UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The hydraulic potential of high iron bearing steel slags Ionescu, Denisa V. 1999

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1999-463583.pdf [ 10.77MB ]
JSON: 831-1.0078677.json
JSON-LD: 831-1.0078677-ld.json
RDF/XML (Pretty): 831-1.0078677-rdf.xml
RDF/JSON: 831-1.0078677-rdf.json
Turtle: 831-1.0078677-turtle.txt
N-Triples: 831-1.0078677-rdf-ntriples.txt
Original Record: 831-1.0078677-source.json
Full Text

Full Text

THE HYDRAULIC POTENTIAL OF HIGH IRON BEARING STEEL SLAGS by DENISA V. IONESCU Metallurgical Eng., Polytechnical Institute of Bucharest, 1980 M.A.Sc. (Metals and Materials Eng.), University of British Columbia 1995 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES ( Department of Metals and Materials Engineering ) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA September, 1999 © Denisa V. Ionescu, 1999 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of £ SfcfeUkdj ^ The University of British Columbia Vancouver, Canada DE-6 (2/88) 11 ABSTRACT The incorporation of additives to the clinker or to the raw materials stream is a common practice in cement manufacture. However, steel slag, unlike its ironmaking parent the blast furnace slag, it is not a conventional admixture for cement. Currently most steel slags are slow cooled rendering stable crystalline compounds with minor hydraulic value. Nevertheless, if steel slags would be quenched and granulated, the resulting glassy product might display increased hydration and strength development potential. The use of steel slag in cement could contribute to important savings for both cement and steelmaking industries and provide a solution for the environmental problems linked to CO2 emissions and costs of transport and disposal. The purpose of this research is to explore the thermodynamics and kinetics of steel slag hydration in an effort to produce a cement additive, or a more promising material of near Portland cement composition. An important criteria used in the assessment of slags as potential cements is the presence of a glassy phase. At present, it is not very clear why glass enhances the hydration process. However, it is known that the free energy of formation for glasses is less than for crystals so that glasses are easier to hydrate compared to crystalline materials. In the particular case of steel slag, the glassy phase would have to contain high amounts of iron. Steel slags are known to display iron levels approximately 10 times higher than Portland cement and commonly used blast furnace slags. However, the effect of high Fe203 levels on the setting and strengthening of cement paste is not clearly understood due to the fact that most cement additives do not present this characteristic. I l l The present work looks at the progress made in recycling steel slag as cement additive, the complexity of the hydration process in slags, the possibilities of improving the hydration potential of slags at laboratory and industrial level, and the problems that still need to be addressed. However, the focus is on the glassy phase present in quenched steel slag and its influence on the hydration rate. B-SEM, Image Analysis, XRD techniques and a series of isothermal calorimetric experiments on synthetic as well as oxidized industrial steel slags vis-a-vis Portland cement assist in this endeavor. Temperature is a thermodynamic and kinetic factor modifying the enthalpy of hydrate formation (heat of hydration) and accelerating the hydration reaction. Hydration tests were carried out at temperatures ranging from 25 to 70 °C to determine the heat release, the rate of reaction and the apparent activation energy for steel slag and Portland cement hydration. The kinetics of hydration were explored in synthetic steel slags in both amorphous and fully crystalline form. The mechanism of hydration for both amorphous and crystalline steel slag was found to be a combination of nucleation and growth and diffusion, with higher reaction rates for the glassy slags. The higher reactivity in the glassy slags was explained by a lower activation energy when compared to the crystalline parent. Also, it was confirmed that slags have higher activation energies than Portland cement. As expected, and comparably to granulated blast furnace slags, quenched steels slags exhibited a significant hydraulic potential and hydrated at longer times, thus being expected to contribute to the late (> 5-180 days) strength development in slag - cements with compression strength superior to pure Portland cement. IV TABLE OF CONTENTS ABSTRACT ii TABLE OF CONTENTS iv LIST OF FIGURES viii LIST OF TABLES xii LIST OF SYMBOLS xiv GLOSSARY xv ACKNOWLEDGMENTS xvii Chapter 1 - INTRODUCTION 1 Chapter 2 - LITERATURE REVIEW 5 2.1 General Considerations on Portland Cement 5 2.2 Chemical Characteristics and Physical State of Slags Suitable for Cement-Making 10 2.3 The Chemistry of Steel Slags 14 2.4 State of the Art in Steel Slags Recycling as Cement Blends 18 2.4.1 The Challenges of Recycling Steel Slags vs. Blast Furnace Slags 19 2.5 Types of Glasses Encountered in Slags 24 2.5.1 Ferrite Glasses Present in Portland Cement Clinker and in Metallurgical Slags 30 2.5.2 Defects and Unstable Atomic Arrangement in Granulated Slags 33 2.6 The Hydration of Portland Cement 35 2.6.1 Development of Structure in Portland Cement Pastes 35 2.6.2 Setting and Hardening 37 V 2.6.3 Heat Evolution During the Hydration Process 38 2.6.4 Hydration of Individual Phases in Portland Cement 40 Tricalcium Silicate 40 Dicalcium Silicate 43 Tricalcium Aluminate 44 Calcium Aluminoferrite 44 2.6.5 Factors Influencing Cement Hydration 47 Cement Composition 47 Temperature 48 Fineness 49 Water/Cement Ratio 49 Age of Paste 50 Admixtures 50 2.6.6 Kinetics of the Hydration Process 51 2.7 The Hydration Process in Glassy Slags 53 2.7.1 Blast Furnace Glasses 54 Glass Composition and Reactivity 55 2.7.2 Hydration Mechanism in Slags and Slag Cements 57 2.7.3 Hydration Mechanism in Other Glassy Materials 64 C-A-S Type Glasses 64 Monocalcium Aluminate Type Glasses 65 2.8 The Activation and Properties of Slag Cement 69 2.8.1 Mechanism of Hydration 70 2.8.2 Products of Hydration 71 VI 2.8.3 Calcium Chloride as an Admixture in Cement 73 Chemical Effects of CaCl2 Addition 74 CaCl2 Effects on Cement Hydration 74 2.9 Energetics of Slag Hydration 78 Chapter 3 - SCOPE AND OBJECTIVES 86 Chapter 4 - METHODOLOGY 88 Chapter 5 - RESULTS AND DISCUSSION 92 5.1 Mineralogical Observations 92 5.1.1 XRD and B-SEM Analysis of Synthetic Slags 92 5.2 Hydration Study 97 5.2.1 Heat of Hydration of Portland Cement 97 The Influence of Temperature on the Heat of Hydration 98 5.2.2. Heat of Hydration of Synthetic Steel Slags 109 Quenched Slag (quasi-amorphous) 109 Annealed Slag (fully crystalline) 119 5.2.3. Hydration Behavior of Industrial Steel Slags 131 5.2.4. Hydration of Mixtures of Portland Cement and Alta Steel Slag(quenched) 132 5.2.6. Hydration of Synthetic Steel Slags in the Presence of CaCh 134 Chapter 6 - CONCLUSIONS 138 Chapter 7 - RECOMMENDATIONS 140 Chapter 8 - REFERENCES 142 APPENDIX 151 Appendix I - 152 Appendix II- 156 viii LIST OF FIGURES F i g . 1 - T e r n a r y d i a g r a m C a O - S i C v F ^ O s ( m o d i f i e d ) 11 F i g . 2 - R e l a t i o n s h i p B e t w e e n C o m p r e s s i v e S t r e n g t h a n d H e a t E v o l u t i o n o f 2 0 A l u m i n a D o p e d L D S l a g P a s t e s [9] F i g . 3 - S c h e m a t i c t w o - d i m e n s i o n a l r e p r e s e n t a t i o n o f g l a s s a c c o r d i n g to the 2 5 n e t w o r k t h e o r y [16] F i g . 4 - T w o - d i m e n s i o n a l r e p r e s e n t a t i o n o f s i l i c a g l a s s , a c c o r d i n g to the c r y s t a l l i t e 2 5 t h e o r y [16] F i g . 5 - T y p i c a l p r o d u c t s o f p o l y m e r i z a t i o n r e a c t i o n s o f i o n s o f g e n e r a l f o r m X 0 4 n " 2 8 [16] F i g . 6 - S c h e m a t i c R e p r e s e n t a t i o n o f the H y d r a t i o n a n d S t r u c t u r e D e v e l o p m e n t i n 3 7 C e m e n t P a s t e [47] F i g . 7 - S c h e m a t i c D e s c r i p t i o n o f S e t t i n g a n d H a r d e n i n g o f a C e m e n t P a s t e [47] 3 8 F i g . 8 - T y p i c a l C u r v e o f H e a t E v o l u t i o n i n the H y d r a t i o n o f O r d i n a r y P o r t l a n d 3 9 C e m e n t [12] F i g . 9 - D i s t r i b u t i o n o f Si A m o n g S i l i c a t e A n i o n T y p e s i n C 3 S P a s t e s [10] 4 3 F i g . 1 0 - R a t e s o f H y d r a t i o n o f the V a r i o u s C o m p o u n d s i n a P o r t l a n d C e m e n t [47] 4 7 F i g . 1 1 - E f f e c t o f a T e m p e r a t u r e o n the D e g r e e o f H y d r a t i o n as a F u n c t i o n o f 4 8 C u r i n g T e m p e r a t u r e [47] F i g . 1 2 - E f f e c t o f w/c R a t i o o n the H y d r a t i o n o f P o r t l a n d C e m e n t [47] 4 9 F i g . 1 3 - S c h e m a t i c D i a g r a m S h o w i n g the A c t i v a t i o n E n e r g y a n d the 51 T h e r m o d y n a m i c F o r c e f o r a R e a c t i o n [40] F i g . 1 4 - Si N K R S p e c t r a o f O r d i n a r y P o r t l a n d C e m e n t ( left ) a n d G r a n u l a t e d B l a s t 5 4 F u r n a c e S l a g (r ight)[58] F i g . 1 5 - T h e D e v e l o p m e n t o f the O H " C o n c e n t r a t i o n i n the P o r e W a t e r o f P o r t l a n d 5 6 C e m e n t P a s t e w i t h F l y A s h a n d F i n e Q u a r t z F l o u r ( T = 2 0 ° C , w / ( m i x ) = 0 . 4 5 ) [58] F i g . 1 6 - S c h e m a t i c S h o w i n g the P r o g r e s s i v e D e v e l o p m e n t o f M i c r o s t r u c t u r e 5 9 A r o u n d a S l a g G r a i n i n C e m e n t P a s t e [60] L I S T O F F I G U R E S (cont.) F i g . 17 - M i c r o s c o p i c Z o n i n g of a Partially Hydrated Slag G r a i n in Cement [60] 59 F i g . 18 - Spatial Distributions of Slag Hydrat ion Products [60] 62 F i g - 1 9 - Relationship of log (1- M\-a ) and log t o f the Hydrated Samples [16] 6 7 F i g . 20 - T h e Degree of Hydrat ion of Cement Pastes in the Presence of C a l c i u m 75 Chlor ide in Compar i son with Plain Paste, Measured by X - r a y Analys i s [75] F i g . 21 - Degree of Hydrat ion of a T r i c a l c i u m Silicate Paste in the Presence of 75 C a l c i u m Chlor ide; Measured by Non-evaporable Water Content [75] F i g . 22 - General ized Gibbs Free Energy-composit ion D i a g r a m , Showing the 78 Relation Between States [60] F i g . 23 - Schematic Representation of Free Energy for Granulated Blast-Furnace 80 Slags [81] F i g . 24 - Rate of Hydrat ion Heat vs. time [61] 84 F i g . 25 - B - S E M of Sample SS330, Showing 60 % Glass (matrix), 37 % C a l c i u m 95 Silicates (gray areas) and 3 % Periclase and Holes (dark areas) F i g . 26 - B - S E M of slag SS340 showing: 1) Increased A m o u n t o f Glass (62-70 %), 95 2) Larnite , 3) "Oxide" Phase, 4) Periclase F i g . 27 - B - S E M on the SS340 Sample. Periclase Crystals as R o u n d and Dendrit ic 95 Structures A r i s i n g F r o m a Glassy Matr ix F i g . 28 - B - S E M on Sample SS340 Showing Concentration of Periclase Crystals in 95 the Glass Matr ix F i g . 29 - B - S E M of Sample SS340, Glass Matr ix with Compos i t i on S imi lar to the 95 Ferrite So l id Solution Series (light gray areas), C 2 S (round and fingered dark gray crystals), Periclase (black areas) F i g . 30 B - S E M of Sample SS430, Glass (light gray), Larnite (dark gray), Periclase 95 (black) F i g . 31 - Heat of Hydrat ion for Portland Cement at 25 ° C and 0.5 w/c 98 F i g . 32 - Heat Evo lut ion Rate for Portland Cement vs. Temperature 99 F i g . 33 - Cumulat ive Heat vs. T i m e for Portland Cement 100 L I S T O F F I G U R E S (cont . ) F i g . 34 - Reciprocal o f Heat L iberat ion vs. Reciprocal o f React ion T i m e for PC at 102 Var ious Temperatures F i g . 3 5 - Determinat ion o f n and k for PC Us ing the M o d i f i e d A v r a m i M o d e l 105 F i g . 3 6 - Determinat ion o f n and k for PC Us ing the M o d i f i e d Jander M o d e l 106 F i g . 37 - Apparent Ac t i va t ion Energy fo r PC Us ing the A v r a m i M o d e l 107 F i g . 38 - Apparent Ac t i va t i on Energy for PC Us ing the Jander M o d e l 108 F i g . 39 - Heat Evo lu t ion Rate Curve fo r the Amorphous F o r m o f the Synthetic Slag 110 SS330 F i g . 40 - Cumula t ive Heat fo r SS330 (amorphous) at D i f fe rent Temperatures 111 F i g . 4 1 - Doub le Reciprocal o f Heat L iberat ion and React ion T i m e fo r SS330 111 (amorphous) F i g . 42(a) - The A v r a m i M o d e l fo r Slag SS330 (amorphous) Over the Ent i re 24 hr 114 Test Interval F i g . 42(b) - The A v r a m i M o d e l fo r Slag SS330 (amorphous); Da ta F i t t i ng in the 114 Accelerat ion and Decelerat ion Periods F i g . 43(a) - The Jander M o d e l A p p l i e d on Data Col lected fo r Slag SS330 (amorphous) 115 Over the Ent i re 24 hr Test Interval F i g . 4 3 ( b ) - The Jander M o d e l fo r Slag SS330 (amorphous); Data F i t ted in the 115 Accelerat ion and Decelerat ion Periods F i g . 44 - Rate o f Heat evo lu t ion D u r i n g the Hydra t ion o f T r i c a l c i u m Si l icate [10] 116 F i g . 4 5 - Ar rhen ius Relat ionship Us ing the A v r a m i Calculat ions fo r Slag SS330 118 (amorphous) F i g . 4 6 - Ar rhen ius Relat ionship Us ing the Jander Calculat ions fo r Slag SS330 119 (amorphous) F i g . 47 - Heat Evo lu t ion Rate Curve for Synthetic Slag SS330 (crystal l ine) 120 F i g . 48 - Cumula t ive Heat fo r SS330 (crystal l ine) at D i f fe rent Temperatures 120 F i g . 49 - Reciprocal o f Heat L iberat ion vs. Reciprocal o f React ion T i m e in SS330 121 (crystal l ine) at Var ious Temperatures xi LIST OF FIGURES (cont.) Fig. 50 - Avrami Model for Slag SS330 (crystalline) 122 Fig. 51 - Jander Model for Slag SS330 (crystalline) 123 Fig. 52 - Arrhenius Plot for Slag SS330 (crystalline) 124 Fig. 53 - Hydration of Slag SS340 (quenched) at Different Temperatures 125 Fig. 54 - Hydration of Slag SS340 (crystalline) at Different Temperatures 125 Fig. 55 - Hydration of Slag SS430 (quenched) at Different Temperatures 126 Fig. 56 - Hydration of Slag SS430 (crystalline) at Different Temperatures 126 Fig. 57 - Calorimetric Curve of C 3 A at Different Temperatures 129 Fig. 58 - Calorimetric Curve of C 4 AF at Different Temperatures 129 Fig. 59 - Calorimetric Curve of C3S at Different Temperatures 130 Fig. 60 - Calorimetric Curves for Mixtures of PC and Alta Steel Slag (quenched) 132 Fig. 61 - Effect of Slag and S0 3 (gypsum) Addition 133 Fig. 62- Hydration Curve for Slag SS330 (quenched) with 2% CaCl2 136 Fig. 63 - Hydration Curve for Slag SS330 (crystalline) with 2% CaCl2 136 Fig. 64- Hydration Curve for Slag SS430 (quenched) with 2% CaCl 2 137 Fig. 65 - Hydration Curve for Slag SS430 (crystalline) with 2% CaCl 2 137 xii LIST OF TABLES Table I Common Abbreviations Used in the Cement Industry [10,11,13,14,15] 7 Table II- Minerals Associated with Steel Slag [4] 14 Table III- Comparison of Cements and Slags Chemical Compositions [4] 16 Table IV - Chemical Composition of Laboratory LD Slags with Alumina Additions 21 [9] Table V - Chemical Compositions and Compression Strength for a Series of 30 Calcium Ferrites Mortars [26] Table VI - Slag Reaction Products [60] 72 Table VII - Chemical Composition of Synthetic Slags (Calculated Values) 92 Table VIII - Chemical Compositions of Synthetic Slags (Chemical Analysis) 93 Table IX - Phases Detected by B-SEM and XRD Techniques 96 Table X - Heat of Hydration at 24 hr for Portland Cement 99 Table XI - Values of Q; and t 5 0 for Portland Cement 102 Table XII - Constants n and k for Portland Cement Using the Modified Avrami and 106 Jander Models Table XIII - Heat of Hydration at 24 hr for Synthetic Slag SS330 (amorphous) 110 Table XIV - Values of Qi and t 5 0 for Synthetic Slag SS330 (amorphous) 112 Table X V Sequence of Hydration of the Calcium Silicates [10,40] 116 Table XVI - Constants n and k for Slag SS330 (amorphous) Employing the Modified 117 Avrami and Jander Models Table XVII - Heat of Hydration at 24 hr for Synthetic Slag SSS30 (crystalline) 121 Table XVIII - Values of Qj and t 5 0 for Synthetic Slag SS330 (crystalline) 122 Table X I X - Constants n and k for Slag SS330 (crystalline) Using the Modified 123 Avrami and Jander Models Table X X Values of Qi, t 5 0 and E a for Synthetic Slags SS340 and SS430 in 124 Amorphous and Crystalline Form LIST OF TABLES (cont.) Table XXI - Heat Released in Mixtures of PC and Alta Steel Slag 133 Table XXII - Final Heat Release for Slags SS330, SS340 and SS340 in the Presence 134 of CaCl2 LIST OF SYMBOLS a Degree of hydration E a Apparent activation energy (KJ/mole) A G Free energy change (KJ/mole) k j Rate constant in Jander equation ( s 1 ) k A Rate constant in A v r a m i equation (s"1) n Kinet ic parameter in A v r a m i and Jander equations Q i Estimated final heat (J/g) R Gas constant ( J / ° C m o l e ) t T i m e (hr) to T i m e at the end of the induction period (hr) tso T i m e at which 50 % hydration occurs (hr) T Temperature ( ° C or K ) XV GLOSSARY [31,106] Calcium Chloride A crystalline solid, CaCl2; used as a drying agent, as an accelerator of concrete and other purposes. Calorimeter An instrument for measuring the heat exchange during a chemical reaction such as the hydration of cement. Cement Kiln A kiln in which the ground and proportioned raw mix is dried, calcined and burned into clinker at a temperature of 1420 to 1650 °C; can be of rotary, shaft, fluid-bed, travelling grate type; fuel may be coal, oil or gas. Cement, Hydraulic A cement that sets and hardens by chemical interaction with water and is capable of doing so under water. Cement, Portland A hydraulic cement consisting predominantly of calcium silicates which react with water to form a hard mass. Clinker Concrete C-S-H Early Strength Ettringite Ferrite Phase Glass The material coming out of the cement kiln after burning. It appears as dark, porous nodules. It is ground and mixed with small amounts of gypsum to give cement. A composite material that consists essentially of a binding medium within which are embedded particles or fragments of aggregate; in Portland cement concrete, the binder is a mixture of Portland cement and water. Amorphous calcium silicate hydrate formed in the hydration of C3S and C2S. It has a variable stoichiometry. Strength of concrete or mortars usually developed at various times during the first 72 hr after placement. First hydration product formed when tricalcium aluminate ( C 3 A ) hydrates in the presence of gypsum. Calcium aluminoferrite solid solution with the composition range C 6 A 2 F -C 6 A F 2 and a mean close to C 4 A F . Amorphous, solid substance without an ordered structure over a range greater than several times the dimensions of the individual unit cells. It is formed by a continuous transition from the liquid to the solid state. On heating, glass softens to form a viscous liquid. Considered to be a supercooled liquid that cannot crystallize because of its high viscosity (>1012Pas). Granulated Blast Furnace Slag Glassy, granular material formed when molten blast furnace slag is rapidly chilled, as by immersion in water. xvi GLOSSARY (cont.) Gypsum Heat of Hydration Heat of Solution Calcium sulfate dihydrate, CaS04-2H20(CSH2). Added to cement to regulate setting. Heat evolved by chemical reactions with water, such as that evolved during the setting and hardening of Portland cement, or the difference between the heat of solution of dry cement and that of partially hydrated cement. Heat evolved or absorbed when a substance is dissolved in a solvent. Hematite Hydration Induction Period A mineral, iron oxide (Fe 203) used as aggregate in high density concrete and in finely divided form as a red pigment in colored concrete. The reaction of cement with water. Period of dormancy in the early hydration of cement. This results in concrete remaining in a plastic state for several hours after first mixing with water. Paste Pozzolan Setting Slag cement Steel Slag Soundness Mixture of cement and a limited amount of water. When the mass has reacted with water and developed strength it is called hardened cement paste. A siliceous or siliceous and aluminous material, which in itself possesses little or no cementitious value but will, in finely divided form and in the presence of moisture, chemically react with calcium hydroxide at ordinary temperatures to form compounds possessing cementitious properties. Transformation of cement paste or concrete from a fluid-like consistency to a stiff mass. A hydraulic cement consisting mostly of a blend of granulated blast furnace slag and hydrated lime. A by-product in the steelmaking production, it is usually slow cooled in air, and has little hydraulic properties; used as an aggregate in concrete or as a fertilizer in agriculture. The freedom of a solid from cracks, flaws, fissures or variations from an accepted standard; in the case of cement, freedom from excessive volume change after setting. XVII ACKNOWLEDGMENTS I would like to express my gratitude and thanks to my supervisors, Dr. T. R. Meadowcroft and Dr. P. V. Barr for their advise, expertise and encouragement during the course of the project. Many, many thanks to the faculty, staff and fellow graduate students in the Department of Metals and Materials Engineering for their advise, suggestions, technical as well as moral support, great intellectual exchanges and good times spent together. The list of people I should thank is very long and I apologize for not mentioning them here. However, brighten up, I'm planning to write a book about it all ! (Any resemblance with the real characters will be mere coincidence). Financial assistance from NSERC (Canada) and SJDOR (Venezuela) is gratefully acknowledged. Chapter 1 - INTRODUCTION 1 CHAPTER 1 INTRODUCTION The production of metals via pyrometallurgical processes inevitably involves the generation of oxide slags. Due to chemistry restrictions, slags have few commercial applications, causing a significant problem to metal producers in terms of cost of transportation, disposal, and environmental concerns. Thus, from the perspective of the steelmaker, slag disposal and/or recycling is a significant problem for which there is no current solution. Steel slags, which are produced at a rate of 100-200 kg per tonne of product, consist of burnt lime and calcium silicates, and therefore might be useful as additives in cement production, downstream of the energy intensive and CO2 generating clinkering process [1]. In the production of cement clinker, which constitutes the main raw material for cement-making, for each tonne of clinker, about a tonne of CO2 is released into the atmosphere. Wi th the introduction of ever more strict environmental regulations, the substitution of cement clinker by other cementitious materials, amongst them steel slags, offers an attractive method of reducing CO2 emissions by the cement industry. Current international accords stipulate very strict target values that all countries subscribing to these agreements must comply with and several organizations have been created for the purpose of finding technical solutions to this problem using the concept of sustainable development. The cement industry is one of the largest sources of CO2 emissions. Although blast furnace slag is a relatively common additive to cements, at present, steel slags are used only as minor additives in Portland cement (PC) manufacture due to problems arising from their mineralogy and chemical Chapter 1 - INTRODUCTION 2 composition which cause quality problems in the product. Canada generates about 1 million tonnes of steel slag and 9 million tonnes of Portland cement annually, and adopting a modest substitution ratio of 1:10, all slag could be recycled if these quality issues could be successfully addressed. In Portland cement production, 80% of the total energy is consumed in burning the raw mix. The use of slags as additives to the clinker mix avoids this step. The granulation process does not demand too much energy, as slag grinding consumes only 10 to 30% more energy than clinker grinding. The saving of energy in comparison to common Portland cement production can be up to 60% [2]. Portland cement is by far the most important cement in terms of the quantity produced. It is made by heating a mixture of limestone and clay, or other materials of similar composition, to a temperature at which partial fusion occurs. The product, called clinker, is ground and mixed with gypsum, the latter acting as a retarder of the setting process avoiding a brisk hardening and hence the generation of fractures or other types of defects