

Mineral name Structural formula

Larnite -Ca ₂ SiO ₄

Srebrodolskite Ca ₂ Fe ₂ O ₅

Tricalcium silicate Ca ₃ SiO ₅

Spinel (Fe,Mg,Mn,Al) Me ²⁺ Me ³⁺ ₂ O ₄

Wustite FeO

Lime CaO

Periclase MgO

1.2.2 EAF slag

EAF slag is produced as a by-product during the melting of scrap, when processed in the electric arc furnace at temperatures above 1600°C.

Phosphorus, sulphur, silicon and sometimes carbon are removed by lancing oxygen into the melt, forming a liquid oxide slag. In Sweden, two types of EAF slag are produced; carbon and high-alloyed steel. EAF slag which originates from the carbon steel manufacturing usually has a high content of iron and basicity B 2 (CaO/SiO 2 ) around 2.5, while the EAF slag generated from the production of high-alloyed steel might differ considerably. The most common minerals found in the two EAF slags are listed in Table 2.

Table 2: Most abundant minerals usually found in EAF slags.

Mineral name Structural formula Larnite -Ca 2 SiO 4

Srebrodolskite Ca 2 Fe 2 O 5

Brownmillerite Ca 2 (Al,Fe)O 5

Spinel (Fe,Mg,Mn,Al,Cr) Me ²⁺ Me ³⁺ 2 O 4

Wustite FeO

Periclase MgO

Mineral name Structural formula Bredegite Ca 14 Mg 2 (SiO 4 ) 8

Merwinite Ca 3 MgSi 2 O 8

Akermanite Ca 2 MgSi 2 O 7

Gehlenite Ca 2 Al 2 SiO 7

Cuspidine Ca 4 F 2 Si 2 O 7

Periclase MgO

Me ²⁺ Me ³⁺ O Electric arc furnace slag (carbon steel)

Electric arc furnace slag (high alloy steel)

1.2.3 AOD slag

In the manufacture of high-alloyed stainless steel the AOD (Argon Oxygen Decarburization) converter is often the first refining step after the electric arc furnace. By decreasing the oxygen content in the decarburization gas gradually, using inert gas (nitrogen and/or argon), a significant removal of carbon is achieved without extensive losses of chromium to the slag. The AOD slag often shows disintegrating properties due to the formation of - Ca 2 SiO 4 . The most likely minerals to be found in the AOD slag are listed in Table 3.

Table 3: Most abundant minerals usually found in AOD slag.

Mineral name Structural formula

Ingesonite -Ca ₂ SiO ₄

Larnite -Ca ₂ SiO ₄

Merwinite Ca ₃ MgSi ₂ O ₈

Melilite (Ca,Na) ₂ (Al,Mg,Fe)(Si,Al) ₂ O ₇

Fluorite CaF ₂

Spinel (Fe,Mg,Mn,Al,Cr) Me ²⁺ Me ³⁺ ₂ O ₄

Lime CaO

Periclase MgO

1.2.4 Secondary metallurgical slag

Secondary metallurgical slag is produced during the treatment of crude

steel, and arises from the subsequent treatment of steel produced both in

the BOF and EAF processes. Regardless of whether they originate in ore-

or scrap-based steelmaking, secondary metallurgical slags often have a

high content of CaO, SiO 2 , Al 2 O 3 and MgO. The most common minerals

found in the secondary metallurgical slag are listed in Table 4.

Table 4: Most abundant minerals usually found in secondary metallurgical slag.

Mineral name Structural formula

Ingesonite -Ca ₂ SiO ₄

Larnite -Ca ₂ SiO ₄

Bredegite Ca ₁₄ Mg ₂ (SiO ₄ ) ₈

Mayenite Ca ₁₂ Al ₁₄ O ₃₃

Tricalcium aluminate Ca ₃ Al ₂ O ₆

Cuspidine Ca ₄ F ₂ Si ₂ O ₇

Spinel (Fe,Mg,Mn,Al,Cr) Me ²⁺ Me ³⁺ ₂ O ₄

Lime CaO

Periclase MgO

1.3 UTILIZATION OF STEELMAKING SLAGS

An analysis of the focus areas in slag-related research in the periods 1996–

2000 and 2001–2004 indicates that slag recycling issues have gained importance in research, reflecting the increasing awareness of sustainable production and environmentally friendly processes [41,42], Within this research, the emphasis is primarily on finding and evaluating slag- containing products, such as:

• metallurgical fluxes

• cement

• aggregates for road and waterway construction

• soil improvers and fertilizers

The usage of iron and steel slags in external applications such as road

building and cement is nothing new. The earliest reports on the utilization of

slags refer to Aristotle, who used slags as a medicament already around

350 BC [43]. During the following centuries, slag was mainly used as construction material. The Romans used slag in road construction about 2000 years ago. In 1813, the first road made of slag in modern times was built in England. However, slag has not only been used in road construction. In 1589, the Germans used slag from iron making when manufacturing cannonballs [44].