in the final concrete. The ground clinker from the kiln contains four main phases: tricalcium silicate (C3S), (3-dicalcium silicate (C2S), tricalcium aluminate (C3A), and ferrite solid solutions with a composition between C2F and C6A2F, often approximating to brownmillerite (C4AF). Characteristics of the hydrated product, such as the rate of hardening, the rate and extent of heat liberation and the resistance of concrete to corrosion (eg. by sulphate solutions), are influenced by the proportions of these four main components. Industry standards suggest that a 'good' cement clinker should contain the calcium silicates and aluminates, together with at least 15 % glassy phase and very low levels of free calcia and magnesia [3]. Because alumina and iron oxide act as fluxes during cement burning (clinkering), in order to minimize the temperature and time required to complete the clinkering process both are included in the raw material stream for cement making. However, the amount and valence state of the iron oxide are known to influence the cementitious properties Chapter 1 - INTRODUCTION 3 of the clinker. Although Portland cement contains only small amounts of iron oxide, acceptable cements with levels of iron comparable to steel slags (eg. high alumina and Erz cements) are produced. This indicates that, despite the very poor cementitious properties exhibited by 'raw' steelmaking slags, the relatively high Fe content is not the base problem. Steel slags are characterized by high amounts of iron (30-40%), mostly in the form of wustite (FeO) and hematite (Fe2C>3). Recent research [4] has shown that, by changing the iron valence state from Fe 2 + to Fe 3 +, the hydraulic potential of steel slag (defined as reactivity of slag towards water and capability of developing high compression strengths) can be much improved, presumably due to the enhanced glass forming capability of the molten slag during solidification. The glass forming capability of a slag is also known to be strongly influenced by the cooling rate. Currently, most steel slags are slow cooled, rendering stable crystalline compounds with insignificant hydraulic potential. However, an amorphous or quasi-amorphous product with superior hydration potential can be obtained from steel slags if the cooling rate is increased to levels similar to those imposed on cement clinker [4]. Although most researchers acknowledge the importance of the glassy phase in the cement clinker, studies have been commonly centered on blast furnace slags, which have a low content of iron oxide (0.3-2.0 %), and little emphasis has been given to the role of glasses in the actual hydration process. Consequently, the understanding of the thermodynamics and kinetics aspects of the hydration process of steel slags, the nature and role of steel slag glassy phases and the hydration potential of the ferrite phase (which is responsible for englobing most of the iron oxide contained by the slag) is still limited. Reflecting the increased interest of the steelmaking industry in finding a viable solution for slag disposal, three recent projects at UBC [4, 5, 6], have dealt with important aspects in the fabrication of steel slag cement, including the iron solubility limit in silicate glasses [5], the Chapter 1 - INTRODUCTION 4 oxidation state of iron in glasses and steel slags [4], and the compression strength of cements containing steel slags [6]. The present study extends this work by examining the role of glassy phase formation on the hydraulic potential of steel slags. Particular attention was given to glasses with compositions close to the ferrite solid solutions, which are known to display cementitious behavior. A l l the materials were analyzed in both their fully crystalline and quasi-amorphous forms, the intent being to understand and quantify the effect of the amorphous phase on the hydration process. Chapter 2 - LITERATURE REVIEW 5 CHAPTER 2 LITERATURE REVIEW 2.1 General Considerations on Portland Cement In its broadest sense, 'cement' denotes any kind of adhesive. In civil engineering, it denotes a substance which can be used to bind together sand and stones, or other aggregates, into a concrete. Hydraulic cements, of which Portland cement is a familiar example, harden by reacting with water and yield a water-resistant product. The hydraulic process begins with the mixing of water and solid 'cement' to produce a slurry paste. After a few hours, the paste develops rigidity (sets) and then steadily increases in compressive strength (hardens) by chemical reaction with water (hydration). A cement that continues to increase its strength while under water is called hydraulic. In this process, nearly all of the cements rely on the formation of hydrated calcium silicates, aluminates or alumina sulphates. Cement clinker, the main raw material for cement-making, contains four main cementitious phases: tricalcium silicate (C3S), P-dicalcium silicate (P-C2S), tricalcium aluminate (C3A), and ferrite solid solutions, with composition between C 2 F and CeA2F with C4AF as a mean. Portland cement characteristics are influenced by the proportions of these four main phases. The anhydrous calcium silicates and aluminates differ markedly from each other in their reactivity towards water. C2S exibits various polymorphs which can behave quite differently. For example, P-C2S, reacts slowly but is strongly hydraulic, while y-C2S is inert. The aluminates, C3A and d 2 A7 react rapidly, but pastes made from them develop little strength. Other compounds, such as gehlenite (C2AS), wollastonite (P-CS) and calcium hexaluminate Chapter 2 - LITERATURE REVIEW 6 (CA6> are inert at room temperature. Moreover, the main phases present in Portland cement are not pure. For example, C 3 S contains small amounts of M g 2 + and A l 3 + and these variations in composition appear to influence reactivity. For the ferrite phase, the reactivity increases with the Al/Fe ratio. The fact that an anhydrous calcium silicate or aluminate reacts with water does not necessarily mean that it is a hydraulic cement. For example, C3A reacts rapidly but develops low strengths. The determination of the compression strength assesses the hydraulicity rather than chemical reactivity. Nevertheless, there is usually a correlation between the increase in strength in concrete and the degree of hydration of cement [7, 8, 9]. Moreover, the hydration products have to be taken into account. A summary on the hydration process and the main hydration products in cement pastes is presented in the chapter discussing the hydration of Portland cement. The presence of numerous chemical compounds and constituents associated with cement lead to rather complicated formulae and mineralogical denominations. For brevity, a list of constituents relevant to the present research and their abbreviations, is provided in Table I. in r f 1—l rf 60 03 o 3 W I 'S 1 « H P Q Z H Z W S w u H Z Q P z o H z O O u S va >, u U O ca 3 §<S 2 o E o. o So| •S 'S u J3 | 1 c o, O Ml w £3 O 00 X CJ „ c o 2 o * S 6 o 00 s-o c >-> ca o u oo 00 ca .2 S o •-• "3 S 'C 23 H -5 6 |o "ca o -3 ca ca .ts o u a u o oo 11 41 A 5 -< u u oo u I a 00 U 00 (S u CO. too .fi o u , fi J3 "o I I T3 1 1 — CD c/i OH cn 60 C O J O _T pa £ J so C Q J O _T pa mJ 3 •a O <L> C V s> S . £5 on ^ 3 o fi 8 S 5 'S "* • J= -5 cn 60 > cn T S? 2 U 3 •° § —* fi J= a 60 « a °£ S u p < DC .5 E CU O C3 ' c 03 ' C O H 3 C3 C3 "2 ca a a ca c o 60 ca c o 60 ca X o £ (•> s o e , >> on OH o eu E . « Z T3 H~> c3 J3 D. 3 cn s "ea u ea HH - a a ca o ca HH ca o u 3 'S o HH T 3 >, . C E 3 'o la U 3 E eu HW ea £ H o t« OH <1 O 3 .2 I ' M HW K s -a § «! O O <S" 6 6 cs ea u u V O M3 GO C M 6 6 0 6 ca U cs U CN 6 6 0 s 6 ea u U u O V) c *4—> O o CS c o 0 0 CS X . « BC o 'c 73 o c o o 0 0 CS X CJ CS c o 0 0 CS X CJ J 3 o 0 0 CS X CJ o 0 0 CS X CJ o o 2 o o •S CJ V—< CA , o O J CS -a CJ 11 6 o o oo •e 'B B O a CJ c S3 0 0 •a 3 3? 2 DC oo| T 3 C CS CJ T 3 C E 2 £ § .5 0 0 ^ UH 3 „ O CS CL CJ < CJ CJ C CJ 0 0 ( CL CJ ° j= — oo CL is ca cu s z cs cs C CS s 3 73 CS CS 6 _ 3 7s cl cj .2 ts 7s U B cs t~< T 3 > » x: CJ CS c 6 3 CS S 3 7 j 7s U > CJ t-l - O X> CS O •J-. 73 CJ" 'ob CL 3 O UL cj < • t i t o « cd G cj .2 cs cS^DC •a 0 0 CS CS £ 'S o D. , °> \< X oo O CS U cs o X o 5^o s 52 •£ DC I 3 < | U ~ u DC < oo 6 , u PL vo <? U * CS r7 cj co E .2 DC |loo < vO u DC = .S too DC < < U U o Chapter 2.2 - Chemical Characteristics and Physical State of Slags 10 2.2 Chemical Characteristics and Physical State of Slags Suitable for Cement-Making The compositions of slags potentially suitable for cement manufacture fall into the quinary system CaO-Si02-Al203-MgO-Fe203, and may vary significantly in hydraulic properties depending on the chemical composition, the presence of constituents other than those of the quinary system, and the amount of crystalline material. Fig. 1 represents a ternary diagram CaO-Si02-Fe203 modified with the addition of Al203 and MgO and it shows the approximate location of typical oxidized steel slags, blast furnace slags as well as various cements [6]. It can be seen that the lime to silica ratios typical of cements also include steel slag. However, the Fe203 content in steel slags (30-40 %) is very much higher than in cements (3-13 %). The presence of certain constituents in the slag has been shown to lead to the deterioration of the final hardened cement [1, 16, 17, 18]. For example, unassociated MgO (periclase) is detrimental because of the volume increase experienced by this constituent when it hydrates to form brucite. For CaO there are some restrictions that result from the phase change of P to y dicalcium silicate (2CaO-Si02) which is associated with a volume increase of the slag [1]. However, several works have shown that rapid quenching of the slag prevents the phase change [19, 20]. Although some research has been done into the suitability of high-magnesia or high-lime slags for cement production, the entire range of slag compositions available for cement manufacture has not been studied to determine the extent to which these are suitable for inclusion into cements. In the present work, the concept of physical state of slags relates to the crystallinity level of slags. Depending on the chemical composition and the cooling rate, slags may have different crystallinity levels. One criterion suggested in the assessment of slags as potential cements is that crystalline material should not exceed 15 % of the mass of quenched slag, since crystalline Si0 2 + Al 2 0 3 1. - Oxidized steel slag 2. - Portland cement 3. - Erz cement 4. - High alumina cement 5. - Blast furnace slag 6. - Synthetic slag SS330 7. - Synthetic slag SS340 8. - Synthetic slag SS430 Fig. 1- Ternary diagram CaO-Si0 2-Fe 20 3 (modified) [6] Chapter 2.2 - Chemical Characteristics and Physical State of Slags 12 slags possess little or no hydraulicity [1]. Slags that cannot be successfully quenched to yield acceptable high amorphous (glassy) structure are considered unsuitable for cement fabrication. However, it should be noted that Portland cement clinker is normally 75-80 % crystalline [18, 21], and it exhibits an increased hydration potential due to its specific mineralogical composition, predominantly due to the presence of tricalcium silicate. In an amorphous slag, Ca 2 + ions englobed in the voids of the aluminosilicate skeleton (and acting as a modifier of the structure) will be the first ones to react with water, the reactivity of the slag being determined by the rate of this interaction. If the slag is less basic, Ca 2 + will not occupy all the voids in the structure and then other ions, eg. A l 3 + will act both as network former and as network modifier, by occupying some of the free voids. In this case, A l 3 + will be more unstable, hence more reactive. This explains the positive role of an increased content of AI2O3 in slags with reduced content of CaO. When Mg 2 + ions are present, they will play the role of network modifiers. Other ions like Mn 2 + and S2", will act as both network formers and modifiers, depending on the nature of the slag and the thermal equilibrium attained. Iron is normally considered as a crystal producing element. However, the influence of Fe on crystallinity will depend on the form of the iron. Hematite (Fe20s) in the tetrahedral form acts as a glass former, while in the octahedral form it shows a natural tendency to crystallize [22]. Magnetite (Fe304) in tetrahedral form, is a glass former, while wustite (FeO) is not. The explanation for this varied behaviour is that, whereas Fe 3 + can have tetrahedral or octahedral coordination [23] and therefore can be linked to the silica tetrahedra, Fe 2 + has a sixfold coordination, leaving non-bridging oxygens in the silicate structure. The bivalent iron Fe 2 + behaves only as a network modifier and promotes the formation of small orthosilicate ions, which are more apt to crystallize upon cooling of slags. A recent study [4] shows that by raising the oxidation state of the bivalent iron to trivalent iron in the molten steel slag it is possible to Chapter 2.2 - Chemical Characteristics and Physical State of Slags 13 i m p r o v e the g l a s s f o r m i n g p o t e n t i a l . F e 3 + acts as n e t w o r k f o r m e r , t h e r e b y d e c r e a s i n g the b a s i c i t y a n d s h i f t i n g the i o n i c s t ruc tu re f r o m the o r t h o s i l i c a t e c o m p o s i t i o n t o w a r d s l a r g e r m o r e c o m p l e x i o n i c r i n g s . T h e t h e r m a l h i s t o r y ( m e l t i n g t e m p e r a t u r e a n d c o o l i n g rate) i n f l u e n c e n o t o n l y the ra t io o f the p h a s e s p r e s e n t i n the s l a g , bu t a l s o the i r s t ruc ture . A n i n c r e a s e i n m e l t i n g t e m p e r a t u r e a n d c o o l i n g rate w i l l p r o d u c e a less o r d e r e d s t ruc tu re , a h i g h e r c o n c e n t r a t i o n o f s t r u c t u r a l d e f e c t s , the m o d i f i c a t i o n o f the c o o r d i n a t i o n n u m b e r o f the a c t i v e i o n s a n d , at a m a c r o s c o p i c l e v e l , a m i c r o p o r o u s s t ruc tu re . A l l these a s p e c t s w i l l l e a d to a n i n c r e a s e d s l a g r e a c t i v i t y [17] . Chapter 2.3 - The Chemistry of Steel Slags 14 2.3 The Chemistry of Steel Slags Steel slags contain all the major oxides present in the raw materials for cement manufacture. It might therefore be expected that the main cementitious components present in cement clinker would also be present in the steel slag. However, the chemistry of steel slag is imposed by the requirements of the refining process and, in addition, the cooling rate after tapping is very slow. Several mineralogical studies on cooled steel slags [24, 25, 26] indicate the possibility of forming calcium silicates, ferrites, wustite and wustite solid solutions. Nevertheless, variations in composition and cooling rate of the molten slag make it difficult to predict which mineral phases will be formed or their distribution. Table II shows the predominant mineral phases associated with steel slag. Table II - Minerals Associated with Steel Slag (modified') T41 Formula Name Synonym Crystallization Melting Point System (°C) 3CaOSi0 2 Alite Tricalcium Silicate Triclinic 1900 2CaOSi0 2 Belite Dicalcium Silicate Monoclinic 2130 p-2CaOSi0 2 Larnite Dicalcium Silicate Monoclinic 0i '-CaOSiO 2 Bredigite Dicalcium Silicate Orthorombic 2CaOFe 20 3 Dicalcium Ferrite Calcium ferrite Orthorombic 1430 4CaOAl 2 0 3 .Fe 2 0 3 Brownmillerite Calcium aluminoferrite Orthorombic 1410 FeO Wustite Ferrous oxide Cubic MgO Periclase Magnesium oxide Cubic 2800 CaO Lime Calcium oxide Cubic 2570 During the cooling of steel slag, dicalcium silicate (C2S) will be the predominant species. C 2 S undergoes a series of phase transformations, passing from a metastable P to a stable y phase. This transformation can be prevented by the presence of impurity ions, which stabilize the P-C2S Chapter 2.3 - The Chemistry of Steel Slags 15 structure. Tricalcium silicate (C3S) is formed when the cooling rate is fast enough to prevent the phase transformation of C3S to lime and C2S. A summary of previous work on iron presence in steel slags [24] shows that iron oxide is found primarily as divalent wustite (FeO) with some trivalent hematite (Fe2C*3). Wustite forms solid solutions with CaO, MgO and MnO, while Fe 2 03 combines with lime and alumina to produce calcium ferrites and calcium aluminoferrites. Ferrites form a solid solution series ranging from C2F to C6AF 2 , with the mean composition close to C4AF (brownmillerite) [24, 26, 27]. Nevertheless, wustite solid solutions are always more abundant in the slag than the ferritic ones [24]. The cementitious properties of steel slag are not substantial, due to the deficiency in C3S, which is the primary strength contributing phase in Portland cement hydration, and the presence of the wustite solid solutions which do not form hydraulic phases. Recent work [4] concluded that, by oxidizing the steel slag, the trivalent iron is capable of combining with calcia and alumina to produce a ferrite solid solution upon cooling from the melt. This ferrite has a composition similar to brownmillerite. This study emphasizes that by oxidizing the slag it is possible to convert the mineral phases into a strength contributing phase compound, namely C4AF. In fact, 30 day compression tests for slag-Portland cement mortars showed strength increases over pure Portland cement mortars, at substitution levels of up to 10 percent. Several other studies report on the use of ground steel slag in cement production [28, 29, 30, 31]. Steel slag was used 'as is' and mixed with other cementitious materials forming pastes with strength comparable to ordinary Portland cement. From these studies, it is difficult to quantify the strength advantages or disadvantages introduced by the steel slag itself. Furthermore, they imply that the high content of free lime present in the steel slag makes it an Chapter 2.3 - The Chemistry of Steel Slags 16 activator (in blends of steel slag and blast furnace slag), while disregarding the hydration potential of the glassy phase and other cementitious compounds contained in the slag. Since one of the problems in the use of steel slag in cement fabrication is the high iron content it is useful to look at cements that, by virtue of their chemical composition, are tolerant to increased amounts of iron. The literature provides information on two types of cements: one is the iron ore or Erz cement, composed essentially of lime, silica and ferric oxide and showing all the physical properties of Portland cement though with a lower rate of strength development; the other is alumina cement [18]. Table UJ shows that the iron oxide content in blast furnace slag is very low, (close to Portland cement), while Erz and high alumina cements have iron contents and lime to silica ratios similar to steel slag. Table III - Comparison of Cements and Slags Chemical Compositions [41 Constituent Portland Blast Furnace Slag* Steel Slag ** High Alumina Erz Cement Cement Cement* CaO 64.1 36-45 35 37.7 64 S i0 2 22 33-42 18 5.3 22 A1 2 0 3 5.5 10-16 3.6 38.5 2.8 Fe 2 0 3 3.0 0.3-2.0 8.8 12.7 6.2 MgO 1.4 3-12 11.5 0.1 0.9 S0 3 2.1 - - 0.1 1.8 MnO - 0.2-1.5 6.5 - -FeO - - 18 3.9 -CaO/Si0 2 2.9 0.9-1.4 1.9 7.1 2.9 CaO+MgO/Si0 2+Al 20 3 2.4 - 2.2 0.8 2.6 *Lea [18], **Stelco Slag Essentially, blast furnace slag contains CaO, MgO, Si02 and AI2O3. Its mineralogical components are mainly aluminum melilite, magnesium melilite and P-C2S. Most of the minerals do not have hydraulicity, except for small amounts of P-C2S. Slow-cooled B F Chapter 2.3 - The Chemistry of Steel Slags 17 slag is not hydraulic and, as Shushan et al. [38] indicated, only glassy or microcrystal granules resulting from fast cooling display hydraulic activity. As in Portland cement clinker, the hydraulic potential of steel slag comes from its silicates and ferrite-aluminate minerals. In fact, steel slag might be regarded as a low quality clinker. The steel slag components are rather complex, all of them minerals, and primarily, solid solutions. A small amount of Fe203 and A1 2 0 3 dissolves into C3S, P2O5 into C 2 S, and C2F is the solid solution that contains AI2O3 [35, 36]. Another finding is that when the amount of steel slag increases in a mortar, water requirement decreases, however, strength is only slightly lower than for cement clinker. Steel slag blended cement containing 50-60 % Portland cement clinker, has a 28 day compression strength equivalent to Portland cement, and its late strength development can be even greater than that of Portland cement [34, 36]. It is known that steel slag has similarities to clinker and it is able to form hydration pastes and contribute to the late strength development of steel slag-cement mixtures. Nevertheless, extensive investigations have to be still carried out before the factors controlling volume stability and hydraulicity are understood quantitatively. Also, more emphasis has to be given to the ferrite phase and hence to the glassy phase, as it has been generally observed that glass is a necessary presence for a sound cement. In terms of using steel slag as an addition to conventional Portland cement, the key consideration is whether it is placed in the flowsheet up or downstream of the clinkering kiln. Addition downstream of the kiln would reduce energy demand by the kiln in direct proportion to the substitution level and also contribute to the reduction of CO2 emissions, since lower quantities of materials are processed inside the kiln [28]. Inserting upstream of the kiln really offers few advantages since one cheap raw material is just replacing another. Chapter 2.4 - State of the Art in Steel Slag Recycling as Cement Blends 18 2.4 State of the Art in Steel Slag Recycling as Cement Blends For both ecological and economic reasons, the recycling of industrial by-products via the cement industry is an active area of research. The cementitious materials most frequently used are silica fume, fly ash and blast furnace slag. The use of blast furnace slag as a cementitious material has been practiced in Europe since the late 1800s, but it was not until World War U that slag cement started to be used extensively as an energy saving measure. At present, slag cement represents -20% of the total cement production in Europe but only 1% in the USA [32]. In many countries the production of blended cement containing up to 80% of slag exceeds 50% of the total cement production. These blended cements are used especially in structures exposed to sulphate attack and in massive structures, where a low heat of hydration is beneficial. The basic material is granulated blast furnace slag of specified specific surface area (250 to 700 m2/kg) [2]. Granulated, vitreous blast furnace slag is often used, but partly crystalline slag (3 to 40% crystalline phase) may also be incorporated. The advantage of finely ground slag is its long term stability in comparison with Portland cement, even in moist air. Blast furnace slag ground to 310 m2/kg and stored in a silo for 2 1/2 years showed only slightly retarded setting and lower early strengths. In general, steel slags require prior treatment before use as cement additives. The following section reviews the likelihood of performing slag treatments at industrial level without introducing technical and economical challenges in the steelmaking process, and without impairing the quality of steel itself. At present, steelmakers avoid modifying their process flow in order to accommodate the treatment of the slag immediately after tapping, citing high costs and technical problems. However, this trend may change in the future. Several studies [32, 34, 35, 36] already indicate the possibility of using different methods for altering the chemistry and Chapter 2.4 - State of the Art in Steel Slag Recycling as Cement Blends 19 mineralogy of steel slags in the steelmaking shop. However, more information is required on how to improve the hydraulic properties of the steel slag and, consequently, render it more suitable for cement-making. 