In Europe, the total slag production in 2006 reached approximately 45.5 million tonnes (blast furnace slag included). 80% of the slag produced was used in external applications such as road construction and manufacturing of cement, Figure 2.

Figure 2: Use of steel slags in Europe 2006: 45.5 million tonnes [1].

In Sweden today, by far the largest external application area for slag is within road construction. The physical and chemical properties of the steel slags often make them suitable for ballast in asphalt manufacturing. The highly basic steel slags easily react with the acid bitumen, forming a high- performance asphalt product. Tests performed by the Danish Road Directorate show an increase of 30% in durability when comparing normal mastic asphalt to asphalt made with EAF slag. The increase in porosity compared to ordinary stone material also helps to keep the road free of water, thereby reducing the risk for aquaplaning while at the same time lowering the noise generated by the rolling tires. Apart from road construction, steel slags may be suitable for applications such as reinforcement of riverbanks and in the manufacture of cement and fertilizer.

As a fertilizer, converter lime combines many positive characteristics.

Thanks to its mineral composition, it promotes plant growth. The immune system of the plants is especially strengthened by the high concentration of silica in the slag [45]. Apart from external recycling of steel slags, internal recycling is also applied. Tata Steel [46], Sidera [47] and Ferriere Nord [48]

recycle their ladle slag back into the process as slag former. Apart from use as slag formers, ladle slag can also substitute some of the cement needed when manufacturing briquettes, which contributes to the reduction of CO 2

emissions.

1.4 CRYSTALLIZATION THEORY

As a new phase is formed in a system, the reaction is always associated with the Gibbs free energy of the system. When comparing Gibbs free energy of two mineral arrangements to determine which is the more stable, it is necessary to have a common basis for comparison. The widely adopted convention in chemistry is to compare Gibbs free energy levels based on the free energy of formation from the elements, G f . The free energy of formation is the energy difference between the elements in their standard state and the elements chemically bonded to form a mineral at a certain temperature and/or pressure. The stability of a mineralogical arrangement may be expressed in terms of Gibbs free energy according to reaction (1).

s) f(reactant )

f(products reaction

cal

mineralogi G G

G = − ⁽¹⁾

If, at a specific temperature – pressure condition, G reaction < 0, the product is more stable and the reaction may occur spontaneously. If G reaction > 0, then the reactants are more stable. The equilibrium condition is obtained if

G reaction = 0; thus, there will be no reaction taking place between the product and the reactants. All systems endeavour to minimize the total Gibbs free energy in terms of obtaining a stable state [49-51].

1.4.1 Nucleation

Homogeneous nucleation is by definition the formation of nuclei within one

phase. Conversely, where foreign elements are present, the result is

heterogeneous nucleation, i.e. formation of crystals through more than one

phase. In all chemical systems, the atoms are constantly moving, hence,

bumping into other atoms, and forming a variety of chemical combinations.

Some combinations, called embryos, will have the structure and composition of a mineral that could crystallize from the melt through nucleation. Most embryos will be small, consisting of only a few atoms, while some will be larger. Whether these embryos will grow to form minerals or not depends both on the stability and the size of the embryo. If the new embryo has a higher Gibbs free energy of formation than the melt, the embryo will not have enough power to grow. Instead, the embryos will break apart and constituent atoms will return to the lower energy level represented by the melt. A stable nucleus which can grow further will only exist if the embryo reaches a certain critical radius, r c . If a spherical crystal shape is assumed, then the critical radius of a nucleus that becomes stable can be obtained from equation (2), where , G v , G  and G f correspond to the surface energy per unit area, the volume energy, the surface energy and the free energy of formation [49-51].

( ) ( )

( ^{G G} ) ^{4 r 2 }

3 r 3 4

G 

G v

G tot _{f crystal f melt} ¸ ¸ +

¹

·

¨ ¨

©

§ −

= +

= (2)

In Figure 2, equation (2) is plotted as a function of the radius. As can be noticed, the Gibbs free energy shows a maximum at a certain radius. Only at radius higher or equal to this value is there a driving force for nuclei to grow, the critical radius. The relation between critical radius and G is expressed as;

T K

2

G 2

r c = − = ⁽³⁾

where K and T correspond to a constant and the degree of undercooling.