2.4.1 The Challenges of Recycling Steel Slags vs. Blast Furnace Slags The steel industry has been actively exploring options to improve the hydraulicity of steel slags. Conjeaud et al [9] proposed altering the chemistry of steel slags by adding a synthetic slag-forming flux during the normal steelmaking process. The transformation of the waste slag into a by-product equivalent in quality to blast furnace slag is done without changing the technology or compromising steel quality. The flux contains 30-50% CaO, 22-29 % AI2O3, 9-12 % MgO, 14-19 % Fe203 and 5-6 % S i 0 2 . By adding the flux, the alumina level in the steel slag is increased which transforms the hydraulically inert calcium ferrite to the substantially hydraulic calcium aluminoferrite. When alumina is higher than 15 %, an intense vitrification of the slag occurs. Measurements of heat release versus time during the setting process are commonly made by isothermal conduction calorimetry. For synthetic slags with > 15 % AI2O3 these measurements show more heat release than ones with 10 % AI2O3. It is not clear why. One might infer that the higher heat is due to increased amounts of calcium aluminoferrite, or may be due to the combined effects of higher amounts of calcium aluminoferrite and increased amount of glassy phase. The research concluded that a slag must render at least 83 J/g in order to display useful hydraulic behavior. Only slags with over 15 % AI2O3, and hence glassy, complied with this criteria. Another interesting observation [9] is that normal cements generate considerable heat of hydration within 24 hours, but slags do not follow the same pattern and the setting peak and hence the heat appears at longer times. The heat evolution is slow and an accelerator is generally Chapter 2.4 - State of the Art in Steel Slag Recycling as Cement Blends 20 required. The correlation between heat evolved and mechanical strength development in slag-cement pastes is clearly shown in Fig. 2. Fig. 2 - Relationship between compressive strength and heat evolution of alumina doped steel slags from the L D process [9]. During hydration the amount of heat evolved depends on the quantity and nature of hydrates formed. It should be noted that the hydrates are the binding agents which give rise to cohesive forces, since there are other factors that influence the outcome, such as morphology and packing factors. The heat evolution is only a partial basis for the interpretation of the hydration behavior. Nevertheless, calorimetric results are generally accepted as one measure of the hydraulicity of slags. Furthermore, Conjeaud et al [9] reported the chemically altered slags contained P-C2S, C3A, calcium aluminoferrites, periclase and (Ca, Fe, Mn, Mg)0 solid solutions. Iron was present mostly as FeO (16-20%), magnesio-wustite and aluminoferrite. Free lime was less than 4 %, to Chapter 2.4 - State of the Art in Steel Slag Recycling as Cement Blends 21 avoid expansion problems during the hydration process. The chemical compositions are given in Table IV and it can be seen they were characterized by a high iron oxide content (around 20 %). T A B L E IV - Chemical composition of Laboratory L D Slags with Alumina Additions [9] CaO (%) Si0 2 (%) FeO (%) Fe 20 3 (%) A1 20 3 (%) T i0 2 (%) MgO (%) P 2 0 5 (%) MnO (%) Loss (%) LD110 41.9 11.4 10.6 10.2 11.2 0.6 7.8 1.6 1.1 1.9 LD210 42.3 11.2 6.1 17.6 11.1 0.5 4.4. 1.8 4.9 1.0 LD120 37.3 10.2 10.1 7.1 21.5 0.6 7.0 1.5 1.0 1.7 LD230 32.9 8.7 11.2 6.6 30.8 0.4 3.4 1.4 3.8 1.0 In these slags, the role of AI2O3 was to participate in linking the F e 2 0 3 in the calcium aluminoferrite phase and enhance the formation of a glassy phase. Slags hydrated in spite of high levels of wustite, and wustite solid solutions, being present. The samples with higher amounts of A1 2 0 3 and higher amounts of glassy phase displayed higher heats of hydration (125-146 J/g above 15 % A1 2 0 3 vs. 63-71 J/g below 15% A1 2 0 3 ) . Furthermore, it was concluded that these slags were fully crystalline and their hydraulic activity was not enhanced by quenching, although mentioning that an intense glassification is present above 15 % A 1 2 0 3 when slags are composed entirely of glass and (Ca, Fe, Mn, Mg) O solid solutions and producing high heats of hydration. In view of these findings and the lack of clarity, it would be interesting to further explore the hydration behavior of glassy steel slags. These aluminous steel slags are more reactive than blast furnace slags at early ages. At longer times though (beyond 90 days), mixtures of slags:Portland cement (50:50 ratio) produce compression strengths close to pure Portland cement pastes. The products of hydration formed by these slags are calcium hydrate and calcium ferrite hydrate, which react rapidly with C 0 2 from Chapter 2.4 - State of the Art in Steel Slag Recycling as Cement Blends 22 the atmosphere producing calcium-mono-carbo ferrite hydrate, analogous in structure to calcium-mono-carbo-aluminate hydrate. If the alumina content is high, then calcium aluminate hydrate forms, which reacts with CO2 to form calcium-mono-carbo-aluminate hydrate. Fe(OH) 3 was not detected, and if present, it was assumed to be amorphous. Recently, a new BOF slag treatment technology was reported at Thyssen Stahl [32, 33] in which the slag is treated outside the steelmaking vessel in the slag pot, without therefore compromising the steelmaking process or increasing the tap-to-tap times. Although the resulting slag is intended solely for recycling as aggregate, it is interesting to note that the slag basicity is lowered by adding sand (target binary basicity CaO/Si02 < 3) and oxidized to encourage the formation of calcium ferrites. As shown previously [4], the oxidation of Fe 2 + to Fe 3 + , favors the glassification of the slag, with the additional benefit of producing the heat necessary to dissolve the sand into the slag. Moreover, the oxidized iron binds the free lime into calcium ferrites. More than 200 industrial trials have been successfully performed. The equipment for the treatment of slags is placed immediately downstream of the converter in the steel shop. After tapping the slag into the slag pot, the pot is moved by the slag ladle transfer car into the treatment stand. Sand and oxygen are simultaneously added via pneumatic injection. Following the treatment in the slag pot, the material is transported and poured into specially prepared slag pits. Although tests have been carried out to determine the influence of cooling rate on the characteristics of the solidified slag, the results have not yet been reported. The study concludes that the treated slag has good volume stability and resists leaching of heavy metals and is therefore viable as concrete aggregate. Wachsmuth et al [37] discussed the necessity of chemically altering steel slags after tapping by way of a special treatment which allows reducing the amount of reactive free lime. The presence of free lime represents one of the problems in the use of steel slag, since it causes Chapter 2.4 - State of the Art in Steel Slag Recycling as Cement Blends 23 expansion during hydration. Using S E M techniques they determined that the first precipitate during the solidification of steel slag was the dicalcium silicate and depending on the slag composition, the second precipitate could be lime. However, lime was also present as primary undissolved lime added to protect the lining. In either category it exists in different forms: the residual lime in a spongy and grainy form and the precipitated lime on grain boundaries between wustite and C 2 F, or together with C 3 S or C 2 S. The spongy form is infiltrated by iron oxide rich slag. These are particles in the process of dissolution, and probably, form the calcium ferrite phases. The proposed treatment consists of holding the slag in the slag ladle for a certain time, prior to pouring, so that residual lime becomes dissolved. Pilot tests have been carried out with promising results. Another work highlighting the hydraulic potential of steel slag and its contribution to the late strength development [35] focused on crystalline steel slags, containing 0- C 2 S, C4AF and (Fe,Mg)0 and (Ca,Fe)0 solid solutions. Although blends of blast-furnace: steel slag: Portland cement (50:30:20) showed lower initial strength development than blast-furnace slag: cement mixes, after 4 weeks, the strength was equivalent to Portland cement. The products of hydration were found to be C H , C 4 A H 1 3 , "monocarbonate" C 3 A C H 1 1 , C 3(A,F)H6, and eventually C-S-H. After 3 months a great part of the Ca-Al-hydrates had converted to hydrogarnet. Moreover, a difference in the hydrate morphology appeared depending on the hydration medium; e.g., water reacted slags exhibited tabular plane crystals, while NaOH hydrated slags were tabular, but bent. Chapter 2.5 - Types of Glasses Present in Slags 24 2.5 Types of Glasses Encountered in Slags Since cement clinker, as well as slags used in cement fabrication, contain glassy phases, it is useful to explore the nature of glasses. Most substances when heated above their melting point and cooled in the absence of nucleating substances can be undercooled; that is, they remain liquid at temperatures below the stable melting point. An undercooled liquid becomes more viscous as its temperature is lowered and changes into a rigid, isotropic solid. Glasses can be regarded as undercooled liquids [16]. The structure of glass has been explained in terms of the random network theory and the crystallite theory. According to the random network theory, glass constituents can be divided into two classes: network formers and network modifiers. The formers, characteristically SiC"2, AI2O3, B2O3 and P2O5, provide cations which link together appropriately by sharing oxygen atoms to form randomized three-dimensional networks of tetrahedra. The modifiers, such as Na20, K 2 0 and CaO provide ions which occupy spaces in the network in such a manner that properties such as specific volume are little affected by the number of spaces filled. In order to produce an overall neutrality, bonds must form between formers and modifiers. As a general rule, for a tetrahedral type structure, if the formula of a glass is A m B„0 and B is the glass former element, the minimum value of n is 0.33 and the maximum 0.50. Fig. 3 shows a representation of glass according to the network theory [16]. The crystallite theory postulates that glass contains small regions of crystalline order (containing most cations), linked together by amorphous regions (formed by residual anions). Fig. 4 represents the structure of the silica glass according to the crystallite theory [16], the Chapter 2.5 - Types of Glasses Present in Slags 25 • Network-forming ion # Network-mod ifying ion Q Bridging oxygen ion © Non-bridging oxygen fan Fig. 3 - Schematic two-dimentional representation of glass according to the network theory [16] Fig. 4 - Two-dimentional representation of silica glass, according to the crystallite theory [16] Chapter 2.5 - Types of Glasses Present in Slags 26 shaded regions being the more crystalline. It is not possible to decide between the two theories on the basis of radial intensity distributions obtained from X-ray diffraction, as the crystallites postulated are only a few angstrom units in size. However, when analyzing the radial distribution of intensity and calculating the interatomic distances both theories give reasonable agreement. Support for the crystallite theory is based on the chemical behavior of certain glasses, in which part of the glass may be leached out, leaving the main network intact with little change in mechanical properties [16]. Ironmaking slags contain between 20 to 40 % silica and, therefore, they consist of silicate structures and possibly silicate glasses. The structure of a glassy silicate slag can be approached by considering first a crystalline form of silica such as quartz. The latter has a three dimensional framework of SiC>4 tetrahedra [10]. Each tetrahedron, except on the surface of the crystal, shares all four corners with other tetrahedra. The O-Si-O and Si-O-Si angles are 109.5° and 140° respectively. The term "crystalline" means that the pattern of tetrahedra repeats at regular intervals in all directions. In a vitreous silica, the O-Si-O angles are close to 109.5°, but the Si-O-Si angles differ from 140° and the pattern is no longer of a repeating nature. The structure of the silicate glass is further modified by the weakening of some Si-O-Si bonds: Si - O- Si + O2" -> Si - O" + " O - Si The negative charges are neutralized by the incorporation of metal cations which are called 'modifiers'. The framework is broken down and the glass will include isolated straight and branched chains. Some of the Si positions may be occupied by atoms of other elements, especially A l . Each replacement of S i 4 + by A l 3 + leaves a negative charge that will be further balanced by the introduction of other metal cations. An increased amount of C a 2 + or similar ions Chapter 2.5 - Types of Glasses Present in Slags 27 will break down the tetrahedral framework, thus being associated with increased depolymerization or fragmentation of the tetrahedral framework. In the case of silicate and aluminate structures (which distinguish cements), two features are found. One is the polymerization of silicate tetrahedra into larger groups, pairs, finite groups, chains, sheets and frameworks, and the other is the formation of coordination polyhedra of oxygen atoms around calcium ions and other cations that might be present [10, 39, 40, 41]. The interaction of these two features, together with the ordered or statistical substitution of cations gives silicate structures their complexity as copolymers of silicate and another matching cation polyhedra. Fig. 5 gives the typical products of polymerization reactions for Si, as well as other elements, showing their complexity. The influence of the glass structure on the hydraulic reactivity of slag has been studied but the opinions are not always in agreement. However, there is a general consensus in the sense that slags should be as disordered as possible [10]. A glassified slag, which under equilibrium cooling conditions would form a crystalline solid, exists in a metastable state. This type of slag is considered to have an excess of internal energy attributable to quenching, therefore the non-crystalline solid has a greater energy content than the parent crystalline one [4]. The cementitious potential of an amorphous slag should be directly attributable to the glass and its ability to hydrate. Several studies [11, 16, 42] reported on the fabrication of synthetic glasses with composition approximating blast furnace slag. Compositions included gehlenite (C2AS), and mixtures of it with C2S. Glasses were activated with Ca(OH)2, cement clinker or gypsum. It was concluded that artificial glasses did not follow the same pattern of behavior as commercial slags. Other group of researchers fabricated 34 glasses in the system CaO-Al 2 03 -S i02 and tested them in conjuction with Portland cement clinker [11]. They found that glasses and slags have similar Chapter 2.5 - Types of Glasses Present in Slags 29 results in terms of compression strength resistance. Nevertheless, it is difficult to correlate the results because the type and amount of activator depends on the composition of the slag and also because Ca(OH)2 appears to have a different activation mechanism compared to cement or gypsum. Studies on glasses with composition approximating steel slags are not reported in the literature. In order to better understand the hydration behavior of steel slag, attention should be given to the composition and reactivity of this type of glasses. Therefore, this project will examine steel slags with increased amounts of glassy phase, as well as quasi-amorphous synthetic slags resembling the glassy alumino-ferrites present in Portland cement clinker. Section 2.5.1 - Ferrite Glasses Present in Portland Cement Clinker and in Metallurgical Slags 30 2.5.1 Ferrite Glasses Present in Portland Cement Clinker and in Metallurgical Slags The glassy phase in Portland clinker is the aluminoferrite phase. This phase consists of a solid solution series ranging in composition from C 2 F to C 6 A 2 F . A composition typical of the ferrite phase found in clinkers is Ca2AlFeo.6Sio.15Tio.05O5, with the mean composition corresponding to C4AF, also termed browmillerite. In spite of these variations, the aluminoferrite phases are commonly expressed in terms of their ideal compositions: 4CaOAl 2 03 -Fe 2 03. Dragoi et al [26] studied the hydraulic properties of a series of alumino-ferrites using their compression strength development. The compositions and compression strength data are shown in Table V . T A B L E V - Chemical Compositions and Compression Strength for a Series of Calcium Ferrites Mortars [26] No. M A I Probable CaO A1203 Fe 20 3 3 days 7days 1 4 days 28 days Compound % % % daN/cm**1 daN/cm 2 daN/cm 2 daN/cm 2 1 1.91 C 8 A 3 F 49.07 33.51 17.42 26.40 13.21 164.30 189.00 2 1.70 C7.85A2 .65F 50.59 31.11 18.30 23.90 14.94 102.80 165.30 3 1.50 C6.64A2.34F 48.18 31.04 20.65 35.90 18.02 130.00 193.00 4 1.26 C 6 A 2 F 48.07 29.19 22.74 41.20 22.01 154.70 215.30 5 1.10 C 5 . 5 1 A 1 7 i F 48.06 21.17 24.77 31.70 14.20 61.00 111.00 6 1.00 C 9 . 1 6 A 1 5 6 F 47.59 26.21 26.19 77.70 37.60 142.60 237.90 7 0.90 C4.9cA1 .40F 47.62 24.78 27.60 22.90 18.05 83.60 125.70 8 0.70 C 4 . 3 9 A . 1 . 0 9 F 47.64 21.55 30.18 33.30 19.45 50.80 54.80 9 0.64 C 4 A F 46.19 21.03 32.78 11.50 8.50 23.30 21.31 10 0.50 C4.56AFL28 45.53 18.18 36.28 35.60 18.12 63.10 82.80 11 0.32 C 6 A 2 F 44.44 13.49 42.07 25.30 17.17 40.60 80.80 12 0.10 C i 5 A F 6 44.30 5.38 50.32 9.50 7.89 16.10 39.30 (*) daN=10 x IN Ferrite samples were fabricated at 1250 °C [9, 45], and were air cooled. The study defines the ratio A l 2 0 3 / % F e 2 0 3 as the ferric module (M Ai). X-ray diffraction analysis indicated the presence of C 3 A in ferrites with high MAI (1.5-1.9), and of C 2 F in ferrites with low MAI (0.1-0.64). An interesting observation which coincides with the research done by Conjeaud et al [9], is that the Section 2.5.1 - Ferrite Glasses Present in Portland Cement Clinker and in Metallurgical Slags 31 ratio between the glassy phase and the crystalline phase gets higher with an increase in the ferric module. Compression strength is higher for samples with an increased ferric module, thus with an increased presence of glassy phase. However, the strength diminishes for a high amount of C 3 A , probably due to the hydro-aluminates transformation from hexagonal to cubic in the hardened paste. The products of ferrite hydration are hydro-aluminoferrites and hydro-ferrites. Another interesting feature is that the setting time decreases with the increase in the ferric module and therefore with the amount of glass present. Not only is the glassy phase more reactive, but it also offers other benefits such as its capability of bearing an increased amount of MgO. This could be seen in the case of blast furnace glasses where MgO becomes non-detrimental for the hydration of slag cement [42], and avoids an undesirable expansion. One of the conditions for obtaining a sound clinker is to have less than 5 % MgO [16, 17]. This stems from the fact that in clinker, MgO separates as periclase (MgO), with a remainder present in solid solution with the calcium silicates and in the vitreous phase. Periclase is known to hydrate slowly forming Mg(OH) 2 while undergoing a 120 % increase in volume and causing unsoundness in the cement paste [18]. However, the behavior of magnesium is different in blast furnace slag. A composition of 15-20 % MgO in this type of slag does not produce these effects, since magnesium is not present as periclase, but as a constituent of the vitreous system [42]. Synthetic glasses of composition similar to blast furnace slags, hydrated in the presence of saturated lime solution for 28 days give hydration products of solid solution types such as: (C ,M) 4 AH X , C H and H G (hydrated gehlenite). The presence of Fe in the (C,M) 4 A aqueous phase complicates the issue. Fe 2 + is able to substitute for C a 2 + in different ways in the hydrated calcium containing phases, while Fe 3 + can substitute for A l 3 + in (C,M) 4 A and C 2 A S H 8 phases. The green color of the hydrated sample could indicate that Fe 2 + is present as ferrous hydroxide. Mascolo [42] compares synthetic glasses Section 2.5.1 - Ferrite Glasses Present in Portland Cement Clinker and in Metallurgical Slags 32 with industrial blast furnace slags indicating that the industrial slags give the same hydration products, but in lower quantity. He also concludes that the quantitative difference can be attributed to a partial devitrification of the industrial slag, occurring during its storage, with a consequent lower activity. Thus, it can be inferred that a slag with higher amount of glassy phase would display a higher activity. Information on the hydration process of ferrite glasses is scarce. However, it appears that the research done in this domain recognizes the importance of the glassy phase in the hydraulic properties of mortars. How and why the glassy phase is hydrating are aspects that need to be clarified and tailored to each specific type of glass, since the rates of hydration and products of hydration are going to vary according to the composition, and also to the degree of crystallinity. Section 2.5.2 - Defects and Unstable Atomic Arrangement in Granulated Slags 33 2.5.2 Defects and Unstable Atomic Arrangement in Granulated Slags Since granulated slags possess a disordered atomic arrangement, it is important to mention findings related to the influence of defects or unstable atomic arrangements on the hydration process. However, the literature only provides information for two of the main cementitious compounds present in PC, namely the C 2 S and the C3S. In comparing the rates of hydration of y-C2S to p-C 2S, it was found that y-C2S has a much 0 9-4-slower rate of hydration. One explanation for this is that the arrangement of O " around Ca is regular in y-C2S, but irregular in the p form. This implies that thermodynamic factors influencing the hydrolysis of this silicate, namely its lattice energy and the heats of hydration (solvation) of its constituent ions, are of primary importance. However, it looks more probable that rate factors are predominant in inhibiting hydration, since y-C2S hydrates almost as rapidly as P-C 2S, if the pH is prevented from rising by frequently replacing the aqueous phase with fresh water. The reactivity of C3S can be enhanced by rapid cooling [11], thereby shortening the dormant period. Fierens and Verhaegen (1976) detected defects (unspecified) in the lattice of samples of C3S by thermoluminescence spectroscopy and linked the increased defect concentration with an increased cooling rate (between 1300 and 1100°C) and with shortening the dormant period. They also found that defects could be produced by exposing C 3 S to ultraviolet radiation before submitting it to hydration. Also, they suggested that there is no protective layer of first hydrate, water being chemisorbed at this stage by interaction with surface defects in C 3 S, and that the length of the dormant period depends on the probability of C-S-H nucleation, which increases with the increasing defect concentration. Section 2.5.2 - Defects and Unstable Atomic Arrangement in Granulated Slags 34 In a silica glass, the tetrahedra are linked to produce a random network, with large numbers of disrupted silicon-oxygen bonds. Silicon and oxygen atoms exposed at surfaces and discontinuities will provide the sites for reaction. The reactivity of a given type of silica thus will depend on the degree of crystallinity (or the degree of its structural disorder) and the number of interrupted silicon-oxygen-silicon (Si-O-Si) bonds [44]. With some experimental observations still in dispute, there is still uncertainty about the role of defects presence in the cementitious phases. Although cement specialists believe there is a correlation between the existence of a metastable arrangement and the hydration rate in slags, there are no clear reports or data on this issue, and consequently even less information for the specific case of chemically altered, and further quenched, steel slags. Chapter 2.6 -Hydration of Portland Cement 35 2.6 The Hydration of Portland Cement The hydration of Portland cement pastes may be divided into 3 stages. The 1 s t stage starts with a fast hydration until 10-15 % of cement is hydrated (pre-dormant period), followed by a very slow hydration rate, caused by the formation of a coating on the cement grains (dormant period). When 15-20 % of cement is hydrated, the coating is ruptured and a fast reaction starts, which lasts until about 30 % of the cement is hydrated. Then diffusion through the very narrow pores between the hydration products becomes the controlling step [46]. This is the 2 n d stage of reaction. In the 3 r d stage, the hydration slows down due to the accumulation of hydration products, but it continues. It is unclear whether the diffusion of water through the hydration products to the unhydrated grains is the slowest and therefore rate-determining step, or if it is the diffusion of the hydrated ions from the surfaces of unreacted grains through the hydration products to water filled pores in which they can precipitate and form new hydration products. The formation of hydration products is accompanied by the formation of small pores and hence the term "diffusion through the hydration products" is used to refer to the diffusion through pores, and not solid matter. As the hydration progresses, the pores become smaller and the diffusion becomes slower. The hydration process is accompanied by the development of a cement paste structure. 