Figure 2: The Gibbs free energy of formation G tot of crystal nuclei is the sum of the surface energy G  and the volume energy G v . Only nuclei larger than the critical radius r c are stable with respect to the melt.

For crystals to grow, it is necessary to heavily undercool in terms of reaching sufficient activation energy and thereby overcome the difficulties related to the surface energy. The consequence of equation (3) will therefore be that a strong undercooling (high T) provides a high driving force and a small critical radius, which results in a high probability for nucleation to occur [49-51].

1.4.2 Growth

Once the nuclei are stable, the further growth is characterized by a gradual exchange of atoms at the surface of the nuclei. In order to predict the degree of solidification, both time and rate of growth must be considered.

r

Radius (r)

-  Gibbs fre e ener gy  +

Unstable Stable

G

0

G

G

Figure 3 schematically shows the temperature dependence between growth and nucleation.

Figure 3: The degree of reaction as a function of temperature.

T 1 = low degree of undercooling. The nucleation will be the rate-determining step for the crystallization. A small amount of nuclei with high growth will result in few, but large, crystals.

T 2 = high degree of undercooling. The growth will be the speed-dependent step in the crystallization. Many small nuclei with low growth rate result in numerous small crystals [49-51].

Rate of growth

T Under cooling

Rate of nucleation

T

(liquid slag)

Nucleation

Growth

T

T

1.5 AIM AND SCOPE

During the past 35 years, intensive research has been carried out related to the origin of the steel slags generated from the steel industry worldwide.

This research has contributed to the development of new products and the conservation of natural resources. Despite the effort that has been invested, approximately 3.2 million tonnes of slag was still sent for deposit in Europe in 2006. In order to develop new applications for slags, the fundamental properties that determine the quality of the product must be further investigated. Most of the research conducted so far has focused on solving questions related to volume expansion and disintegration. This research has led to the development of new methods in order to prevent the formation of, as well as to treat, steel slags showing these properties. In recent years, questions dealing with leaching have attracted increased interest. So far, few studies have been published linking the leaching properties with variation in process parameters. Thus, the work conducted within this thesis presents a study regarding different types of steel slags and how the fundamental properties of the steel slag are influenced by variation in process parameters. The aim of this thesis was;

I. To investigate the possibility of modifying steel slags in hot stage.

II. To determine how different cooling conditions influence the mineralogy and the leaching behaviour of some steelmaking slags.

III. To determine the long-term stability of some EAF slags.

IV. To determine how some of the individual minerals found in the slag behave when dissolving in aqueous media.

V. To understand the chromium distribution in some EAF slags.

2 MATERIALS AND EXPERIMENTAL PROCEDURE

2.1 MATERIAL

The chemical compositions of the materials used in the different studies are shown in Table 5.

Table 5: The chemical composition of the material used.

wt%

Material CaO SiO 2 MgO Al 2 O 3 Cr 2 O 3 MnO FeO Fe 2 O 3 Fe met

Paper II &III

Ladle slag 42.5 14.2 12.6 22.9 0.3 0.2 0.5 1.1 0.4 BOF slag 45.0 11.1 9.6 1.9 0.1 3.1 10.7 10.9 2.3 EAF-slag 1 45.5 32.2 5.2 3.7 4.8 2.0 3.3 1.0 0.1 EAF-slag 2 38.8 14.1 3.9 6.7 2.7 5.0 5.6 20.3 0.6

Paper IV

EAF slag 1 28.8 11.8 8.5 4.9 2.0 6.1 25.5 4.9 4.8 EAF slag 2 42.4 30.1 5.0 3.2 6.8 2.7 7.0 0.0 0.3 EAF slag 3 26.4 31.0 18.1 9.4 7.0 2.2 3.6 0.0 0.4 Paper V

Mayenite 48.5 51.5

Merwinite 51.2 36.5 12.3

Akermanite 41.1 44.1 14.8

Gehlenite 40.9 21.9 37.2

Ingesonite 65.1 34.9

Tricalcium aluminate

62.3 37.7

Paper VI

EAF slag 36.7 14.0 11.2 6.0 3.2 5.1 10.9 10.3 0.6

2.2 EXPERIMENTAL PROCEDURE 2.2.1 Chemical analyses

The total composition of each material was analysed by Ovako Steel AB

(Sweden) with inductively coupled plasma emission spectroscopy, ICP, and

x-ray fluorescence spectroscopy, XRF. The content of Fe and Fe oxid was

determined through titration. The analyses were done in duplicate and the results are presented as a mean value.

2.2.2 Specific surface area and density

The specific surface area was determined according to the BET-method with a Micromeretics Flowsorb 2300 and density was measured with a Micromeretics Multivolume Pycnometer 1305 on material prepared for leaching, <4 mm.

2.2.3 Glass analyses

The glass content was analysed by Scancem Research AB using optical microscopy according to ER 9103. A representative slag sample was ground and the fraction 32-40 m was used. In polarized light, optical isotropic grains, “glass”, have a different colour compared to crystalline anisotropic grains. If the grain contains more than 50% isotropic material it is identified as glass.

2.2.4 X-ray diffraction analyses

For XRD, all samples were prepared in a ringmill. The samples were analysed with a Siemens D5000 x-ray diffractometer, using copper K

radiation. XRD patterns were recorded in the 2-theta range 10 to 90°, in

0.02°/step. Initially, XRD patterns were recorded by counting 1 s/step, and

some samples were later rerun counting 3-8 s/step. The phase

identification was made by reference patterns in an evaluation program

supplied by the manufacturer of the equipment.

2.2.5 Scanning electron microscope investigation

The mineralogy of the slag samples was examined in a Philips XL 30 SEM using a beam operation voltage of 20 kV and spot size 6. Semi-quantitative and qualitative elemental analyses were performed with an energy dispersive spectrometer (EDS) fitted with an Everhart and Thornley detector behind a berylium window. Before mapping the texture and mineralogy of the fractured sample surface using the secondary electron (SE) image signal, the samples were sputter coated with a conductive layer of gold.

2.2.6 Leaching test

All the slag samples were crushed to a particle size of <4 mm and leached according to the one-stage batch test EN 12457-2 [52], except for two samples, ladle slag and granulated EAF slag 1 (Paper II), which were leached according to the two-stage batch test EN 12457-3 [53]. The filtrates were analysed by the laboratory Analytica AB (Sweden). The leaching tests were done in duplicate and the results are presented as a mean value.

2.2.7 Thermodynamic calculations in Factsage

Thermodynamic calculations were conducted using Factsage [54] using compound database FS53base.cdb, FToxid53base.cdb and solution database FToxid53soln.sda. FToxid-slag and FToxid-MeO were used.

During calculation, FS53base.cdb was suppressed contra

FToxid53base.cdb to exclude duplications in the data set.

2.2.8 Titrations

Titration of synthetic manufactured slag minerals (paper V) was conducted using a Radiometer Copenhagen ABU 901 Autoburette attached to a Radiometer Copenhagen PHM 290 pH meter. Assayed HNO 3 , 0.1 M was used as titrant. The choice of HNO 3 was made in order to minimize possible formation of complexes. For each experiment, approximately 0.05 g of mineral was used together with 100 ml of deionized water (Milli-Q). The titration was performed on a sized fraction, 20-38 m. The pH electrode was calibrated before each experiment with adequate standard solution (pH 4, 7 or 10). The temperature in the reaction vessel was kept constant (25°C), using a water bath. A magnetic stirrer was used for mixing during titration and the stirring speed was kept constant throughout the titration.

During titration, nitrogen was injected into the reaction vessel, protecting the system from CO 2 and formation of carbonates.

2.2.9 Carbonation analyses

The degree of carbonation was measured using thermal decomposition (TGA-MS) according to the method described by Huijgen et al. [22]. TGA- MS analyses were performed in a thermo-gravimetric analysis system (Netzsch STA 409) coupled with a quadruple mass spectrometer (QMS).

The samples were heated in alumina crucibles under an oxygen

atmosphere at 20°C/min from 25 to 1000°C. Weight loss was measured by

the TGA, while the gas was analysed for CO 2 and H 2 O. The analyses were

divided into three steps: Step (1) 25-105°C; step (2) 105-500°C; step (3)

500-1000°C. These steps represent (1) moisture, (2) organic elemental

carbon and MgCO (if present) and (3) CaCO (inorganic carbon),

respectively. During heating, the samples were kept isothermally at 105,

500 and 1000°C for 15 min, giving enough time for the reactions to fully

occur. The third weight loss from the TGA curve (m 500-1000°C ) was used to

describe the calcium carbonate content.

3 R ESULTS AND DISCUSSION

This chapter briefly summarizes the six appended papers. The purpose, methods used, and the main conclusions are presented for each paper.