2.6.1 Development of Structure in Portland Cement Pastes When water is added to cement, a super-saturated solution of Ca(OH)2 is formed as a result of the hydrolysis of the calcium silicates. Sulphate and alkali ions, small amounts of silica, alumina and ferric iron are also present in the solution. Ca(OH)2 and C6AS3H32 (ettringite) precipitate out and a dense C-S-H gel coating is formed on the cement grains. This coating Chapter 2.6 -Hydration of Portland Cement 36 retards the hydrat ion and expla ins the existence o f the dormant per iod , last ing 1-2 hours, dur ing w h i c h the paste remains plast ic and workable . T h e end o f the dormant per iod , and the in i t ia l set, are attributed to the C - S - H coat ing break up and the cont inuat ion o f hydrat ion. There are several explanat ions h o w this occurs [47] . The force caus ing the break up of the coat ing is the osmot ic pressure due to the difference between the ion concentrat ion in solut ion at the gel -cement interface and that in so lut ion outside the coat ing. An oth er explanat ion is that the coat ing disintegrates s i m p l y because of the decompos i t ion o f the unstable hydrate. T h e v o l u m e of the hydrat ion products is more than twice that o f the anhydrous cement. Hence , as the hydrat ion proceeds, the hydrat ion products gradual ly f i l l spaces between the cement grains. Points o f contact are fo rmed caus ing the st i f fening o f the paste. A t a later stage, the concentrat ion o f hydration products and the contact points restrict the m o b i l i t y o f cement grains to such an extent that the paste becomes r ig id , and the f ina l set is attained. A schematic d iagram representing the hydrat ion process and the format ion o f the paste structure is shown in F i g . 6. Cement grains appear as shaded areas, Ca(OH)2 crystals as hexagonal shapes, ettringite needles as heavy l ines and C - S - H particles as thin ones. D u r i n g the dormant per iod cement grains are separate and the hydrat ion products are m a i n l y Ca(OH)2 and ettringite. A f t e r an hour, the C - S - H gel begins to f o r m , w i th particles in the shape of l o n g f ibres. Intergrowing w i l l cause st i f fening, w h i l e the increased vo lume of sol ids w i l l decrease the porosi ty o f the paste. Chapter 2.6 -Hydration of Portland Cement 37 Age: Mhutes Hours Days Dormant period Setting Hardening Fig . 6 - Schematic representation of the hydration and structure development in cement paste [47]) 2.6.2 Setting and Hardening B y mixing cement with water, a plastic workable paste is formed. For some time, the characteristics of the paste remain virtually unchanged (dormant period). A t a certain stage (initial set) the paste begins to stiffen to such a degree that, although still soft, it becomes unworkable. The paste continues to stiffen until it can be regarded as a rigid solid (final setting). Chapter 2.6 -Hydration of Portland Cement 38 The resulting solid is known as the hardened cement paste or cement stone. The hardened paste continues to harden and gain strength, the process being known as hardening. The various stages of setting and hardening are shown in Fig. 7. • Add i t ion of wate r » . g P l a s t i c a n d w o r k a b l e paste - I n i t i a l se t -S t i f f and unworkable paste • F i n o l set-R i g i d s o l i d ga in ing s t r e n g t h wi th t ime - E E O Fig. 7 - Schematic description of setting and hardening of a cement paste [47] 2.6.3 Heat Evolution During the Hydration Process When Portland cement reacts with water, heat is evolved and, with an appropriate conduction calorimeter, the rate of heat liberation can be followed isothermally as shown in Fig. 8. The complexity of the hydration process is apparent from the three peaks in heat liberation, which imply there are three maxima in the rates of the hydration [10, 12, 48]. The first one is the highest, but of short duration, followed by one in which the heat-liberation rate is relatively low, although never zero. This dormant period ends with an acceleration in the heat release producing a broad second peak with a maximum at approx. 9-10 hours. Sometimes a third peak is detected, Chapter 2.6 -Hydration of Portland Cement 3 9 however, this is not a characteristic for all types of cements and the peak's height varies considerably from one sample to another. After three days, heat evolution rates are low and the heat of hydration is determined usually by a heat of solution method, in lieu of calorimetric techniques. 4rr i 2oo at 11 3 0 seconds O - i o 2 0 3 0 4 0 Time / hours Fig. 8 - Typical curve of heat evolution in the hydration of ordinary Portland Cement [12] In order to interpret the reactions occurring in the hydration of cement and link them to setting and development of strength in the hardened paste it is necessary to understand: - how the hydration reactions of the individual compounds in cement contribute to the heat peaks; - the causes of the considerable changes in heat-evolution rate; - the nature of the bonds between the hydration products in the hardened paste. Generally speaking, the hydration of the cement may involve either a 'through solution' mechanism or direct topochemical (solid state) reactions. In the first mechanism the reactants Chapter 2.6 -Hydration of Portland Cement 40 dissolve to produce ions in solution. The ions then combine and the resulting products precipitate out. In cement, due to the low solubility of its constituents, hydrolysis rather than dissolution is considered to be significant. In the second mechanism the reactions take place directly at the surface of the solid without the cement constituents going into solution. Hence, reference is made to topochemical or solid state reactions. In the hydration of cement, both mechanisms are probably involved. It seems that the first mechanism is involved in the early stages, and the second operates during the later stages. As cement is made up of several compounds, its hydration involves a number of chemical reactions which take place simultaneously. The products of hydration of the individual compounds are similar to those resulting from their hydration when forming part of the cement. It is convenient, therefore, to consider the hydration of individual compounds as well the hydration of the cement as a whole. The present state of knowledge of the mechanisms and the rates of hydration of the four principal cementitious phases of Portland cement are further discussed. 2.6.4 Hydration of Individual Phases in Portland Cement Tricalcium silicate. This compound has been widely researched. Its hydration is similar to the hydration of alite, which is an impure form of C3S encountered in cement clinker. The products of hydration, at ambient temperature, are calcium silicate hydrate (in which the C/S ratio is lower than in the hydrous silicate) and calcium hydroxide. The reaction can be written as: 3CaO • Si02 +(y + z)H20 = xCaO • Si02 yH20 + zCa(OH)2 Chapter 2.6 -Hydration of Portland Cement 4 1 where x+z=3, but x, y and z are not necessary integers. Employing the nomenclature commonly used in the cement industry (Table I), the reaction becomes: C 3 S + (y+z)H20 -> CxSUy + zCH The value of x can be determined by indirect analysis involving determination of calcium hydroxide by chemical extraction and by direct analysis involving the determination of the x-radiation wavelength excited by a beam in an electron microscope. However, y is difficult to determine because there is no distinct break in weight loss between drying and dehydrating this compound. The calcium silicate hydrate formed by the hydration of C 3 S is poorly crystalline [10, 12, 39, 47], yielding only a few broad, weak bands in its X-ray diffraction pattern. This, and its uncertain composition, resulted in the general use of the notation C-S-H to represent it, where the hyphens emphasize that the composition is indefinite. C-S-H is described as a gel or a xerogel, this being the solid phase formed from a gel when removing the water from it. The three main x-ray diffraction bands correspond to the mineral tobermorite (C5S6H5) and thus, C-S-H is called tobermorite gel. It consists of imperfectly aligned and possibly rumpled tobermorite-like layers which may be expanded or contracted normal to the layer plane by addition or removal of water from the interlayer spaces. Four types of C-S-H gels have been reported in the literature. They are almost amorphous and have different morphologies. However, the nature of the atomic arrangement and composition still remains an open question [49]. According to Taylor [11] the small degrees of order in C-S-H may reside in its Ca-0 parts. Lately, some similarities of the C-S-H gel with another type of mineral, jennite, have been reported [50]. Calcium hydroxide, or portlandite constitutes 20-30 % of the solid content of fully reacted Portland cement or C 3 S pastes. Generally, it is regarded as a crystalline compound, although some researchers consider that amorphous, as well as crystalline C H , is formed in Chapter 2.6 -Hydration of Portland Cement 42 cement and calcium silicate pastes, since there is a discrepancy between determinations by x-ray diffraction and extraction methods [10, 40]. CH's X-ray peaks dominate the powder diffraction patterns of fully hydrated pastes. Under the microscope, C H crystals appear as plates or prismatic shapes rarely exceeding a few micrometers and may be intergrown with C-S-H. When C3S hydrates, the orthosilicate ions (isolated Si04 tetrahedra) are initially converted to disilicate ions (Si207 6") and subsequently to higher polymeric forms [51, 52]. The degree of polymerization of the silicate ions is generally examined by trimethylsilylation (TMS) of the hydrated pastes [12]. Silicate units are hydrolyzed and then end-blocked by trimethylsilylation to prevent polymerization or rearrangement. The trimethylsilyl derivatives, which pass into solution, are then analysed by chromatographic and/or spectroscopic methods. These are standard methods used in organic or polymer chemistry. The analysis of complex silicates requires additional research work, but the following facts have been already established. Hydration involves the polymerization of the monomeric (SiO/") silicate units originally present in anhydrous calcium silicates, first to dimeric (Si2C>76~) and then to polymeric silicates. The polymeric silicates are chains of five to eight silica tetrahedra linked together. For C3S pastes, hydrated at room temperatures, the amount of monomer corresponds approximately to the amount of unreacted C3S found in X-ray diffraction. Little monomeric silica is formed in C-S-H, thus the decrease in amount of monomer can be used as a measure of degree of hydration. The polymerization of Si04 4" tetrahedra can be also observed in C2S and in cement pastes [10]. Fig. 9 shows the development of polymers in a C3S paste. The process of polymerization runs parallel with the increase in strength. It is not clear, though, whether the polymerization of the anions causes this change. It has been suggested that Si-O-Si links are formed between C-S-H particles giving a stronger material. On the other hand Chapter 2.6 -Hydration of Portland Cement 43 this increase in strength might be due to microstructural changes rather than molecular. Research is still needed in this area. " T I M E (lot) K a l e ) day* y « o r t Fig. 9 - Distribution of Si among silicate anion types in C3S pastes [10]. Dicalcium silicate. The hydration of pure P-C2S is similar to the impure form found in clinker, also called belite. Dicalcium silicate may come in four polymorphic forms: a, a ' , P and y. The P-C2S represents the main hydraulic form. This is explained as due to the fact that oxygen atoms surround Ca ions in an irregular form in P-C2S, but in a regular form in the other polymorphs [12]. The P-C 2S displays an extremely low hydration rate and the products of hydration are C-S-H and C-H (portlandite). At 40 days, the rate of hydration reaches a maximum (20% hydration) and unlike C3S, it exhibits a second acceleration period starting after 60 days, suggesting that scaling of the hydration products has occurred. Therefore, C 2 S contributes little Chapter 2.6 -Hydration of Portland Cement 44 to peak II (Fig. 8) in the hydration of Portland cement and to the early strength development, but has a significant contribution to the late strength development. The hydration of alite and belite in cement is essentially similar to the hydration of pure C3S and C2S. The presence of impurities in alite and belite affects the composition and properties of their hydration products. Nevertheless, it may be generally assumed that their hydration products are a C-S-H gel of average composition C3S2H3 and the C H . Alite and belite make up 70% of cement, so the set cement will consist mainly of their hydration products and its properties will be determined accordingly [47]. The presence of C H makes cement pastes highly alkaline (pH 12.5), hence sensitive to acid attack, with the advantage of protecting steel against corrosion. Tricalcium aluminate. C3A reacts rapidly with water forming crystalline hydration products with different lime to alumina ratios. 2 C 3 A + 21H 2 0 -> C4AH13 + C 2 A H 8 The products appear as platelets with hexagonal symmetry, resembling portlandite. They are metastable with respect to cubic hydrogarnet, to which they are rapidly converted above 30 °C, a temperature easily reached due to the heat liberated in the hydration process [47]. If lime is present in solution, the formation of C4AH13 is favored. In cement, C 3 A hydrates differently because of the presence of gypsum. The reason for gypsum addition is to delay cement setting by forming an ettringite layer on the surface of the tricalcium aluminate grains. Calcium aluminoferrite. The hydration products formed by the ferrite phase, JCC2A-(1-X)C2F, are similar to those formed by C 3 A , with A l 3 + ions partially substituted by Fe 3 + ions. The reactivity of ferrite solution increases with increasing values of x. The ferrite Chapter 2.6 -Hydration of Portland Cement 45 phase is deficient in lime relative to its hydration products [12]. If there is a lime supply, as when C3S is present in cement, then the products of hydration are the iron-substituted hydrates (C 4(AP)Hi 3) and the hydrogarnets (C3AH6). If there is a low lime content, hydrous ferric oxide will be the hydration product. In Portland cement, the ferrite phase reacts with gypsum and Ca(OH)2 to produce needle-like crystals of a solid solution consisting of a sulphoaluminate and sulphoferrite [47]. The hydration of the aluminoferrite phase is complicated by the formation of a ferric hydroxide gel, and by other elements present in the PC interstitial phase, with the result that the overall system is not entirely understood. The composition of the amorphous ferric hydrate is unknown [53]. Under the conditions prevailing in cement pastes, ferric iron is very insoluble and cannot migrate very far, which may account for the low tendency of these amorphous phases to recrystallize into recognizable hydrates [54]. The low solubility of these hydrates may be responsible for the low rates of setting and hardening of iron rich cements, especially if the iron rich hydrates can impede hydration of the silicate phases. A cement paste moist cured for 23 years contained a hydrogarnet phase of approximate composition C2Ao.6Feo.4SH4, which suggests that this might be the ultimum equlibrium hydration product [54]. Some indirect evidence has been found indicating the formation of an alumina-rich phase, possibly C 2 A H 8 or A H 3 , at very early ages of C 4 A F hydration, preceding the formation of the AFt phase (in the presence of lime-gypsum solutions). This would be consistent with the mechanism proposed by Tang and Gartner [55], which assumes that an unknown, probably amorphous, C A S H phase is the initial retarding layer. This phase may also include ferric hydroxide and it could be a transition layer between the underlying aluminate phases and the ettringite coating. Chapter 2.6 -Hydration of Portland Cement 46 The heat evolved by the initial reactions of the interstitial phase usually dominates the heat-evolution profile of cements in the 1 s t hour or so, and can therefore be studied by calorimetric techniques in many cements, without significant overlap with heat output from the C3S first stage hydration. The major phases in Portland cement may be considered to hydrate independently [54, 55] and the interactions between phases can only be described qualitatively. The hydration of C3S is accelerated by soluble sulphates and the lime produced influences the rate and course of hydration of C3A and C4AF. C3S makes the major contribution to peak II (Fig. 8). The hydration of C 3 A to form ettrigite also makes a significant contribution to the second peak, but if the amount of sulphate ions is reduced, then peak HI appears. The role of the ferrite phase is less well established. It is suggested that its contribution to peak II is small, partially because its heat of hydration is lower than C 3 A. The preceding discussion and calorimetric results suggests a mechanism, which is a combination of ideas previously developed by Gaidis and Tang [54, 55]. The initial reaction of the clinker interstitial phase occurs at a rate determined by the intrinsic "reactivity" of the aluminates and ferrites. The initial product is a hydrate on or close to the grain surface, incorporating various species, including anions and organic molecules from the surrounding aqueous phase. This incorporation can occur between cationic sheets of [Ca2(Al, Fe)(OH)6]+ compositions. Growth of the hydrate layer leads to a decrease in the rate of hydration of the underlying interstitial phase as the coating thickens and spreads over the surface. Later the hydration of the silicate phases (primarily alite) follows, with the nucleation and growth of C-S-H, as discussed in a previous paragraph. Chapter 2.6 -Hydration of Portland Cement 47 2.6.5 Factors Influencing Cement Hydration Cement Composition The rate of hydration of the individual cement constituents varies considerably. C3A, for example, reacts with water almost instantaneously and most of its hydration tales place within 24 hours. C2S, on the other hand, reacts slowly and its hydration continues for weeks and months. Fig. 10 shows the degree of hydration for the four main constituents in Portland cement. Age, days Fig. 10 - Degree of hydration for cementitious compounds in Portland cement [47] After 24 hours, 65% of C3A is hydrated as compared to 15% of belite. During its early stages, the hydration is selective, i.e. the degree of hydration of cement is determined by the degree of hydration of the individual compounds. Such a selective hydration is to be expected because at this stage water is virtually in direct contact with all the cement constituents. Consequently, each compound hydrates independently of the others at its characteristic rate. However, with time, a layer of C-S-H gel is formed around the cement grains. As the thickness of the layer increases, the rate of hydration is dependent on the rate of water diffusion through the Chapter 2.6 -Hydration of Portland Cement 4b layer, and less on the rates of hydration of the individual compounds. Under such conditions, it may be expected that hydration will penetrate cement grains at an equal rate, and also all cement compounds will hydrate at the same rate, i.e., when cement is 50% hydrated, each one of the constituents is also 50% hydrated [47]. The rate of hydration increases with temperature (Fig. 11), particularly in the early stages, provided that the rise in temperature does not cause the drying of the paste. Such drying might slow down or even stop the hydration. However, the final degree of hydration is apparently not affected by the curing temperature [47]. The degree of hydration is controlled by the density and thickness of the C-S-H layer which encapsulates the cement grains. This layer retards hydration and when it attains a certain thickness, it prevents hydration from taking place altogether. The fact that the degree of hydration is independent of temperature, implies that temperature does not affect the thickness and density of the C-S-H gel layer. Nevertheless, there are still scientific debates on this issue. Fig. 11 - Effect of temperature on the degree of hydration as a function of curing temperature Temperature Curing time, hours [47] Chapter 2.6-Hydration of Portland Cement Fineness of Cement The hydration rate increases with fineness of the cement. The finer the particles, the greater the surface area exposed to water and consequently, the higher the rate of hydration, particularly at early ages. This is of utmost importance in the production of rapid-hardening Portland cement. However, the ultimate degree of hydration is not affected by the fineness of cement [47]. Water/cement Ratio Some experimental data are presented in Fig. 12. It can be seen that initially, the w/c ratio does not significantly affect the rate of hydration [47]. Later, however, the rate of hydration decreases, and this decrease will occur earlier if the w/c ratio is lower. Hence, the lower the w/c ratio, the lower the average hydration rate. This effect was attributed to the reduction in space available for the hydration products. Curing time, hours Fig. 12 - Effect of w/c ratio on the hydration of Portland cement [47] Chapter 2.6 -Hydration of Portland Cement 50 Age of paste The rate of hydration of all cements decreases with time until a stage is reached and hydration stops. Hydration is conditional on the presence of water; however, hydration may eventually stop, leaving behind more than 50% unhydrated cement, even with sufficient amounts of water (i.e. when paste is immersed in water). The decrease in the hydration rate may be attributed to the existence of a thick C-S-H layer. It is expected that at a certain thickness of the layer, diffusion and hence hydration may stop completely. Theoretical calculations indicate that hydration is expected to stop when the thickness of the layer reaches 25 pm. Accordingly, an unhydrated core is always left inside cement grains which have a diameter greater than 50 pm. This fact partly explains the restrictions imposed on the coarseness of cement by several standards. A minimum specific surface area is required in order to reduce the amount of cement left unhydrated. In general, the size of cement grains in ordinary Portland cement ranges from 5 to 55 pm [47]. For slags, some standards suggest a particle size having max. 20 % retained on the 45 pm sieve, or a specific surface area of min. 2750 cm2/g [56]. Admixtures Gypsum is commonly used in Portland cement as a retarder. Numerous materials can be added to PC to retard or to accelerate the setting of cement. More details on this topic are presented in the chapter discussing chemical admixtures. Chapter 2.6 -Hydration of Portland Cement 51 2.6.6 Kinetics of the Hydration Process The progress of hydration can be observed by using a conduction calorimeter and data can be obtained for further calculating kinetics parameters. For example, an apparent activation energy (perhaps better described as a temperature coefficient of the hydration rate) can be derived for Portland cement hydration. The literature on the kinetics of reaction deals extensively with the concept of activation energy and the corresponding Arrhenius equation. An energy barrier must be overcome in order for a reaction to proceed. Fig. 13 shows the distribution of energies at a certain temperature. Fig. 13 - Schematic diagram showing the