3.1 LITERATURE REVIEW - HOT STAGE MODIFICATION, PAPER I When slag recycling issues are studied, the cold slag and its properties are generally considered to be fixed. The whole high-temperature slag treatment process, which results in the slag product, is too often ignored and disregarded. Even when the slag treatment process is considered to only begin at the moment of slag/metal separation to avoid making compromises towards metal or process quality, there is still considerable potential to influence the chemical composition and the mineralogy of the cold slag during the hot stage of slag processing.

Figure 5: General overview of the possible stages in slag processing. The stages of

interest for this article are indicated by the dotted ellipse.

The aim of this review article is to give an overview of the scientific studies dedicated to the hot stage of slag processing, i.e. from the moment of slag/metal separation to complete cooling at the slag yard, Figure 5. Using in-depth case studies on Ca 2 SiO 4 -driven slag disintegration, chromium leaching and influence of microstructure, it is shown that the functional properties of the cooled slag can be significantly enhanced by small- or large-scale additions to the high-temperature slag and/or variations in the cooling path, even without interfering with the metallurgical process. The technology to implement such hot stage processing steps in an industrial environment is currently available. No innovative technological solutions are required. Rather, advances in hot stage slag processing seem to rely primarily on further unravelling the relationships between process, structure and properties. This knowledge is required to identify the critical process parameters for quality control. Moreover, it could even allow purposeful alteration of slag compositions and cooling paths to tailor the slag to a certain application.

3.2 A MINERALOGICAL INTERPRETATION OF THE SOLIDIFICATION /SOLUBILITY OF STEEL SLAGS. PAPERS II & III

Four types of steel slags, a ladle slag, a BOF (basic oxygen furnace) slag

and two different EAF (electric arc furnace) slags, were characterized and

modified by semi-rapid cooling in crucibles and rapid cooling by water

granulation. The aim of this study was to investigate the effect of different

cooling conditions on the properties of slags with respect to their

mineralogy, leaching and volume stability. Optical microscopy, X-ray

diffraction, scanning electron microscope and a standard leaching test have

been used for the investigation. According to both the XRD and the SEM

difference in particle size distribution due to the different cooling conditions.

A summary of the phases found in the slag samples can be seen in Table 6.

Table 6: A summary of phases identified in the slag samples.

Original Rapid Original Semi-rapid Rapid Original Semi-rapid Rapid Original Semi-rapid Rapid Minerals:

Ca

SiO

-Ca

SiO

-Ca

SiO

-Ca

SiO

Ca

Mg(SiO

)

Ca

Fe

O

Ca

(Al,Fe)

O

CaAl

SiO

Ca

Al

SiO

Ca

Al

O

Fe

O

(Fe,Mg,Mn)O

MgO

CaO (ss)

Spinel (ss)

Ladle slag BOF-slag EAF-slag 1 EAF-slag 2

Ladle slag

Ladle slag is difficult to handle and store, due to its disintegrating properties. The XRD reveals that the ladle slag is the only one that becomes almost completely amorphous by granulation. The major mineral in the original slag is mayenite, Ca 12 Al 14 O 33 , followed in order by free MgO,

-Ca 2 SiO 4 , -Ca 2 SiO 4 and Ca 2 Al 2 SiO 7 . The -form may undergo a phase

transformation during cooling at 400-500°C to -form and the volume

increase (>10%) causes a pulverization of the slag [5]. The formation of -

Ca 2 SiO 4 is a plausible explanation for the disintegration. It was not possible

to filter the leaching solution of the original slag, which might be due to

cement-forming properties of the slag. One crystalline phase, undissolved

MgO, was identified in the granulated ladle slag, Table 6. With SEM and

mapping of selected elements, two phases were identified: a matrix consisting mainly of calcium, silicon and aluminium (glass matrix) enclosing small fragments of MgO (1), see Figure 6. The MgO particles are well distributed in the matrix (2).

Figure 6: Scanning electron micrograph of rapidly cooled ladle slag.

Dark fragments of (1) MgO in a matrix (2) with high content of calcium, silicon and alumina.

BOF slag

The original BOF slag has high specific surface owing to high content of

fines and pores compared to the granulated slag. According to the XRD

results, Table 6, the major phase in the original BOF slag is larnite, -

Ca 2 SiO 4 . With SEM and mapping of selected elements, silicon and calcium

were detected in the same phase, agreeing with the identification of larnite

as the major phase. Parts with high coexistence of iron, manganese and magnesium were also distinguished with SEM; possibly the (Mg,Fe,Mn)O solid solution also found with XRD. In the semi-rapidly cooled BOF slag, four crystalline phases were identified with XRD. A wustite-type solid solution containing magnesium, iron and manganese (Mg,Fe,Mn)O, - calcium silicate (-Ca 2 SiO 4 ), calcium ferrite (Ca 2 Fe 2 O 5 ) and a calcium, manganese oxide (Ca,Mn)O phase. All phases except for the (Ca,Mn)O were also detected with SEM and mapping, Figure 7, particle 1-3.

Figure 7: Scanning electron micrograph of the semi-rapidly cooled BOF- slag. (1) (Mg,Fe,Mn)O, (2) Calcium silicate, (3) Calcium ferrite.

In the rapidly cooled BOF slag, three crystalline phases were identified

through XRD. A wustite-type solid solution containing magnesium, iron and

manganese (Mg,Fe,Mn)O, tricalcium silicate and -Ca 2 SiO 4 . These phases

were also confirmed by SEM, phase 1-3, Figure 8, where phase 2 is tricalcium silicate (Ca 3 SiO 5 ), i.e. the euhedral prismatic microphenochrysts which, according to Goldring and Juckes [55], are typical for Ca 3 SiO 5 and phase 3, the matrix phase, crystallizing last, probably containing the - Ca 2 SiO 4 seen in XRD. All phases identified in the rapidly cooled BOF slag, agree with the thermodynamic calculation, Figure 9, i.e. indicating that the fast cooling enables the presence of metastable phases, such as Ca 3 SiO 5

and -Ca 2 SiO 4 at lower temperatures.

Figure 8: Scanning electron micrograph of the rapidly cooled BOF -slag.

(1) (Mg,Fe,Mn) oxide, (2) Calcium silicate, (3) Matrix containing (Ca,Si,Ti,V,Mn,Fe) oxides.

As seen in Paper 2, Figure 3, the composition of the wustite-type solid

solution is different when comparing the semi-rapidly and the rapidly cooled

BOF slag. According to the thermodynamic calculations, Figure 9, MgO is already present as crystals in the liquid slag at 1600ºC. The slight change in position which occurs in the diffractogram is explained by a higher concentration of MgO in the wustite-type solid solution. As the slag is cooled rapidly, neither FeO nor MnO has the same possibility of crystallizing and forming solid solution with MgO, due to i the later crystallization in comparison to MgO, Figure 9. The latter was further confirmed by the SEM analysis. According to semi-quantitative analyses, the solid solution contains 51 at% MgO, 42 at% FeO and 7 at% MnO, in the semi-rapidly cooled slag, while the solid solutions in the rapidly cooled slag were made up of 78 at% MgO, 16 at% FeO and 6 at% MnO.

When a phase can form thermodynamically, the crystal size will depend on the temperature to which the crystals are exposed and the duration of the exposure. There was a significant difference in crystal size between the two modified BOF slags, Figure 7 and 8. The size of the crystals present in the semi-rapidly cooled slag varied between 40-200 m, indicating that these minerals have had more time to grow. In the rapidly cooled BOF slag, the variation in crystal size is larger compared to the semi-rapidly cooled BOF slag. The wustite-type solid solution (phase 1) and the tricalcium silicate (phase 2), Figure 8, have a crystal size varying between 20-100 m. The matrix (phase 3), Figure 8, has a much smaller grain size than the other two phases discussed. The smaller crystal size of this silicate matrix can thus be explained in terms of not having the same time to develop. Based on the thermodynamic calculations, it can be concluded that both the wustite-type solid solution and the tricalcium silicate were present in the liquid slag at the time the rapid cooling with water was carried out.

However, -Ca 2 SiO 4 is expected to form during rapid cooling with water.

The leaching of calcium and iron is reduced in the granulated BOF slag.

Iron is present in the matrix, as discussed above, and the leaching is very low in all three slag samples. Calcium, on the other hand, is also present in the major silicate phase, Ca 3 SiO 5 . The solubility of silicon is increased in the granulated slag compared to the original. The leaching result shows that the dissolution of the minor elements is not prevented by the rapid cooling procedure. Vanadium is most soluble in the granulated BOF slag, correlating to the silica leaching, indicating a more enhanced leaching from the fine-grained silicate matrix.