activation energy and the thermodynamic driving force for a reaction. [40] The fraction of the particles possessing sufficient energy to overcome the activation barrier is given by the shaded area. The chance of this occurring is given by the Boltzman law, which expresses the energy distribution of particles in a substance. The law was originally applied to Chapter 2.6 -Hydration of Portland Cement 52 the kinetics of reactions in vapour phase, but is frequently assumed to operate for the constituent particles (ions, atoms, molecules) of solids, and may be stated as: Ni=N/Z-exp(-Ei/KT) (1) where N=number of constituent particles, N,=number having a specific energy Ei, K=Boltzman constant, Z=distribution function, T=absolute temperature. The equation can be written in many different forms, depending on the nature of the problem, but for a Boltzman distribution there is always a factor of the form exp (-Q/RT). The reaction rate constant, k, is expressed as: fc=A-exp-(-Q/RT) (2) where Q is the activation energy of the chemical reaction. The most common method for determining Q from experimental data is to assume it is independent of temperature. The reaction given above can be converted to log form, giving: aQ \ogk = b - ^ = - (3) where a=log e/R= 1/4.575 and b=log A (units are gram calories), A=P-Zo (P=probability factor and Zo=number of binary encounters leading to reactions between particles). The standard method is to plot log k against 1/T. The slope will give Q, the activation energy. More information on this subject is presented in the chapter dealing with the Energetics of Hydration. Chapter 2.7 - The Hydration Process in Glassy Slags 53 2.7 The Hydration Process in Glassy Slags After discussing the PC hydration a review of the literature on glassy slags is in order. The hydration of glassy blast furnace slag requires breaking of the bonds and the dissolution of the three-dimensional network structure of the glass. The network contains interlinked S i 0 4 and A10 4 tetrahedra and A106 octahedra which are attacked by the OH" in a high pH environment (~ 12). In order to maintain the hydration of the slag active, it is necessary to supply additional hydroxyl ions, as the amounts of hydroxyl ions and calcium ions derived from the slag alone are not sufficient. A second factor is the formation of an appropriate hydrate structure on the surface of the slag, which also assists in further promoting slag hydration. The conditions in which the hydrate forms, its morphology and crystallinity depend largely upon the concentration of the hydroxyl ion [57]. Some slag hydration takes place immediately after mixing with water and a protective layer is formed on the surface of the slag, which inhibits the water penetration to the slag particle and further dissolution of the slag. However, Nurse [16] noted that slags, in general, are not very reactive when in contact with water and that the addition of activators is required. Based on the theories regarding the structure of active components of PC clinker, (primarily C 3 S and C 3 A ) and the quasi-crystalline state of glassy slags [18], the hydration mechanism in glassy slags, when using Ca(OH)2 as an activator, is the following: the C a + 2 cation, from the ionized Ca(OH)2, alters the quasi-crystalline structure of the slag making its mineral composition similar to that of tricalcium compounds (silicate and aluminate) in Portland cement. This replacement mechanism is a surface phenomenon that allows for the dissolution of atoms in the surface layer of the slag. The consequent hydration reaction will form insoluble salts precipitating from solution, such as tobermorite, ettringite, calcium aluminium hydrate, etc. Chapter 2.7 - The Hydration Process in Glassy Slags 54 The extent of slag hydration can be measured by several methods; e.g. (i) dissolution of the hydrated products in organic acid solution, (ii) measurement of non-evaporable water, and (iii) determination of the heat of hydration [10]. The following sections will look at glassy slag hydration in more detail, and draw comparisons with the hydration mechanism in Portland cement, pure cementitious compounds, blends of slag-cement, fly ash and silica fume. 2.7.1 Blast Furnace Glasses Glass content in granulated blast furnace slag usually varies between 90 and 100 % [58]. In the system Si0 2 -Al 2 03-CaO, at 10% MgO, the chemical composition of the glassy slag lies in the mellilite area, C 2 A x M i . x S 2 . x , with x varying between 0 and 1. Gehlenite and akerminite also belong to this area, but phases with mellilite composition appear to display the highest hydraulic activity. Furthermore, it was observed that hydraulicity is higher when the content of two specific glass modifiers increases, namely A l and Mg. A larger percentage of these elements increases the disorder in the glass, and promotes dissolution. Structural defects of the glass are also believed to enhance its reactivity [11, 43]. Pietersen [53] showed that the chemical shift in the M A S - N M R spectrum of slag is broader, reflecting a disordered structure. Moreover, the glassy blast furnace slag spectrum is similar to the Portland cement one (Fig. 14). Fig. 14 - Si N M R spectra of Ordinary Portland cement (left) and granulated blast furnace slag (right) [58] Chapter2.7 - The Hydration Process in Glassy Slags 55 Glass composition and reactivity The reactivity of mineral admixtures is associated with the rate of dissolution of the admixture, but more often is related to the strength of concrete and mortar. Although the latter is a practical way to assess the efficacy of the admixture, it is not a real measure of the chemical reactivity. Moreover, physical or physicochemical effects, such as a better dispersion of cement particles into water and an increased cement reactivity due to admixture particles acting as crystallization nuclei, are also contributing to the overall reactivity. Considering that the dissolution is an important step in the hydration of admixtures, several factors influencing this process are subsequently mentioned. Most of the past research deals with fly ash dissolution, nevertheless, knowledge gathered in this area could be useful in explaining the dissolution of slag grains. Alkalinity -It has been established that the dissolution of fly ash and slags is strongly pH dependent. This means that the alkalinity of the pore water is of extreme importance [58]. The pore water composition of cement paste with fly ash and without fly ash was investigated by squeezing cement pastes at various stages of hardening. The OH" concentration decreases when fly ash is added, and becomes even lower when inert crystalline material (e.g. crystalline quartz flour) is present ( see Fig 15). In the early stages of cement hydration the composition of the pore water may be regarded as a saturated lime solution with some sulphate (gypsum). The pH of the pore solution is less than 13. After lweek, the lime and sulphate concentrations decrease to a very low level, and the concentration of hydroxyl increases rapidly to a pH exceeding 13. Consequently, a significant dissolution of fly ash will occur and in time a further increase of pore water alkalinity will take place. This phenomenon explains the incubation period for the pozzolanic reaction of fly ash. Chapter 2.7 - The Hydration Process in Glassy Slags 5 6 For silica fume, such incubation is not present, because the pH is above 12 from the beginning. Likewise, blast furnace slag is activated directly upon mixing as the pH is approx. 12. • • d a y s time (h) Fig. 15 - The development of the OH" concentration in the pore water of Portland cement paste with fly ash and fine quartz flour (T=20 °C, w/mix=0.45)[58] For most pozzolans, a decrease in alkalinity was observed at later stages. This is attributed to the hydrolysis of the glass network at later stages when the polymerized silica tetrahedra are broken, with the consequence that an increased alkali content can be detected in the C-S-H gel. As previously described, the effect of slags on pore fluid composition is a complex issue. Nevertheless, most researchers have focused on the pH. However, in slags, due to a high CaO/Si0 2 ratio, the pH is well in excess of the minimum 10.5 suggested in the literature [58]. The presence of alkalis will further raise the pH. Moreover, the main buffering action arises from the pair Ca(OH) 2 and C-S-H. The pH is not dependent on the amounts of these phases, but on their continued presence in contact with the pore fluid. Hence, it is not important that slags consume Ca(OH) 2, because one product of hydration is C-S-H, which can also buffer a high pH. However, blast furnace slag-cement develops a lower hydroxyl concentration compared to Portland cements [58], therefore being less reactive. Chapter2.7 - The Hydration Process in Glassy Slags 57 Temperature - Temperature influences the development of hydroxyl concentration and also it enhances the activity of the fly ash, with the result that at higher temperatures, a pozzolanic reaction may be initiated at lower alkalinity [58]. For Portland cements, an increase in temperature will result in a more permeable concrete due to the development of a coarser pore structure. For cements with mineral admixtures, like fly ash, an increase in temperature could have a small, or even adverse effect, decreasing the hydration rate. This could be due to an enhanced pozzolanic activity at slightly elevated temperatures, that would result in an increased amount of C-S-H-gel precipitating in the pore system. Water/cement ratio - The alkalinity increases with w/c ratio. However, at low w/c ratio (e.g. in high strength concrete) fly ash is more effective than at high ratio. It has been observed that, in general, slags need less water than Portland cement. The slag/water ratio could vary between 0.4-0.5, without affecting the hydration rate. 2.7.2 Hydration Mechanism in Slags and Slag Cements Cements containing slag additions are denominated slag cements. A classical case is blast furnace slag cement. It is generally considered that the properties of concrete at large, especially those related to durability, are improved by the incorporation of glassy blast furnace slag into the cement. The disadvantages of blast furnace slags cements are lower early strength and longer setting time. However, these features can constitute an advantage for some applications, or else could be modified by certain activators and accelerators. A discussion of the mechanism of hydration in such cements is subsequently presented. Chapter 2.7 - The Hydration Process in Glassy Slags 5 8 Many researchers have followed the continuous hydration for 1 or 2 years and noted that after 6 months to 1 year the hydration slows greatly. As opposed to studies showing the cessation of hydration at one point in time, Glasser [60] noted that the residual glass will continue to exist in a metastable state and will react with water as long as water is present in the paste. However, the hydration rate is expected to decrease with time. At the reaction interface, the specific surface area will diminish with time and the glassy slag grains will become coated with reaction products, which will tend to act as passivating layers. Complex structures develop when Portland cement is used as an activator of slags. Fig. 16 and Fig. 17 show the progressive development of the microstructure around a grain of slag. The initial stage is represented by a 10-40 pm slag grain which comes in contact with water. The next stage, normally attained within 3 to 28 days, is characterized by the development of a gel-like layer. This layer forms rather rapidly and is associated with the initial calorimetric heat evolution. Its morphology is featureless, but it appears to be significantly denser than the normal cement hydration products and has a higher Si content. To distinguish it from the lower density cement products, it is termed inner slag hydrate or ISH. As the ISH thickens, it loses Ca, A l and Si relative to the concentration of these elements in the glassy slag. Magnesium however, is retained and is concentrated. The composition and microstructure of the ISH slowly evolves with time leading to the development of a zone characterized by incipient crystallization, along its outer margins, where a hydrotalcite-like phase and a second phase, rich in Fe and Ti , are precipitated. The crystallinity tends to improve with time. The innermost portions of the ISH zone are constantly replenished from the slag glass, but at a slower rate, as the residual glass core contracts in size. The ISH slag interface thus tends to exhibit less differentiation at its outer margin, where it is in contact with the products of hydration. Chapter 2.7 - The Hydration Process in Glassy Slags 59 Fig. 16 - Schematic showing the progressive development of microstructure around a slag grain in cement paste. [60] Fig. 17 - Microscopic zoning of a partially hydrated slag grain in cement [60] Chapter 2.7 - The Hydration Process in Glassy Slags 60 While these changes occur in the ISH, the cement hydration products (CHP) also are influenced by material transport and the CP H morphology takes on a radial structure. Eventually (bottom of Fig. 16) the slag grain is completely consumed, leaving behind a porous network of hydrotalcite platelets and a Fe-Ti phase. This region is in turn surrounded by a dense shell of C-S-H and AFm. The orientation of platelets gives rise to a "house of cards" morphology. One feature observed in the hydration of coarse grains in particular, is a rhythmic accumulation of the hydration products. There is a repeating pattern forming bands that can be compared to the annual growth rings of a tree, although the analogy is not quite correct because pastes are cured isothermally. Interesting enough, no two grains appear to be identical in terms of the relative thickness of individual zones, nor the number of repetitions. These periodic structures are related, probably, to the phenomenon known as Liesegang rings [60]. The latter are typical where an aqueous solution is placed in contact with a gel. Precipitation occurs along the contours of diffusion gradients, often in a rhythmic way within the gel at places where it becomes locally supersaturated. The "two-reservoir" model explains this phenomena. In slag-cement blends, one reservoir is represented by the core of glassy grains, the other by the porous cement paste and its associated pore fluid. The most important concentration gradient is that of water, which is abundant in the cement hydration product, but absent in the slag. In addition, other chemical potential gradients occur; for example: Mg, Ti and S2" may be enriched in the slag, while Ca and SO4 2 " will be enriched in cement. However, each species has its own characteristic diffusion properties and the overall pattern of diffusion occurs in response to the drive to lower the free energy of the reacting system, comprising cement, slag and pore fluid. Chapter 2.7 - The Hydration Process in Glassy Slags 61 Cements have too little SiC>2 and AI2O3 to form dense and impervious gels, but slags have higher contents allowing this to happen. Thus, the diffusion membrane develops spontaneously around slag grains. The Liesengang rings occurring in the slag hydration product, can be detected by S E M , and are made prominent by the precipitation of Mg, A l , Fe, Ti , and Mn oxides. Attempts have been made to observe the extent of reaction and spatial distribution of the reaction products by immersing the samples in Ca(OH)2 solution, or embedding them in Portland cement paste, and curing at 60 °C to 80 °C. A linear hydration kinetics can be obtained showing the depth to which the slag is hydrated. There is some concern in extrapolating these data to the lower temperatures encountered in normal hydration. However, the conclusions appear to be broadly in line with simple interface kinetic models and with selective dissolution experiments, without any indication of a plateau [61, 62]. As previously mentioned, when Portland cement is added to slag, the microstructure becomes complex. A relatively dense skin of hydration product is observed to form at the slag interface; this differed from hydration in the presence of Ca(OH)2 solution where the reaction products did not present significant densification [60]. In general after 28 days of cure a material covering the surface of slag consisted of Ca, Si, A l and H 2 0 . Its composition resembles the CSH. Also skeletal structures are present at the interface with a "card house morphology" as seen in Fig. 18. The latter are characterized by their high Mg content [60]. These structures are explained by the dissolution of slag components, resulting in the migration of its components beyond the original interface between slag and cement; Ca and A l are believed to be the migrating cations, while OH" ions diffuse in the opposite direction. Chapter 2.7 - The Hydration Process in Glassy Slags 62 Fig. 18 - Spatial distributions of slag hydration products [60] The differences between Portland cement and blast furnace slag cement are the following. Hydration products in pure PC paste precipitate close to the cement particles, so that relative large capillary pores remain in the space between the particles. On the other hand, in BFSC and fly ash, the reaction products precipitate more into these pores [60]. Ultimately, this will result in a finer pore structure due to the filler-like effect (better particle size distribution, better dispersion, etc.) Concluding on these rather complex phenomena, one must emphasize that the reactive components in mineral admixtures are represented by the amorphous glassy structures. Fly ash, silica fume, activated metakaolinite and glassy slags consist of polymerized tetrahedra which share Si atoms at all four coordinates fitting well into the Zachariasen model for glasses and Chapter 2.7 - The Hydration Process in Glassy Slags 63 amorphous compounds. The dissolution process involves breaking these Si-O-Si bonds [58]. The dissolution of the mineral admixtures will depend on the alkalinity of the cementitious system. Silica fume, metakaolinite and slag dissolve at pH levels existing already from the very beginning after mixing (~ 12). Fly ash glass is more alkali resistant, and the pH should increase above 13, before any noticeable dissolution. The reactivity of admixtures is influenced by temperature. The increased pozzolanic reactivity results in an increased amount of C-S-H gel deposited inside the pores. It is suggested that the C-S-H gel originated from the reaction between mineral admixtures, lime and water, is deposited between the mineral admixtures and cement particles, instead of in the vicinity of the cement particles, as in the case of pure Portland cement. This leads to a finer size distribution. Moreover, it has been shown that the finer pore system of pastes with mineral admixtures results in a large reduction of the transport of ions, liquids and gases, although it does not influence the strength of pastes significantly. The addition of admixtures causes substantial decrease in the interfacial zone between cement and aggregate, due to improved particle packing [58]. The microstructure of slag cement blends differ significantly from the ordinary Portland cement, as does their internal chemistry. Slag grains develop membranes which control diffusion. This densifies the paste matrix which, as a result, often exhibits low permeability. The later stages of reaction are marked by ongoing mineralogical and structural changes; Ca(OH) 2 is gradually consumed and Mg forms hydrotalcite. The existence of chemical gradients lead to the appearance of Liesegang rings or membranes, made prominent by the precipitation of Mg, A l , Fe, Ti and Mn oxides [60]. Chapter 2.7 - The Hydration Process in Glassy Slags 64 2.7.3 Hydration Mechanism in Other Glassy Materials The literature on this matter discusses glasses with a variety of compositions [63, 64, 65, 66]. However, most of them are dealing with compositions belonging to other systems and/or with ranges beyond the interest of the present research. Since there is no information available on the hydration process of high iron silica based glasses, belonging to the CaO-Si02-Fe203-MgO-Al 203 system, this section reviews issues related to other types of glasses and their possible mechanism of hydration in an attempt to find similarities with the hydration of glassy steel slags, to be discussed in a later chapter. C-A-S Type Glasses Cements based on this type of glasses, (45 %CaO, 35 %A1 20 3 and 20 %Si02), when combined with gypsum, set and produce ettringite [63, 64]. This process occurs without expansion, which commonly accompanies the formation of ettringite. Randomly oriented ettringite is observed in the pastes suggesting a simultaneous formation of ettringite by a through-solution reaction (rapid dissolution of the starting constituents in the liquid phase and precipitation of ettringite from the solution) in contrast to a topochemical ettringite formation and the resultant oriented growth of the ettringite crystals, which is the case for development of expansion. The early strength at ambient temperatures is rather low, whereas the strength found at 7 and 28 days are similar to that of ordinary Portland cement. B-SEM reveals the presence of significant amounts of non-reacted glass even in pastes hydrated beyond 28 days. In the presence of gypsum, these glasses do not yield any crystalline hydration products. Gehlenite hydrate (GH) and/or hydrogarnet (HG), formed in the absence of gypsum, are not formed under these conditions. It appears that the Si0 2 is converted into an amorphous calcium silicate hydrate (C-S-H) phase and possibly also into an amorphous S i 0 2 n H 2 0 . Chapter 2.7 - The Hydration Process in Glassy Slags 65 Of other cements, supersulphated cement (Portland clinker + blast furnace slag + gypsum) is the closest to C-S-A cement. In supersulphated cement, ettringite is formed in a reaction between the glassy phase in the slag and gypsum. The reaction is not accompanied by significant expansion. However, due to an increased amount of S i 0 2 and lower A 1 2 0 3 content in the glassy phase and due to the presence of Portland clinker, an amorphous C S H phase, rather than ettringite, is the main product of hydration [63, 65]. MacDowell et al [66] also conducted studies on calcium aluminosilicate glass cements. Quenched from 1600 °C to room temperature, after grinding and mixing with water, glass cements hydrate without activators and develop good mechanical properties. Gehlenite (C 2AS) is a phase encontered in high alumina cement and in blast furnace slags. Lea [18] recognized this compound as hydraulically inert, while other researchers suggest that gehlenite type glass in granulated blast furnace slags can be hydrated in the presence of activators. Yet, MacDowell [66] reported producing a stable gehlenite hydrate cement directly from CaO, A l 2 0 3 , S i 0 2 glasses without activators, showing that these glasses are very reactive. Moreover, the study gives a good correlation between the heat of hydration and compression strength, indicating that microcalorimetry is an effective method for estimating the hydraulicity of cements. Monocalcium Aluminate Type Glasses Very little or contradictory results are found in the field of high alumina compounds concerning the influence of the degree of crystallinity. Sorrentino et al [67], discussed the hydration mechanism of amorphous or micro-crystalline compounds, tested as additives to calcium aluminate cement to improve the properties of mortars. Hydration tests showed that Ci 2 A7, in pseudo-amorphous form (hereafter denominated CA), hydrates slower than the fully crystalline form up to 7 days, and slightly faster Chapter 2.7 - The Hydration Process in Glassy Slags 66 after 28 days. The authors concluded that the presence of amorphous or microcrystalline C A inhibits the hydration process or has no strong effect, when the CaO/Al203 ratio is between 1 and 2. However, they noted that a small quantity of glass phase whose composition is not the same as the main crystal of calcium aluminate cement, namely the C A , can contribute largely to the hydration in the early stages, due to the presence of reactive microcrystalline matrices, and that the simultaneous presence of CA(amorphous) and Ci2A7(crystalline) contribute to an increase in the hydration rate. According to Sorrentino, crystalline C12A7 is faster to hydrate, a conclusion coinciding with Hisashi et al. [68]. On the contrary, Nakagawa et al [69] report that amorphous C A hydrates immediately after contact with water, and sets in 3 to 5 seconds, which is faster than crystallized Q 2 A 7 . The hydration kinetics of amorphous C A differs from crystalline C i 2 A 7 , as shown by Nakagawa et al [64]. The analysis is done using an expression given by Kondo and Ueda [70]. The equation for the reaction rate is: where a is the degree of hydration and t the hydration time. When n equals 1 , the reaction rate is constant regardless of the thickness of the reaction layer; for n=2, the reaction rate must be controlled by diffusion through the reaction layer. If n>2, the resistance to diffusion is supposed to be increased due to the reaction layer becoming denser. For amorphous calcium aluminate (CA), it was found that n has a constant value of 4 (at 1 5 sec. after water addition). This is attributed to a rapid increase in diffusion resistance, since particle surfaces are covered by amorphous hydrates and A H 3 gel. For the crystalline form (C12A7), n is 1.5 in the time interval of 1 5 sec. to 5 min. after water addition. The value of n is 3 (4) Chapter 2.7 - The Hydration Process in Glassy Slags 67 in the interval of 5 to 30 min. In this case it is suggested that the reaction rate is controlled by surface chemical reaction at first, and that diffusion resistance at the particle surface increases later. Fig. 19 shows that the rate of reaction for amorphous calcium aluminate is higher than for the crystalline form, for a period of 10 min. A violent reaction occurs immediately after the amorphous phase is placed in contact with water, and a dense membrane of amorphous hydrates forms, further retarding the reaction due to an increased diffusion resistance. After 10 min. and attaining approx. 