Liquid slag

Ca

SiO

(s) Ca

SiO

(s)

MgO(s) CaO(s)

Ca

SiO

(s) FeO(s)

Ca

Fe

O

(s)

MnO(s) Ca

Fe

O

(s)

Ca

SiO

(s) Ca

SiO

(s)

Liquid slag Ca

SiO

(s)

MgO

CaO(s) FeO

MnO(s) γ −

γ − α′−

α′−

BOF-slag

T(ºC)

Ph a se d is trib u tio n wt %

25 203 380 558 735 913 1090 1268 1445 1623 1800

0 10 20 30 40 50 60

Figure 9: Thermodynamic calculation of BOF slag using Factsage.

EAF slag 1

In the semi-rapidly cooled EAF slag three crystalline phases were identified with XRD (Table 6) i.e., a spinel containing magnesium and chromium (magnesiochromite, MgCr 2 O 4 ), calcium magnesium silicate (merwinite Ca 3 Mg(SiO 4 ) 2 ) and -dicalcium silicate (-Ca 2 SiO 4 ). Both merwinite (phase 1) and magnesiochromite (phase 3) were found with SEM as well, Figure 10. From Figure 10, it can be seen that manganese is present in the solid solution along with chromium and magnesium. Except for the three phases which were found with XRD, an additional phase was observed with SEM and mapping, i.e., a calcium alumina silicate phase (phase 2), see Figure 10. According to the thermodynamic calculation, Figure 11, only one phase of calcium alumina silicate exists in the system, namely Ca 2 Al 2 SiO 7

(gehlenite). Gehlenite is thermodynamically formed below 1270ºC according to the calculations. The later crystallization of gehlenite agrees well with the texture of the semi-rapidly cooled EAF slag. As seen in Figure 10, both the merwinite and the spinel found have their own specific structures, characterized by sharp edges, while the gehlenite is located between the other two. According to the thermodynamic calculations in Figure 11 both merwinite and the spinel crystallize earlier than gehlenite, which explains the texture of the semi-rapidly cooled EAF slag.

Two crystalline phases were identified in the rapidly cooled EAF slag with

XRD, Table 6, both similar to those that were found in the semi-rapidly

cooled EAF slag, merwinite and the spinel, containing magnesium and

chromium. When comparing the diffractogram in Paper 2, Figure 3, a

broadening of the peak width can be observed as a result of the rapid

cooling with water. Suryanarayana and Grant [56] suggest that this may be

caused by a decreased crystallite size.

Figure 10: Scanning electron micrograph and accompanied

mapping of the semi-rapidly cooled EAF slag. (1) Calcium

magnesium silicate, (2) Calcium alumina silicate, (3) Chromium

containing spinel.

(MgO)(Cr

O

)(s) FeCr

O

(s)

FeCr

O

(s) ss-spinel

Ca

SiO

(s) Ca

Si

O

(s)

Ca

Si

O

(s)

Ca

MgSi

O

(s)

Ca

MgSi

O

(s)

Ca

Al

SiO

(s) MnAl

O

(s)

MnAl

O

(s) ss-spinel Liquid slag

Liquid slag (MgO)(Cr

O

)(s) ss-spinel

Ca

SiO

(s)

Ca

SiO

(s) Ca

Al

SiO

(s)

Ca

SiO

(s) α′−

α′−

α−

α−

EAF-slag 1

T(ºC)

P h a se dis tribut io n wt %

25 203 380 558 735 913 1090 1268 1445 1623 1800

0 10 20 30 40 50 60

Figure 11: Thermodynamic calculation of EAF slag 1 using Factsage.

Furthermore, the differences in crystal size found between the two modified

slags are significant (Figures 10 and 12). The spinel phase has the same

size and texture in both materials, while the rest of the phases vary

considerably, indicating that the spinels were crystallized already at the

time the granulation started, which is also confirmed by the calculations,

Figure 11. The large, well-defined merwinite and gehlenite crystals, which

were found in the semi-rapidly cooled slag, were no longer present in the

rapidly cooled EAF slag. Instead, a mixture of calcium, magnesium,

alumina and silica was found (area 2), Figure 12. According to the theory

on nucleation and growth, area 2 most likely consists of small merwinite

and gehlenite crystals, due to the rapid cooling i.e., the rapid crystallization.

Figure 12: Scanning electron micrograph of rapidly cooled EAF slag. (1) Chromium containing spinel, (2) Complex of (Ca,Al,Si,Mg) oxides.