20 % degree of hydration [67], the reaction rates are inverted, the crystalline phase having a higher rate. O No add. Amorphous • No add. Crystal • CH 1mole A CSHj 1mole 0.01 15 30 5 15 30 (sec) (min) Time (log scale) Conflicting results related to the reactivity of the pseudo-amorphous phases indicate these aspects still needs to be dealt with. From the literature survey, it appears that no work has been Chapter 2.7 • The Hydration Process in Glassy Slags 68 done on glasses with composition similar to steel slags, nor on any other types of glasses containing high iron content (20-30% Fe203). Chapter 2.8 - Activation and Properties ol Slag Cement 69 2.8 The Activation and Properties of Slag Cement Industrial slags which have sufficient latent hydraulic power to be used directly as binders are not commonly encountered, and usually it is necessary to add an activator to ensure that hydration reactions are attained in a reasonable time. In order to decrease the setting time and increase the early strength development of cements, numerous species serve as activators and accelerators. Portland cement represents a traditional activator for slags. As mentioned in a previous section, granulated slags can also be activated by lime, sulphates, alkalies and alkali salts alone or in combination. A brief review on the effects of activators on the hydration of slag-cements, the mechanism involved and the products of reaction is further presented. However, this discussion revolves around blast furnace slags cements or fly ash cements and again, there is a lack of information regarding the activation of steel slags and/or steel slag cements. Since this current research used calcium chloride as an activator for steel slags, a review of the prevalent effects of CaCl 2 addition in cement and concrete mixes is in order. Granulated blast furnace slag consists of a glassy calcium magnesium aluminosilicate with potential cementitious reactivity. The degree of hydraulicity in slags depends on the composition and nature of slag, as well as on the type of activator added. In general, slags with high lime content show higher hydraulic potential. The presence of A1 2 0 3 , MgO, P2O5, C r 2 0 3 and other oxides influences reactivity, however their role is still not fully understood [71]. While some slags show strength development without activators (akermanite type glass), others need activators (gehlenite type glass). Furthermore, activators have variable activation effects on different types of slags [72]. Alkaline activators, such as Ca(OH) 2, NaOH and K O H will accelerate the dissolution of Si and A l ion, breaking the Si-O and Al -O bonds in the slag glass structure, which will then be followed by the precipitation of low solubility calcium Chapter 2.8 - Activation and Properties of Slag Cement 7 0 silicate, calcium aluminate and magnesium aluminates due to the increased ionic concentration in the liquid phase [73]. When gypsum is present in the mix, Ca and A l ions concentrations will be reduced because they interact to form ettringite. When the pH of the liquid phase reaches 12 (the level at which ettringite forms) the hydration of slag is accelerated. This means that in order to maintain slag hydration active, it is necessary to supply sufficient hydroxyl to create a high pH environment for breaking the network of glass and to form ettringite. This is achieved by the addition of activators containing these species. For example, Na2SC>4 is a source of hydroxyl and SO3 through its reaction with calcium hydroxide. However, the amount of additives must be limited, since high levels could cause the degradation of the hardened pastes. To date, their limits can be established only experimentally. 2.8.1 Mechanism of Hydration There has been many attempts to explain the hydration of slags, however the mechanism is still not very well understood [72]. A l l alkalies and alkali compounds whose anions or anion groups can react with C a 2 + to produce Ca-compounds that are less soluble than Ca(OH)2, can act as activators of slag. The anion activator will react with C a 2 + dissolving from the surface of slag and form Ca-compounds during the initial period. The initial pH plays an important role during the initial hydration, while the further hydration and structure formation are determined by the Ca-compounds. Thus, a higher pH does not mean a greater activation and thus it is not necessary to keep the pH of the activator solution over 12 as has been previously suggested [72, 74]. The presence of alkali ions M + will neutralize the anion in the activator solution and provide an alkaline solution, which will ensure the initial dissolution of slag. The presence of alkali cations increases the solubility of Ca in CSH, therefore alkali cations are regarded as catalysts. NaOH is a selective activator, and might give Chapter 2.8 - Activation and Properties of Slag Cement 71 different effects in various slags. At the moment, a fast method of evaluating the hydraulicity of a slag is the A S T M C1073 standard. The slag is hydrated for 24 hr at 50 °C and in the presence of NaOH. However the results may unfairly rank slags that are not NaOH selective. 2.8.2 Products of Hydration In general, the products of hydration are present in an amorphous gel form and are difficult to identify. For this reason, there is little agreement on the exact products and mechanism of formation. For example, Wu et al [73] observed that the products of hydration in blast furnace slag cements are different than the ones in ordinary Portland cement. The CSH formed in blast furnace slag cement is a Si-rich C-S-H, having a different atomic bonding and strength than the Ca-rich C-S-H in Portland cement. The former might be stronger due to a greater proportion of bridging oxygens, which might explain the excellent properties at later stages that blast furnace slag cement has in contrast with Portland cement. On the other hand, Shi and Day [72] noted that the main hydration products in blast furnace slag cement are essentially similar to those formed in Portland cement; CSH gel, C H , C 3 A H 6 and AFt. Also they observed that an increased content of slag will gradually result in the formation of C 2 A S H 8 and C4AH13 and, if the MgO content in slag is high, M 4 A H 1 3 will occur. Glasser [60] showed differences between products of hydration in slags as a function of activator (see Table VI). Summarizing the information given previously, we can say that the main hydration product of slag cement pastes is calcium silicate hydrate (CSH), and other minor hydration products varying with the nature of the activators [60, 72]. The nature of the chemical activator will influence the products of hydration, which in turn will change the hydration process and microstructure formation, finally affecting the strength development. Chapter 2.8 • Activation and Properties of Slag Cement 7 2 T A B L E VI - Slag Reaction Products [60] Nature of Activator Crystalline Phases Comments NaOH, N a 2 C 0 3 , Na-silicate CSH, C 4 A H 1 3 , C 2 A H 8 , Mg(OH) 2 Some Si in C 3 A H i 3 , C/S in CSH less than OPC Ca(OH) 2 CSH, C 4 A H 1 3 C 2 A H 8 Sulfate, e.i., gypsum, hemihydrate, phosphogypsum CSH, AFt, and Al(OH) 3 S acts to some extent as an autoactivator Cement Various: CSH, AFm, AFt, hydrogarnet, hydrotalcite-like phase; also, vicatite, C 3 S 2 H 3 Not all these phases are likely to be encountered in the same paste Activators exhibit selectivity and hence different activators will render different effects on slags. For example, the A S T M C1073 test, using high temperatures and NaOH additions,'is not suitable for the evaluation of slag activity in general, because some slags are nor benefited by this specific activator. The addition of activators, if any, should be based upon activator-optimization testing rather than specifying NaOH as the only possibility. The design of a distinct rapid test procedure might be required. Calorimetric measurements at higher temperatures (> 60 °C) might constitute a base for assessing the hydraulicity of slags in the absence of admixtures, thus avoiding the introduction of an additional variable in the already complex process. Chapter 2.8 - Activation and Properties of Slag Cement 73 2.8.3 Calcium Chloride as an Admixture in Cement As previously mentioned, there are a number of admixtures that are added to cement with the purpose to accelerate or delay the hydration process. In general, retarders and accelerators work by affecting the hydration of the C3A present in cement [41]. For example, the rapid setting of cement which contains no gypsum is a result of the almost instantaneous setting of C 3 A . In such cement, alumina passes into solution and combines with silica to form a gel, poor in lime, which stiffens the mix. In the presence of gypsum, ettringite is formed and coats the C3A grains. The formation of this coating retards the hydration of C3A and setting is delayed until the calcium silicates start to hydrate. Similarly, setting may be delayed by reducing the solubility of the alumina. Such reduction limits the formation of the lime-poor gel, and thereby delays setting even when cement does not contain gypsum. Calcium chloride (CaCb) is probably the most common accelerator, which finds applications mainly in cold weather, when it allows a strength development in the plain unreinforced concrete approaching that of concrete cured at normal temperatures [75]. However, it is interesting to point out that CaCl2 can act as both retarder and accelerator. The addition of CaCl2 to Portland cement results in the formation of a double sulphoaluminate salt. For concentrations lower than 1 %, the presence of CaCb does not affect the solubility of alumina, which remains low. Hence, in small amounts, CaCl2 acts as a retarder. At higher concentrations, the solubility of the alumina increases and CaCl 2 acts, therefore, as an accelerator [47]. That is, the accelerating effect due to increased solubility outweighs the retarding effect brought about by the precipitation of the double salt. In general, 2% CaCl2 is added to cement. Chapter 2.8 - Activation and Properties of Slag Cement 74 Chemical Effects of CaCl 2 Addition The following aspects are relevant to the reactions occurring in Portland cement pastes: (a) There does not appear to be any chemical reaction between calcium chloride and the di-and tri-calcium silicates although their rate of reaction is increased [75]; (b) Calcium chloride does not react significantly with cement pastes for a period of 2-6 hr, although rapid setting can occur in this period; (c) After a brief initial period (2-6 hr), the free calcium chloride concentration progressively drops to almost nil. CaCl 2 Effects on Cement Hydration When analyzing cement hydration it is important to take into consideration (a) the kinetics of reaction products (b) the composition and morphology of hydration and (c) the mechanism of reaction. (a) Kinetics of reaction - Conduction calorimetric studies indicated that the degree of hydration of a cement paste or that of a silicate paste is considerable increased when CaCl 2 is added (Fig. 20 and 21) to the mix. (a) Composition and morphology of hydration products - CaCl 2 addition to Portland cement results in a tobermorite gel with much higher lime/silica ratio than the plain paste. The differences in chemical composition are accompanied by differences in the morphology of the gel. Tobermorite changes to a foil structure, instead of a cigar shaped tube in plain pastes, which renders a more open and accessible structure [75], hence promoting the hydration process. Chapter 2.8 - Activation and Properties of Slag Cement 75 I O O R 20 V-J i i i 1 ' 1 0 0 . 5 I 3 >0 2 8 64 I O O T i m e ( d o y ) Fig. 20 - The degree of hydration of cement pastes in the presence of calcium chloride in comparison with plain paste, measured by X-ray analysis [75] Fig. 21 - Degree of hydration of a tricalcium silicate paste in the presence of calcium chloride; measured by non-evaporable water content [75] Chapter 2.8 - Activation and Properties of Slag Cement 76 Several studies were dedicated to observing the activation of tricalcium aluminate using different compounds. It has been shown that accelerating admixtures react considerably with the C 3 A , resulting in different hydration products to those resulting from a plain cement. In the presence of CaCl 2 an insoluble calcium chloroaluminate complex, also called "Friedel's salt" [76] is formed. C 3 A 1 2 H 2 0 + CaCl 2 = C 3 A-CaCl 2 -12H 2 0 This reaction is said to reduce the C a 2 + concentration in the cement pore fluids to extremely low levels [44] with the result that the C3A reactions are either diminished or retarded [77], at times beyond 60 hr [44]. However, the beneficial effects of CaCl 2 addition on the dicalcium silicate and tricalcium silicate in Portland cement are of greater importance and overcome the drawbacks encountered for the tricalcium aluminate, as calcium silicates are the principal strength contributing compounds in cement. Other studies, focusing on fly ashes, reported that the use of CaCl 2 (at 50 °C) can increase the pozzolanic reactivity of fly ashes and significantly improve the strength. Due to the presence of high alumina content (20%) in fly ashes, a solid solution C4AH13-C 3 A C a C l 2 1 0 H 2 O is formed. It was noted that the chloride activator produces the highest strength in fly ashes with lower calcium content [78]. (c) Mechanism of reaction - In spite of all the data acquired until now, the way CaCl 2 operates is not fully understood. However, there is consensus in the sense that the mechanism of reaction involves the acceleration of the C 2 S and C3S hydration, and that the reaction is catalytic in nature, explained by the fact that tobermorite gel, although altered in character, does not contain chemically bound chloride. Chapter 2.8 - Activation and Properties of Slag Cement 77 Most of the literature on CaCl 2 is directed to the study of C 2 S, C 3 S, C 3 A and Portland cement pastes. However, a recent study attempted to elucidate the role of C4AF in Portland cement with respect to chloride binding. It was concluded the chloride is bound • in the form of a chloro-complex C 3 F C a C l 2 1 0 H 2 O which is derived from C4AF phase [76]. Yet, there is still no data available on how this complex affects the rate of hydration of the cement. C 3 F C a C l 2 1 0 H 2 O and the C 3 A-CaCl 2 10H 2 O have an identical crystalline structure. "Friedel's salt" and its ferrite analogue belong to the family of aluminate-ferrite mono (AFm) phases [76, 79]. On this basis, one may infer that the chloro-ferrite could behave similarly to the chloro-aluminate [79], and hence retard the hydration process. This reasoning might also point towards the possibility that by adding CaCl 2 to high iron content steel slags and/or high alumina content, a retarding effect might be observed. Nevertheless, as stated in an earlier paragraph, only experiments will show the real effect that calcium chloride might have on these type of cementitious materials. Chapter 2.9 - Energetics of Slags Hydration 78 2.9 Energetics of Slags Hydration While cements are almost entirely crystalline, slags intended for cement blending are mainly glassy. The fast cooling of slag produces a vitreous material. This material is thermodynamically metastable as evidenced by its reaction with water and its devitrification through heating at around 1000 °C. The relationships between states are shown in Fig. 22. >-0 500 1000 1500 TEMPERATURE ,°C Fig. 22 - Generalized Gibbs free energy-composition diagram, showing the relation between states [60]. Molten slag has a choice of reaction pathway when cooled: the stable way that leads to crystallization may be arrested if cooling is sufficiently rapid. In this way the slag persists below its equilibrium freezing point as a supercooled liquid, and due to the high viscosity, the long-range diffusion is prevented and the atomic (or ionic) disorder characteristic of the vitreous state is effectively frozen in. Chapter 2.9 - Energetics of Slags Hydration 79 Fig. 22 also indicates the possibility of attaining various states of annealing within the glass and the formation of metastable crystalline phases which may arise by devitrification of the glass. At near ambient temperatures, metastable crystalline phases are close in energy to those of stable crystalline phases. These thermodynamic aspects explain, in general terms, why slags used as cement blends should have as high a glass content as possible. The thermodynamic metastability gives glass the potential of reacting.[80]. Crystalline phases derived from glass almost always have a lower driving potential for hydration. Few exceptions to this generalization occur. One exception is the case of metallurgical slags which crystallize p-Ca2Si04 when cooled, since the latter is known to hydrate readily [60]. However, slag compositions are such that the most abundant crystalline phases resulting from partial crystallization include mainly melilite, merwinite, bredigite and spinel, all of them being less reactive than the parent glass. Mellilite is a Ca, A l , Si, M g solid solution encountered often in slags. Slag glasses are often referred to as mellilite glasses. The term is a misnomer, because if the glass should by chance have the mellilite composition, its atomic arrangement is not characteristic of mellilite. Glasser [60] mentions that some researchers claim that partial crystallization of a glassy slag will enhance the reactivity of the residual glass, however, these results could not be reconfirmed by his research group. There are still differences of opinion in this field of research and hence, additional knowledge is required, often at a more fundamental level. A representation of free energy for granulated blast-furnace slags is presented in Fig. 23. Chapter 2.9 - Energetics of Slags Hydration 80 CD > CD c ai crystal glass hydrate Fig. 23 - Schematic Representation of Free Energy for Granulated Blast-Furnace Slags (Hooton 1987) [81]. The free energy G h for the direct hydration of the glass slag is: G , = G +Gh ^g.h ^g.c *-"c,h (5) where: Gg c = free energy difference between glassy slag and devitrified crystalline phases (such as, akermanite(C2MS2), gehlenite(C2AS), wollastonite(CS), merwinite(C2MS2), etc) Gch = energy difference between the crystalline phases and the hydration products Egh, Egc, Ech are the energy barriers or the activation energies corresponding to the transformations. Chapter 2.9 - Energetics of Slags Hydration 81 Since granulated blast furnace slag hydrates very slowly at room temperatures, the activation energy for hydration is high. Shi and Day [78] confirmed that the literature does not provide reports on direct tests carried out to determine the activation energy of slag hydration. However, several studies have indicated that the addition of slag to Portland cement increases the hydration activation energy of the blended cement and the higher the slag content, the higher the barrier. On that account, the apparent hydration activation energy for blast furnace slag is higher than for Portland cement. In the presence of activators, the activation energy will depend on the slag composition and also on the specific activator used. Shi and Day [78] also suggested the importance of using elevated curing temperature for a rapid evaluation of the hydraulic activity of slags. If the Arrhenius equation is written as follows: i dflnfcJ E — - = — V (6) the expression d[lnfc r]/<i7', at a given temperature T, is proportional to the activation energy Ea. This means that raising the temperature would be more beneficial for the reaction processes with higher activation energies (blast furnace slags) than for those with lower reaction activation energies (Portland cement). Most researchers have determined the temperature dependence of the heat evolution by calorimetry and applied the Arrhenius equation, or some variant of it, to calculate Ea. Ideally, the equation should be applied to a single reaction; however in blended cement systems, heat is evolved from hydration of cement and slag, so difficulties of interpretation arise. In the early stages of hydration, little interaction occurs between cement and slag; both hydrate semi-independently of each other. Each one retains a characteristic induction period, which is shorter Chapter 2.9 - Energetics of Slags Hydration 82 for cement than for slag. Thus, the overall progress of the reaction in a blended system might be better described via the use of a two-term model, one for each pair of semi-independent reactions: cement + pore fluid and slag + pore fluid. However, such models are difficult to evaluate from the experimental data. Therefore, other fundamental approaches have to be pursued. Since the hydration of slag blends is sensitive to temperature, the concept of "equivalent age", or the F factor, has been used to equate results obtained at various temperatures [60]. Regourd [82] calculated the activated energies for cement and slag in a blend using the following approach. She suggested that the problem could be solved by measuring the time (t) necessary to reach an equal degree of hydration and then comparing numerical values, including the equivalent age factor. The ratio of two times could be related to an Arrhenius type relationship: F = — = exp- (7) where Ti and T 2 are the two temperatures. Using this method, Regourd reported activation energies of 46 and 50 KJ-mol"1 for cement and slag, respectively. However, when applying the same concept to higher slag additions the difference in activation energies was higher, 42 vs. 56 KJ-mol"1. Another attempt at calculating the activation energy was done by Taplin [83] and Knudsen [84], using the following relation: t-t0 P = P l 0 = - ( f— (8) They defined P as the change in a particular property (heat included) at a time t with respect to the initial time to and tso the time at which 50% hydration was obtained. Plotting 1/P Chapter 2.9 - Energetics of Slags Hydration 83 against l/(t-tn), where t0 is time taken from the end of the induction period, the resulting energies can be calculated. Reported values for activation energies when using this approach are 43 and 49.1 KJ-mol"1, for cement and slag respectively. Many other researchers found significant differences in activation energies between cement and slag (approx. 