The content of calcium and silicon is high in the EAF slag 1, Table 5. The

solubility of these two major elements, as well as aluminium, iron and

magnesium, is shown in Paper 2, Table 3. The leachability is very low and

varies in the three samples. The solubility of aluminium is reduced

substantially in the semi-rapidly cooled and the granulated slag, which

indicates that one of the matrix-forming phases is stable. On the other

hand, the mobility of silica seems to increase when granulating. There

does not seem to be any obvious correlation between the solubility of the

major and the minor elements. The varying dissolution of the metals

chromium, molybdenum and vanadium is more likely a result of the

presence in different minerals. The solubility of chromium is very low, 20

ppm of the total chromium content, in all three samples. Vanadium, on the other hand, is most leachable in the semi-rapidly cooled slag.

EAF slag 2

The XRD analysis reveals that the slag is very complex and some phases have varying content of substituted ions. The identified main mineral is - Ca 2 SiO 4 in both the original slag and the two modifications. A wustite-type solid solution ((Fe,Mg,Mn)O), Ca 2 (Al,Fe) 2 O 5 and Fe 2 O 3 were also identified, Table 6. A broadening in the diffraction peaks, indicating smaller particle size, could be seen when cooling rapidly. Calcium, iron and silicon are the major elements in the matrix of the EAF slag 2. As can be seen in Appendix 2, Table 3, calcium, aluminium and iron have the lowest leachability in the granulated slag, while silicon as well as the minor elements chromium, molybdenum and vanadium have the lowest leaching in the semi-rapidly cooled slag.

3.2.1 GLASS FORMATION (Paper II)

Daugherty et al. [34] claim that an acid slag M b (CaO+MgO/SiO 2 +Al 2 O 3 )<1

produces a glassy material more readily compared to a more basic slag

when cooled rapidly. The investigated slags are considered to be basic M b =

(1.4-3.9) and should therefore mainly contain crystalline material. The

measured glass content for the original and granulated slag is listed in

Paper 2, Table 1. Both the EAF slag 1 and the ladle slag show significant

changes in glass content. In order to determine the glass formation in the

slag, it is not enough to look at the chemical analyses. It is also important to

consider the chemical analyses of the remnant melt due to high-

temperature crystallization. To better understand the glass formation, the

crystallization path and corresponding melt composition at equilibrium conditions were calculated using Factsage. The calculated M b value is shown in Figure 13.

Figure 13: Glass forming tendency M b as a function of liquid slag temperature.

According to the thermodynamic calculations, the MgO crystallization from the liquid ladle slag starts already at approximately 1800°C. Only ~38% of the total MgO content is present in the liquid slag at 1400°C. The remaining 60% has already been crystallized as pure MgO. This phenomenon can also be seen in Figure 6. When MgO crystallization takes place, the M b - factor in the liquid material is changed from original 1.5 to 1.25 at 1400°C, influencing the glass forming properties in the material.

Due to early crystallization of solid solution, spinel phases, the liquid slag

composition of the EAF slag 1 is changed during cooling.

When the formation takes place, MgO reacts with chromium, forming magnesiochromite (MgCr 2 O 4 ). The formation of spinel is increased as the temperature of the liquid slag decreases, resulting in a lower M b factor. The M b ratio is decreased from 1.41 to 1.34. As seen in Figure 13, both the EAF slag 1 and the ladle slag tend to become more acid as the liquid slag temperature decreases, due to crystallization from the liquid slag.

Neither the BOF slag nor the EAF slag 2 shows any tendency of forming glass when cooling rapidly, according to Appendix 2, Table 1. The high M b

value shows that the slag is also basic at low temperatures, Figure 13.

3.2.2 REACTIVITY, BOF AND EAF SLAG 1 (Paper III)

The leaching from steel slags is generally characterized as a surface reaction, followed by a solid-solid diffusion process, in order to retain equilibrium in the materials [19]. A minimization of the surface area of the slag is therefore likely to enable a decrease in leachability. Leaching and specific surface data regarding these materials are listed in Paper 2, Table 1 and Table 3. The specific surface area data was unfortunately not measured in relation to the semi-rapidly cooled samples. However, since the original and semi-rapidly cooled samples were prepared for leaching in the same way (crushing < 4mm), it is assumed that these materials have a similar specific surface area.

It has earlier been concluded that no distinct changes in the total

leachability could be noticed when comparing the semi-rapidly cooled with

the rapidly cooled materials. However, a decrease in the specific surface

area was noted when the semi-rapidly cooled slag was compared against

the rapidly cooled slag. To gain a better understanding of the reactivity with

regard to the surface chemistry of the differently cooled slags, a reactivity

factor  was introduced and calculated according to equation (4). The