6 to 16 KJ-mol"1). From these differences it is clear that slags would experience a greater dependence of their hydration rate on temperature [82, 85]. Wu et al [85] studied the kinetics of hydration in both Portland cement and blast furnace slag - cement. The apparent activation energy (Ea) for Portland cement was 44 KJ/mol and for slag - cement 49 KJ/mol. The Ea for slag was significantly higher than for PC. Slag hydration was accelerated by increasing the temperature. The effect of elevated temperature (60 °C) on the hydration of slag cement was seen to be much greater than on Portland cement, illustrating the benefit that slag cement gets from thermal activation. The total heat released by pure cement and slag - cement, at 27 °C and 24 hr, was 180 vs. 112 J/g respectively. These values are consistent with a higher strength development in pure cement than in slag - cement at early ages. However, at 60 °C, the total heat liberated by slag - cement was greater than of pure cement (313 vs. 232 J/g). A general hydration model for PC and BF slag was proposed by De Schutter and Taerwe [86] using heat release determinations, a w/c ratio of 0.5, temperatures of 5, 20 and 35 °C, and a relatively wide range for the specific area (a Blaine number varying from 4300 to 5100 cm2/g). The contribution of the 1st peak related to wetting was discarded for two reasons: the calorimetric tests were unstable in the beginning of the reaction and the 1st peak is never produced inside a concrete element, but in the concrete mixer. The apparent activation energy for Portland cement was 33.5 KJ/mol. The total heat release (<2ma*) in Portland cement was 350 J/g. They assumed Chapter 2.9- Energetics of Slags Hydration 84 that the slag hydrates independently of the cement a fact confirmed by other researchers [77]. An interesting observation, revealed by Roy and Idorn [87], and which points towards this conclusion is that the alkalis and lime, released by the residual Portland cement, are retained in the hydration products of the slag fraction, and do not seem to contribute to the hydration of the slag. Fernandez et al [61] conducted a study on the hydration of alkali (water glass) activated blast furnace slag (AAS). Mixtures of activator/slag were prepared with a constant 0.4 ratio. Applying isothermal conduction calorimetry they followed the reactions at 25, 35, 45 and 60 °C, and monitored the heat release for 45 hr. Fig. 24 shows the rate evolution curves at different temperatures. 55 50 \- 25'C 35*C 4S"C 60»C 45 CD 10 10 Time (hour) Fig. 24 - Rate of hydration vs. time [61] Five steps can be distinguished. The first step is associated to the 1s t peak of the curve. It corresponds to the first minutes of reaction and it is assigned to slag partial dissolution. The 2 n d Chapter 2.9 - Energetics of Slags Hydration 85 step corresponds to a period called the induction period in which the rate is low and it is the consequence of a period of low reactivity. The 3 r d and 4 t h steps are associated to the 2 n d peak. These steps are called acceleration and deceleration. In this period a massive precipitation of reaction products occurs. The 5 t h step corresponds to a low reactivity period called decay, or finishing of reaction. Experimental data indicate that the temperature influences the duration and intensity of these steps. In order to determine kinetic parameters, Fernandez et al [61] used the Mori-Minegishi method [62], based on the following equation (9). ( l - O - a ) ^ ) " =kt (9) where, cc= degree of reaction, k= rate constant; f= hydration time; N is a value that varies according to the type of reaction governing the process. N is the slope obtained when plotting log|l - (l - a)^3 j vs. log(t). When TV <1, reaction occurs through a nucleation kinetic, when N=l, the process is governed by a phase boundary kinetic and when N=2, the kinetic corresponds to a diffusion process. At all temperatures, N varied between 1.8 and 2, indicating diffusion control. Slag activation occurs, from the acceleration process point of view, by diffusion. The diffusion takes place through a layer of reaction products surrounding the unreacted slag grain. The literature does not report on any direct tests carried out to determine the activation energy for steel slags hydration. The only two facts known are: - vitreous blast furnace slag has a higher activation energy than Portland cement. - raising the temperature is more beneficial for the reaction processes with higher activation energies than for those with lower reaction activation energies, hence vitreous blast furnace slag hydration is more sensitive to temperature. Chapter 3 - SCOPES AND OBJECTIVES 86 C H A P T E R 3 SCOPE AND OBJECTIVES It has been established [4, 10, 11, 19, 40], that one criterion necessary in a viable slag-cement is the presence of a glassy phase. In the case of a cement produced from steel slags, this glassy phase would have to include high levels of iron oxide (30-40 %). The hydration of glasses with low iron levels (0.3-2.0 %), similar to blast furnace slags, has received attention [26, 42]. However, fundamental studies on the hydration of steel slag glasses and the mechanism responsible for the higher reactivity of glassy slags vs. crystalline ones are lacking. Compared to blast furnace slag, steel slag is regarded as a 'troublesome' raw material and is used only sparingly by the cement industry. For this reason, there is scarce information regarding the kinetics of steel slag hydration in general, and for the glassy slags in particular. A laboratory scale program exploring slag compositions in a more complex system, such as CaO-Si02-Al203-Fe203-MgO, with emphasis on ferrite solid solutions primarily present in the glassy phase, would contribute to a more comprehensive view of the mineralogy of steel slag and thereby establish the compositions with enhanced reactivity. Moreover, the detrimental role of iron will be avoided by incorporating it into the glassy phase. A hydration study of synthetic steel slags, belonging to the above mentioned system, and of oxidized industrial steel slags, would allow for a better understanding of the hydration process in high iron level-cements (eg. Erz cement). The final intention is to increase steel slag additions to the Portland cement mix and hence generate economic benefits, and at the same time provide an environmental solution Chapter 3 - SCOPES AND OBJECTIVES 87 for both the cement and steelmaking industries. Considering all these points, the current work is aimed at gaining additional knowledge in two key areas: (i) Determining the hydration mechanism of high iron content glassy slags relative to Portland cement. In order to observe the influence of glass presence on reactivity, defined as heat released during hydration, the work considers components in both amorphous and crystalline state. As there is no clear explanation on the role of glasses in the enhancement of the hydration process [18], mechanisms accounting for the high hydration rate of glasses (relative to crystalline materials) are postulated, e.g.: different activation energy, higher free energy of formation. These are evaluated against experimental data as outlined in the next section. (ii) Examining the hydration process of oxidized and quenched/slow cooled industrial steel slags, again relative to Portland cement in order to observe differences in the extent of reaction, and possible mechanisms. Chapter 4 - METHODOLOGY 88 C H A P T E R 4 METHODOLOGY The hydration behavior of glassy steel slags was assessed on the basis of: (i) chemical composition, (ii) mineralogical composition and (iii) physical state. The project began with the analysis of Portland cement and moved on to synthetic slags simulating oxidized steel slags, as well as oxidized industrial steel slags. At this point a more complete discussion of the experimental variables is in order. Chemical composition. In establishing the composition range of synthetic slags suitable for cement-making, two factors were considered. Steel slags contain all the oxides present in raw materials for cement manufacture (CaO, Si02, Fe203, AI2O3, MgO) and Portland cement clinker undergoes hydration due to the presence of cementitious compounds such as brownmillerite ( C 4 A F ) , dicalcium silicate (C2S), tricalcium silicate (C3S) and tricalcium aluminate (C3A) [12]. Consequently, the composition embraced synthetic slags with oxides ratios that ensured the formation of these compounds and contained the typical steel slag iron content. Several synthetic slags with composition belonging to the quinternary system CaO-Si02-MgO-Al203-Fe203 were fabricated from laboratory grade oxides and carbonates at temperatures around 1530 °C (more details are provided in Appendix I). These slags had lime to silica ratios (3:1 and 4:1) and iron levels (30-40 % Fe203) typical of steel slags, as well as 10 % MgO and 4 % AI2O3. The aim was to obtain materials comprised of calcium silicates, calcium aluminates, ferrites and amorphous Chapter 4 - METHODOLOG Y 89 phases specific to Portland cement clinker and avoid the variability that could be introduced by industrial steel slag samples which contain a variety of other constituents. It has been shown [4] that the oxidation of bivalent iron to trivalent iron enhances cementitious properties by reducing the O:Si (basicity) ratio of the complex ions in the slag. In the molten slag, the trivalent iron will act as a network former, promoting a shift in ionic structure from small orthosilicate ions towards larger more complex ionic rings, which are less apt to crystallize from cooling. Therefore, in order to promote glass forming potential, steel slags would have to be oxidized prior to melting. In view of these findings, synthetic slags fabricated in the present work contained iron primarely in the trivalent form (Fe203), minimizing the detrimental FeO presence. In addition, oxidized, melted and quenched or slow cooled industrial slags were tested. Mineralogical composition. This characteristic refers to the amount and type of phases to be formed in slags, comprising amorphous phases, inert and cementitious components. The work focused on the determination of phases formed upon melting and quenching of slags. Mineralogical features relative to changes in the cooling rate were observed. Physical state. The crystallinity or glassification level in the samples can be influenced largely via cooling rate. Previous research has shown [1, 10, 40] that the presence of a glassy phase is necessary to render a slag cementitious, hence the work looked at the relationship between crystallinity level and hydration potential. The methods to obtain different cooling rates and hence different crystallinity levels were: air cooling ('slow') and water quenching ('moderate'). The estimated cooling rates for these methods are: Chapter4 - METHODOLOGY 90 Method Approx. order of magnitude ( ° C/sec) Air cooling ('slow') 102 Water cooling ('moderate') 103 Higher cooling rates (106 °C/sec) were attempted to enhance glassification in slags, as well as in laboratory grade cementitious compounds, by using plasma spraying techniques and liquid nitrogen quenching. Some of the results were encouraging, showing a 30-50% reduction in slag crystallinity, however problems arose due to material flowability and torch parameters. In order to use this method, further research is needed. Analytical techniques The analytical techniques used in this project were aimed at measuring: (i) The amounts of C?S, C3JS, C T A , O A F and glass present in synthetic slags. The amount of cementitious components and the differences between crystalline and amorphous structures were assessed by both X-ray diffraction (XRD) and back scattering electron microscopy (B-SEM). These techniques are commonly recommended for cement chemistry and for mineralogical studies. B-SEM in particular is the most suitable for cement slags and calcium silicate compounds since these compounds have different atomic numbers and they can be distinguished onto the image displayed on the cathode-ray tube. Polished thin sections (30 pm thick) mounted on microscope slides were prepared for such purposes. (ii) The heat of hydration (at temperatures ranging from 20 to 70 ° C ) . Assessment was made on the basis of heat of hydration, as well as the activation energy for hydration. The heat Chapter 4 - METHODOLOG Y 91 of hydrat ion, a characteristic l i nked to setting and the development o f strength in the hardened paste, was determined by isothermal conduct ion ca lor imetry , w h i c h records the development of the hydration process as a funct ion o f t ime and temperature. Apparent act ivat ion energies can be der ived f r o m calor imetr ic measurements. D i f ferent act ivat ion energies o f hydrat ion for amorphous mater ials , for example , w o u l d indicate the existence o f a different hydrat ion m e c h a n i s m when compared to crystal l ine mater ials , poss ib ly exp la ined by the existence o f unstable cha in anions in the glassy structure, (compared to SiO*~ r ings present in crystal l ine structures) w h i c h react very fast w i t h water [40] or different hydrat ion products. In order to per form isothermal ca lor imetr ic measurements o f the early stage reactions, a special set-up was designed. Deta i ls regarding the set -up, e . . i . ca lor imeter characterist ics, ca l ib rat ion , other equipment, data acquis i t ion and the subsequent analysis o f the output are p rov ided in A p p e n d i x II. (iv) A c t i v a t i o n energies and the rates of hydrat ion. These parameters were determined by app ly ing models general ly used in analyz ing cement hydrat ion [10, 6 1 , 6 2 , 9 3 , 94 , 95] . (v) Ac t i va tors . M o s t of the w o r k was done on pure slags without addi t ion o f activators. H o w e v e r , upon observ ing that steels slags had a relative l o w hydrat ion rate, a 2 % CaCl2 addit ion was used in some trials in an attempt to enhance react iv i ty . Chapter 5 - RESULTS AND DISCUSSION 92 CHAPTER 5 RESULTS AND DISCUSSION The chemical composition of the synthetic slags and the results associated with the mineralogical observations of slags are further presented. 5.1 Mineralogical Observations 5.1.1 X R D and B-SEM Analysis of Synthetic Slags With the aim to produce high iron level slags (30-40 % Fe 2 0 3 ) containing dicalcium silicate, tricalcium silicate and brownmillerite solid solutions, several synthetic slags were fabricated in an induction furnace at temperatures around 1530 °C and in an oxidizing atmosphere. Ferric oxide, alumina, silica, calcium carbonate and magnesium carbonate were used as raw materials. Lime to silica ratios were 3:1 and 4:1. High density alumina and magnesia crucibles were chosen to contain the slags due to their ability to withstand the high temperatures required and to partially resist the chemical attack of the highly basic slags. All samples were water quenched, in order to obtain amorphous phases. The chemical compositions of the melts are shown in Tables VII and VIII. Table VII- Chemical Composition of Synthetic Slags (Calculated Values) Sample Si0 2 CaO MgO A1 20 3 Fe 2 0 3 CaO+MgO / CaO+MgO Si0 2+Al 20 3 % % % % % Si0 2+Al 20 3 % % SS330 13.52 42.48 10.00 4.00 30.00 3.00 52.48 17.52 SS340 11.00 35.00 10.00 4.00 40.00 3.00 45.00 15.00 SS430 10.00 46.00 10.00 4.00 30.00 4.00 56.00 14.00 Chapter 5 - RESULTS AND DISCUSSION 93 Table VIII- Chemical Composition of Synthetic Slags (Chemical Analysis) Sample Si0 2 % CaO % M g O % A1203 % F e 2 0 3 % FeO % CaO+MgO / Si0 2+Al 20 3 CaO+MgO % Si0 2+Al 20 3 % SS330 14.17 40.29 8.63 4.91 31.41 1.67 2.66 48.92 18.36 SS340 12.72 34.21 7.48 10.65 33.17 1.86 1.78 41.69 23.37 SS430 11.49 43.15 7.86 4.43 32.11 0.90 3.20 51.01 15.92 When plotting the compositions on a modified CaO-Si02-Fe203 diagram it can be seen that these slags are located in the dicalcium silicate phase field (Fig. 1). The X-ray diffraction (XRD) and back scattered electron microscopy (B-SEM) techniques allowed for the identification, as well as the quantification of phases. Fig. 25 shows a B - S E M image of sample SS330, water quenched. The slag consists of around 60 % glass (matrix), 37 % calcium silicates (gray areas) and the 3 % periclase and holes (dark areas). Most of the iron is present in the glassy phase, with small amounts detected in the periclase and silicate phases. Sample SS340 (Fig. 26), water quenched, shows an increased amount of glass (62-70%), probably due to a higher content of AI2O3. Also it is interesting to point out the formation of an "oxide" phase (0.30-2.64 %). A spot analysis on this phase indicates very high amounts of Fe and O, which might suggest the presence of Fe2C>3 or Fe 3 0 4 . It appears that Fe diffuses from the matrix into the "oxide" phase enriching it in Fe and leaving behind a "glass" phase lower in Fe content. The presence of iron oxide might be explained by levels of iron exceeding the solubility limit. Other areas indicate larnite (C 2S) crystals (25-35 %) as dark irregular masses "fingered" at the edges or as dendritic structures. Brownmillerite formation was not likely. The alumina, as well as a great part of the iron, seems to be incorporated into the glassy phase. Incipient periclase crystallization is detected as black round areas and more advanced as dendritic or hexagonally oriented crystals (1.50-2.70 %) arising from the glassy matrix (Fig. 27). Some regions contain Chapter 5 - RESULTS AND DISCUSSION 9 4 high concentration of periclase crystals (33 %) as seen in Fig. 28 explained by the use of magnesia crucibles and a high amount of MgO in the mixture. Sample SS430 displays a glass matrix containing a ferrite phase composition (Fig. 29 and Fig. 30). It might be brownmillerite or a solid solution in the series ranging in composition from C 2 F to CeA 2 F. The formation of this type of solid solution it is very likely, since this series can be obtained at around 1250 °C [29]. Round gray crystals of C 2 S are also detected. The same shape of crystals has been reported in the literature [11]. The presence of periclase crystals (fingered and hexagonal shape) is explained by the fact that the synthetic slag contained quantities of magnesium exceeding the solubility limit with the aim to protect the magnesia crucible from reacting with the melt. The overall distribution of larnite is 27-37 %, and periclase is 4-12 %. The sample contains 66 % glass and the peaks appearing in the X R D pattern are short and broad, making the identification of phases difficult. Nevertheless, correlating the X R D and B-SEM results, it seems that the sample contains the following phases: C 4 A F , C 2 S and MgO. Although they were not detected in the sample inspected under the microscope, knowing the complexity of the system to which these slags belong, it is likely to have also formed calcium iron oxide (calcium ferrite), bredigite, merwinite and maghemite. The search and match program for X R D analysis gives a high probability of formation of these phases, but the PDF lines are not a good match. The industrial slags used in present study were previously characterized by Murphy [4]. He encountered mineralogical species similar to the synthetic slags, as well as C3S and FeO, the latter being still present in some melts done with material that was not completely oxidized. Chapter 5 - RESULTS AND DISCUSSION *».t t' !« airs „. A. «... « * FIG. 25 B- S E M of sample SS330, showing 60 % glass (matrix), 37 % calcium silicates (gray areas) and 3 % periclase and holes (dark areas) FIG. 27 B - S E M on the SS340 sample. Periclase crystals as round and dendritic structures arising from a glassy matrix "oxide FIG. 29 B - S E M of sample SS430, glass matrix with composition similar to the ferrite solid solution series (light gray areas), C 2 S (round and fingered dark gray crystals), periclase (black areas) FIG. 26 B - S E M of slag SS340 showing: 1) increased amount of glass (62-70 %) 2) larnite, 3) "oxide" phase, 4) periclase FIG. 28 B - S E M on sample SS340 showing concentration of periclase crystals in the glass matrix FIG. 30 B - S E M of sample SS430, glass (light gray), larnite (dark gray), periclase (black) Chapter 5 - RESULTS AND DISCUSSION Table IX summarizes the mineralogical observations on these slags. Table IX - Phases Detected by B - S E M and X R D Techniques Slag B-SEM results XRD (phases which were not confirmed by the B-SEM) (*) Obs. SS330 P-C2S, larnite 37% MgO(periclase)+holes 3% Glass 60% Fe 203, maghemite Ca 1 4Mg 2(Si04) 8, bredigite -Glassy -Homogeneous SS340 (3-C2S, larnite 25-35% MgO(periclase)+holes 1.5-2.7% Glass 62-70% "Oxide" phase 0.3-2.64% Fe 2 0 3 , maghemite CaFe 30 5, calcium iron oxide Ca 1 4Mg 2(Si0 4) 8 , bredigite Ca 3Mg(Si0 4), merwinite -Very glassy SS430 P-C2S, larnite 27-37% MgO(periclase)+holes 4.8-14% "Oxide" phase 0.12% Glass 57-67% Fe 2 0 3 , maghemite CaFe305, calcium iron oxide Cai 4Mg 2(Si0 4) 8 , bredigite C 3A,Ca 3Al 206, tricalcium aluminate -Glassy -MgO not homogeneously distributed in the sample (*) X R D results match the B - S E M findings. Nevertheless, there are several additional phases for which the X R D gives a high probability of formation and that have not been detected by B-SEM. These phases are listed in this column. Chapter 5.2 - Hydration Study 97 5.2 Hydration Study As previously mentioned, the present research work measures heat effects by way of isothermal conduction calorimetry, which allows for a continuous monitoring of the heat release for a chosen time frame. However, it must be emphasized that the study covers only the early stages of hydration, when the reaction rates are relatively high and the heat effect is large enough to be properly recorded. It is known that the rate of hydration decreases after the setting period, and that the time required for capturing heat changes would be very long (several days) and hence not practical. A series of calorimetric results for Portland cement, synthetic and industrial steel slags are reported. Slags in both quasi-amorphous and crystalline form are used, with and without activator. The hydration of PC, steel slag and gypsum mixtures is also presented. The differences in reactivity are interpreted by applying the concept of activation energy using three basic linear kinetics models: Knudsen, Avrami and Jander. 5.2.1 Heat of Hydration of Portland Cement Portland cement was chosen as a reference material for contrasting the behavior of synthetic steel slags in their amorphous and crystalline form. The conduction calorimetry profile at 25 °C and a w/c ratio of 0.5 is given in Fig. 31. The result confirms earlier findings [10, 88, 89] in that two exothermic peaks can be observed. The first, appearing early in the test but of short duration, is attributed to various possible reactions including the heat of wetting of the sample and the early hydration of the cement phases ( C 3 A , C3S), while the second corresponds to the setting phase with the rapid hydration of C 3 S being the major contributing reaction [57, 77]. Integration over 24 hours gives an enthalpy of 15 J/g for the first peak and 158 J/g for the second. Chapter 5.2 - Hydration Study 9g The range of results on duplicate tests is ± 9 J/g. The resulting exothermic effect is comparable to values reported in the literature by Killoh et al. [89]. 0.300 0.250 4 !3 0.200 + 3 ta c 0.150 + 5 1 g 0.100 u es X 0.050 Enthalpy (J/g) at 24 hr: -173 (- 15 - 158) 0.000 - 15 J/g 9.2 hr 158 J/g 10 15 Time (hr) 20 25 Fig. 31 - Heat of hydration for Portland cement at 25 °C and 0.5 w/c The Influence of Temperature on the Heat of Hydration In order to determine the temperature effect on the heat of hydration tests, were done at temperatures ranging from 15 to 70 °C for a period of 24 hr. Fig. 32 shows that an increase in temperature will shift the setting peaks towards lower times and will increase the total heat evolved. These findings are in agreement with calorimetric measurements done on different kinds of cements [88]. The first peaks, related to the wetting and the partial dissolution of cement, were removed when smoothing the calorimetric curves; however, they were included in the calculations of the total heat. Chapter 5.2 - Hydration Study 99 Table X provides the details of the heat release. The initial heat corresponding to the first peaks is not greatly affected by temperature. Yet, a marked effect is noticed in the time at which a maximum in the heat evolution occurs during the setting phase, as well as in the total heats. The maximum is reached at lower times when the temperature increases, e.g. 20.3 hr at 15 °C down to 3.6 hr at 70 °C, while the total heat increases from 162 J/g to 379 J/g. Portland cement hydration at different temperatures 0.025 -i , Time (hr) Fig. 32 - Heat evolution rate for Portland cement vs. temperature Table X - Heat of Hydration at 24 hr for Portland Cement Temperature (°C) Total Heat of Hydration, at 24 hr (J/g) Initial Heat (J/g) Heat of Setting (J/g) Time at which Maximum Heat Release Occurs (hr) 15 162 19 143 20.3 25 173 15 158 9.2 40 220 11 209 6.2 50 263 7 256 4.6 60 343 15 328 3.7 70 379 13 366 3.6 Chapter 5.2 - Hydration Study 100 Fig. 33 gives the cumulative heat effect with time. The tests were stopped at 24 hr, however the hydration proceeds beyond this point, albeit at very low rates. The absence of a plateau indicates the reaction is still in progress. Cumulative heat vs. time Portland cement Time (hr) 25 °C -a—40°C —tr- 50°C —•—60°C - * — 70°C Fig. 33 - Cumulative heat vs. time for Portland cement The cement hydration process is heterogeneous and polyphase; it is therefore, not easy to calculate kinetic parameters. The material has a broad particle size distribution and the hydration behavior does not obey the same form of kinetics as single grains. Knudsen [84] proposed a hydration-time equation derived from a Laplace transformation formulation which corrects the degree of hydration of the total particulate system to the degree of reaction of the single particles. By using this approach it was shown that Portland cement is hydrating according to linear kinetics. Since the hydration of cement continues with time and does not show a plateau, the Chapter 5.2 - Hydration Study 101 final heat released must be estimated. The Knudsen equation can be used to estimate the final heat by extrapolating data in the last part of the curve. The Knudsen relation is as follows: 1 1 t50 — = + 5 " (10) Q Q [Q{t-t0)] where Q is the represented property (heat evolution, strength, etc.). In this work Q is the cumulative heat evolution at time t-to, to is the time at the end of the induction period, Qj is the value of Q at infinite time, tso is the time necessary to achieve 50% of Qi. If the results of heat evolution measurements are plotted in a double reciprocal diagram of 1/Q vs. l/(t-t0), a linear equation can be derived and tso and Qi can be calculated from the intercept and the slope. If it is assumed that t 5 0 is an inverse function of the reaction rate constant k, then the ratio of tso determined at two different temperatures (Ti and T 2) can be used instead of k to calculate the activation energy. The Arrhenius relation that gives the apparent activation energy is: — = exp| K2 RTXT2 (11) Before applying this method, the Q values must be obtained by integrating the rate evolution curves at various times. A time to, for the beginning of each acceleration period, must be defined. The first peak of hydration is discarded and a value of heat release is used in the calculations, which is defined as the heat released from the end of the induction period. Fig. 34 shows the double reciprocal curve from which the final heat Qi was determined. It can be seen that the slope (tso/Qi), when l/(t-to) —» 0, is decreasing with temperature and hence the reaction rate is increasing with temperature. The estimated final heats (Qj) and times (ts0) are shown in Table XI. When the temperature increases from 25 to 70 °C, the time at which 50% of the Chapter 5.2 - Hydration Study 102 cement is hydrated decreases from 20.7 to 3.5 hr, meaning that the reactivity increases with temperature. 0.40 --0.45 • • 0.50 J 1 1 1 1 1 1 1 1 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00 1 /(t-to) —•—25 °C -a— 40°C - * - 5 0 ° C ~*~G0°C - * — 70°C Fig. 34 - Reciprocal of heat liberation vs. reciprocal of reaction time for PC at various temperatures Table X I - Values of Q; and I50 for Portland cement Temp. 25°C 40°C 50°C 60°C 70°C Qi (J/g) 221 261 380 349 465 tso (hr) 20.7 11.1 6.0 5.5 3.5 Subsequently, the apparent activation energy (Ea) can be calculated by using two temperatures. The 25 and 40 °C isotherms were chosen in this case, as for these temperatures in particular the final heat is almost the same, which translates to a variation of the final heat with temperature close to zero. This means that the heat release is almost independent of temperature, and accordingly the activation energy (Ea) is independent of temperature, an assumption on which the Chapter 5.2 - Hydration Study 103 Arrhenius relationship is based. This approach has also been used by Wu et al. as well as by other researchers [84, 87, 90]. 20.69 11.11 ln 1.86= 1.934-105-Ea E a= 32 KJ/mole This result is comparable to the apparent activation energies determined by other researchers using calorimetric measurements, e.g. 33 KJ/mole [86, 91] and 39-44 KJ/mol [8, 85]. In addition to the Knudsen's method, two other approaches, the Avrami and the Jander models [62, 92, 93], were used in order to calculate and verify the apparent activation energy and to identify the hydration mechanisms. The shapes of the calorimetric curves are typical of those for phase transformations occurring by nucleation and growth and many researchers have attempted to analyze them by using classical kinetic theory [93, 94, 95]. This requires a general form of the Avrami equation: — ln(l - a)" = Jt(r — r0) ( 1 2 ) where a is the fraction reacted, k is the reaction rate constant, t is time, t0 is the induction time and n is a constant. The fraction reacted or the degree of hydration, defined as a=Q/Gj, can be calculated using the Q; value obtained in the Knudsen procedure. Knowing the final heat, the degrees of hydration at different times are then obtained [61]. However, it must be noted that these degrees of hydration will not coincide with the ones calculated by other methods, such as the quantification of the unhydrous slag remaining unreacted. Fig. 35 shows a plot of the Avrami equation in logarithmic scale. From the slope of the resulting line the reaction order n was calculated. Knowing n, a mean value for k can be obtained [62] by calculating the reaction rate ~25° \M1° 40° -exp Ea (313 -298) 8.314-298-313 Chapter 5.2 - Hydration Study 104 constant at different times for each temperature; it must be borne in mind that the resulting rate constants are not absolute values but are a function of the equation used. On the other hand, the modified Jander model uses the equation: ( l - ( l - a ) 1 / 3 ) n =kjt (13) where, a = degree of hydration, kj = rate constant, t = hydration time, and n varies according to the type of reaction governing the process. The value of n is calculated by plotting ln(l-(l-a) 1 / 3) vs. ln(t) and determining the slope (1/n). When n<\ the process occurs through nucleation, when n=\ by phase boundary kinetics and when n=2, by diffusion. Fig. 36 shows the application of the Jander equation to the PC experimental data, while Table XII summarizes the kinetic parameters. Fig. 35 confirms that the Avrami model can be successfully used in dealing with Portland cement hydration in both acceleration and deceleration periods at periods under 24 hrs. This is in agreement with results reported in the literature by Zeng et al. [93]. The constant n is less than 1 between 25 and 60 °C, indicating the reaction occurs by a nucleation kinetics. At 70 °C it is slightly greater than 1, indicating a change to a phase boundary kinetics. The kinetic parameters are similar to those from a previous study on PC [93]. Fig. 36 indicates that the Jander model fits data for both the acceleration and the deceleration periods. However, it must be noted that the n values are >2, a value typical for diffusion controlled processes. This fact could signal the existence of a more complex hydration mechanism. The high n values can be explained by the fact that PC is a polyphase material [87]. Several phases are independently reacting with water at the same time and at different rates. The final n value, according to basic kinetics theory, represents the sum of individual n values for each of the phases present in the cement grains [96]. In their work, Mori et al. [62] also reported n values of 3 and 4 and associated the high value of n Chapter 5.2 - Hydration Study 105 with the formation of a dense coating around the grains and consequently an increased resistance of the coating to diffusion. The reaction rate constants are within the order of magnitude reported by other researchers [62, 93]. The rate constants change slightly with temperature. Mori [62] theorized that these small differences in values could be attributed to changes in the physical properties of the hydrate coating and not to the effect of temperature on the diffusion process. Avrami model 1.5 i 0.0 1.0 2.0 3.0 ln(t-to) Fig. 35 - Determination of n and &for PC using the modified Avrami model Chapter 5.2 - Hydration Study 106 Table XII - Constants n and k for Portland cement using the modified Avrami and Jander models Avrami Jander Temp. °C n Eq. n fcy*10"5 (s-1) Eq. 25 0.62 1.06 y25 = 1.6236x-4.4559 R2 = 0.9997 4.65 0.81 yi5 = 0.2151x-0.3135 R2 = 0.9981 40 0.72 2.60 V40 = = 1.3882x- 2.884 R2 = 0.9954 6.43 1.13 y4o = 0.1554x-0.1941 R2 = 0.9881 50 0.84 3.15 V50 = 1.1868x-2.3879 R2 = 0.9838 5.91 1.66 y5o = 0.1691x-0.2036 R2 = 0.9932 60 0.90 3.85 yeo = 1.1023x- 2.1099 R2 = 0.9624 5.53 2.73 yeo = 0.1807x-0.2086 R2 = 0.9958 70 1.16 5.56 y7o= 0.8599x- 1.5377 R2 = 0.9824 7.54 4.72 yvo= 0.1326x-0.1621 R2 = 0.9935 Chapter 5.2 - Hydration Study 107 Since the reaction rate constant is dependent on temperature in a form given by the Arrhenius equation: k = Aexp(-£ f l I RT) (14) the slope of lnfc vs. 1/T multiplied by the universal gas constant R will give the activation energy E a . The resulting apparent activation energies (Fig. 37 and 38) are 28.80 KJ/mole (Avrami) and 32.44 KJ/mole (Jander). These values confirm the 32 KJ/mole obtained earlier via the Knudsen approach. Fig. 37 - Apparent activation energy for PC using the Avrami model Chapter 5.2 - Hydration Study 108 Arrhenius plot for PC using the Jander model calculations -9.60 -] -10.00- y=-3.9889x+1.4994 1^  = 0.9483 -10.40-1 -10.80--11.20 --11.60--12.00- 1 1 1 1 2.90 3.00 3.10 3.20 3.30 3.40 1/TXxlO3) Fig. 38 - Apparent activation energy for PC using the Jander model The fact that both models fit the experimental data confirms previous studies [10, 62, 93] on PC showing that the hydration process occurs by nucleation and growth, as well as by diffusion. In polysize specimens the hydration of all grains is not necessarily governed by one and the same rate-determining process at the same time, and the effects of simultaneous action of different reaction mechanisms should be taken into account [87, 103]. It has been theorized that cement grains do not hydrate at the same rate and some will be in the nucleation stage of hydration, while others would be in the diffusion stage through an already formed hydrate layer. Moreover, temperature slightly increases the rate of reaction and hence, it is probable, that both nucleation and growth and diffusion processes may overlap in the acceleration as well as in the deceleration periods. Present results suggest this condition is likely to occur in PC hydration and confirm the findings of Zeng et al. [93]. Chapter 5.2 - Hydration Study 5.2.2 Heat of Hydration of Synthetic Steel Slags 109 This section presents calorimetric results for 3 synthetic steel slags, and a detailed kinetic analysis for one of the slags (SS330) in 2 distinct forms: quenched (amorphous) and annealed (crystalline). Quenched slag (amorphous) As in Portland cement, quenched slags displayed two peaks. For ease of curve smoothing, the first peaks are not shown in the graphs. At 25 °C, the heat release was almost insignificant and the sample did not set in the short time interval chosen (24 hr). The slag hydrates at very low rates and impractical long times would be required to measure the heat evolved at room temperature. This is not uncommon, as slags are know to hydrate at longer times and are added to cement for this reason in particular, to contribute to the late strength development. At temperatures above 40 °C all samples hydrated and were removed with difficulty from the sample holder, a phenomenon which indicates setting has occurred. Fig. 39 shows the heat evolution in slag SS330. The material hydrates within the 24 hr time frame and the heat evolved in the process is increasing with temperature, while the maximum heat release shifts towards lower times. In addition, it can be noted that the setting periods are shorter (between 3-7hr) than those of PC (between 8-14hr). Table XIH gives more details on the heat release process. Chapter 5.2 - Hydration Study 110 Hydration of slag SS330 (quenched) at different temp, (no activator added) 0.05 -i Time (hr) Fig. 39 - Heat evolution rate curve for the amorphous form of the synthetic slag SS330 Table XIII - Heat of Hydration at 24 hr for synthetic slag SS330 (amorphous) Temperature (°C) Total Heat of Hydration, at24hr (J/g) Initial Heat (J/g) Heat of Setting (J/g) Time at which Maximum Heat Release Occurs (hr) 25 24 2 22 8.1 40 176 6 170 8.8 50 197 8 189 3.5 60 225 9 216 1.3 70 253 9 244 0.8 Fig. 40 shows the cumulative heat with time. The total heat evolved increases as the temperature increases. A tendency towards reaching a plateau can be seen at 40, 50 and 60 °C. Chapter 5.2 - Hydration Study Curnulative heat vs. time for slag SS330 quenched 300 n Fig. 40 - Cumulative heat for SS330 (amorphous) at different temperatures 0.50 H — . — . — i — i — I — . — , — , — > — i — , — i — . — , — I — , — . — , — , — I 0.40 0.30 0.20 0.10 0.00 l/«-to) - * - 2 5 ° C m 40°C - * 50°C - » - 6 0 ° C - » - 7 0 O C Fig. 41 - Double reciprocal of heat liberation and reaction time for SS330 (amorphous) Chapter 5.2 - Hydration Study 112 The slopes of the reciprocal curves (Fig. 41) are more temperature dependent for the synthetic steel slag than for the Portland cement (Fig. 34). This coincides with results reported for blast furnace slag [93] and it is indicative of high activation energy processes, the latter being more temperature dependent. Raising the temperature is more beneficial for the reactions with higher activation energy than for those with lower activation energy [72]. This fact stems from the Arrhenius equation written in the following form: 1 , =—V (15) dT RT2 the expression ti[ln kT ] / dT, at a given temperature T, is proportional to the activation energy Ea. Raising the temperature would be more beneficial for the processes with higher activation energies (e.g. blast furnace slags) than for those with lower reaction activation energies (Portland cement). Table XIV provides the Knudsen estimates for the amorphous slag. Table X I V - Values of Q; and tso for synthetic slag SS330 (amorphous) Temp. 25°C 40°C 50°C 60°C 70°C Qi (J/g) 27 191 205 239 258 tso (hr) 22.10 9.04 4.95 2.32 1.63 As anticipated by the heat evolution rates curves, the times corresponding to a 50% degree of hydration (tso) decline with temperature, while the final heat increases. The apparent activation energy (E a) calculated between 40 and 50 °C is 49.61 KJ/mole. This value is not far Chapter 5.2 - Hydration Study 113 from the E a reported for mixtures of cement + 80 % granulated blast furnace slag of 50-58 KJ/mole [61, 87]. Avrami and Jander models were further applied to corroborate the E a and to identify the mechanisms involved in the hydration of the amorphous synthetic slag. The graphs used for determining the constants n and k are shown in Fig. 42(a), 42(b), 43(a) and 43(b). At this point it should be remembered that the calorimetric curve can be divided into five stages [10]: stage I (initial hydrolysis), stage U (induction period), stage HI (acceleration), stage TV (deceleration) and stage V (steady state or decay period) as seen schematically in Fig. 44. Most of the hydration occurs during acceleration and deceleration periods. As the literature is not giving clear information on the kinetics of glassy slags hydration, present results are analyzed against the kinetics for calcium silicates hydration (Table X V ) . Chapter 5.2 - Hydration Study 114 Avrami model ln(t-to) - « - 4 0 o C - & - 5 0 ° C - * - 6 0 ° C - * - 7 0 ° C Fig. 42 (a)- The Avrami model for slag SS330 (amorphous) over the entire 24 hr test interval Fig. 42 (b) - The Avrami model for slag SS330 (amorphous); data fitting in the acceleration and deceleration periods Chapter 5.2 - Hydration Study 115 • Jander model 0.0 --d -0.2 --j -0.3 ---0.4 ---0.5 --3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 4.0 bi(t-to) - m - 4 0 ° C - * - 5 0 ° C - * - 6 0 ° C -at- 70°C Fig. 43 (a) - The Jander model applied on data collected for slag SS330 (amorphous) over the entire 24 hr test interval Chapter 5.2 - Hydration Study 116 \f Stage I Stage n Stage m Stage ET Stage Z \ T I M E Fig. 44 - Rate of heat evolution during the hydration of tricalcium silicate [10] Table X V - Sequence of Hydration of the Calcium Silicates [10, 40] No. Reaction Stage Kinetics / Rate Control Chemical Process Relevance to Concrete Properties I Initial hydrolysis Very rapid / Chemical Initial hydrolysis; dissolution of ions -n Induction period Slow / Nucleation or diffusion Continued dissolution of ions, formation of early C-S-H Determines the initial set m Acceleration Rapid / Chemical Initial growth of hydration products Determines final set and rate of initial hardening TV Deceleration Moderate / Chemical and Diffusion Continued growth of hydration products; development of microstructure Determines rate of early strength gain V Steady state (decay) Very slow / Diffusion Slow formation of hydration products; gradual densification of microstructure Determines rate of later strength gain Hydration is most sensitive to temperature in stage DI when the reaction is chemically controlled. In stage V , when completely diffusion controlled, it is much less temperature sensitive, although the diffusion coefficient of the hydrate barrier will vary with temperature. Chapter 5.2 - Hydration Study 117 Looking at the Avrami and Jander curves for slag, it can be noticed they have an S shape. This curve shape was also obtained by Mori [62], when analyzing C3A hydration, and using the Jander model in the second part of the curve related to acceleration and beginning of decay, and obtaining a roughly linear portion of the curve. The results of the kinetic analysis for slag SS330 in amorphous form are presented in Table X V I . Table X V I - Constants n and k for slag SS330 (amorphous) employing the modified Avrami and Jander models Avrami Jander Temp. ° C n *A*10- 5 (S 1 ) E q . n kj*W5 (s"1) E q . 25 - - - - - -40 0.33 0.31 y 4 0 = 3.0101x- 5.774 R 2 = 0.9835 2.66 0.36 y 4 0 = 0.3759x - 0.8422 R 2 = 0.992 50 0.31 0.82 y 5 0 = 3 .1865x- 4.2206 R 2 = 0.947 2.93 0.80 y 5 0 = 0 .341x - 0.5769 R 2 = 0.9737 60 0.43 1.05 y60 = 2.3348x- 1.5062 R 2 = 0.9751 5.16 1.17 yeo = 0 .1936x -0 .2491 R 2 = 0.9971 70 0.55 2.10 y 7 0 = 1 .8157x-0 .0677 R 2 = 0.9372 5.77 1.98 y 7 0 = 0.1731x- 0.1415 R 2 = 0.9739 The duration of the setting stage (acceleration and decay periods) varies in length with temperature. At 40 °C, this time is approx. 5 hr and decreases to 2.5 hr at 70 °C. Since both the Avrami and Jander models apply for the acceleration and deceleration periods, the hydration of slag SS330 in quenched form occurs by both nucleation and diffusion. Studies done by Fernandez et al. [61] on granulated blast furnace slag show that a diffusion mechanism is present, however these researchers opted for the general Jander equation and didn't apply the Avrami equation to determine its suitability. Nonetheless, Fierens et al [94, 95] show the possibility of encountering nucleation and growth, phase boundary reaction and diffusion in the last part of the hydration process of P-C2S over a 15 hr time period and suggested that the kinetic mechanism Chapter 5.2 - Hydration Study 118 lies between these processes. At this point it should be remembered that P-C2S is one of the main constituents found in our synthetic steel slags. Figs. 45 and 46 give the Arrhenius relationships. The apparent activation energies are 52.32 KJ/mole (Avrami) and 48.17 KJ/mole (Jander). These values are in the same order of magnitude as the one calculated by the Knudsen method (47 KJ/mole). Fig. 45 - Arrhenius relationship using the Avrami calculations for slag SS330 (amorphous) Chapter 5.2 - Hydration Study 119 Arrhenius plot for amorphous steel slag using the Jander model calculations y = -5.9219x+6.4626 R* = 0.9807 2.90 2.95 3.00 3.05 3.10 3.15 3.20 1/ITxlO5) Fig. 46 - Arrhenius relationship using the Jander calculations for slag SS330 (amorphous) Slags SS340 (quenched) and SS430 (quenched) present the same behavior as slag SS330. Their evolution rate curves are provided in the next section when comparing to the crystalline forms. Annealed slag (crystalline) The highest heats of hydration for crystalline slags, obtained at 70 °C, are around 150 J/g. These values are approx. 40 % lower than for the glassy forms. At 40 °C the heat evolved is about 55 % lower than at 70 °C. Also, the level of noise in the data acquisition for these samples represents a much larger fraction of the results. Beyond 50 °C, data acquisition becomes more intensive and stable. This tendency evidences the need for a higher "thermal activation" for the crystalline slags or else a greater temperature dependence. Figs. 47 and 48 indicate the heat flow release. Chapter 5.2 - Hydration Study 0.010 0.008 -f Hydration of slag SS330 (crystallire) at different temp. Fig. 47 - Heat evolution rate curves for synthetic slag SS330 (crystalline) Cumulative heat vs. tire for SS330 crystalline 160 i 0 5 10 15 20 25 Time (br) 25 °C - • - 40°C -tr- 50°C - • - 60°C - * - 70°C Fig. 48 - Cumulative heat for SS330 (crystalline) at different temperatures Chapter 5.2 - Hydration Study 121 The setting curves appear at longer times (> 7hr) and the total heats are lower than for the quenched form. Table X V U provides more data on the hydration behavior of the crystalline slag. The times at which maximum heat release occurs at lower temperatures (25 and 40 °C) are shorter than for higher temperatures. However, it must be noted that these values correspond to a rather short 24 hr test time and they will occur at longer times if hydration time is extended. Table XVII - Heat of Hydration at 24 hr for synthetic slag SS330 (crystalline) Temperature (°C) Total Heat of Hydration, at 24 hr (J/g) Initial Heat (J/g) Heat of Setting (J/g) Time at which Maximum Heat Release Occurs (hr) 25 36 5 31 8.5 40 83 7 76 6.1 50 107 7 100 19.8 60 131 7 124 16.0 70 149 7 142 10.7 0.00 0.45 + 0.50 -I 1 1 1 1 0.40 0.30 0.20 0.10 0.00 1 / (t-to) —•— 25 °C 40°C 50°C B 60°C -*— 70°C Fig. 49 - Reciprocal of heat liberation vs. reciprocal of reaction time in SS330 various temperatures (crystalline) at Chapter 5.2 - Hydration Study Table XVIII - Values of Qi and tsofor synthetic slag SS330 (crystalline) 122 Temp. 25°C 40°C 50°C 60°C 70°C Qi (J/g) 56 99 131 161 167 tso (hr) 48.92 19.19 18.68 23.52 11.52 Using the times t 5 0 at 60 and 70 °C (Fig. 49 and Table XVIJJ), the activation energy is calculated as 66 KJ/mole. The E a for the crystalline slag is higher than for the amorphous one. This is in agreement with the thermodynamic theory suggesting that disordered structures and undercooled liquids, such as glasses, have a higher free energy of formation and hence the E a necessary to overcome the energetic barrier for a reaction to proceed is lower for the amorphous material. (This assumes that both materials pass through the same activated state and that the same products are being formed). Both Avrami and Jander models were successfully used. Figs. 50 and 51 and Table X I X show the results. Avrami model In(t-to) Fig. 50 - Avrami model for slag SS330 (crystalline) Chapter 5.2 - Hydration Study 123 Jander model 0.10 - 1 -0 .40 -\ 1 1 1 1 1 1 0.0 0.5 1.0 1.5 2.0 2.5 3.0 In(t-to) Fig. 51 - Jander model for slag SS330 (crystalline) Table X I X - Constants n and k for synthetic slag SS330 (crystalline) using modified Avrami and Jander models Avrami Jander Temp. ° C n k*W6 (s 1 ) E q . n k*l0-6 (s 1 ) E q . 25 1.67 - y25 = 0.5992x-2.14 R 2 = 0.9883 7.16 - y 2 5 = 0.1396x-0.5698 R 2 = 0.968 40 2.22 0.29 y4o = 0.4503x-0.66 R 2 = 0.9625 16.69 0.25 y 4 0 = 0.0599x-0.22 R 2 = 0.978 50 1.56 0.83 y5o = 0.6417x-0.62 R 2 = 0.9925 12.31 0.41 y 5 0 = 0.0812x-0.20 R 2 = 0.9845 60 1.61 1.68 yeo = 0.621x-0.5359 R 2 = 0.9935 13.96 0.58 yeo = 0.0716x-0.1896 R 2 = 0.9927 70 1.12 1.54 y?o = 0.893x-1.119 R 2 = 0.9966 8.27 1.69 y 7 0 = 0.1208x-0.3148 R 2 = 0.9732 Chapter 5.2 - Hydration Study 124 -10.00 -10.50| -11.00 -11.50 -12.00 -12.50 -13.00 -13.50 + -14.00 -14.50 -15.00 2.9 Anferius plot for crystake steel slag using the Jander model 2.95 y = -8.7025x+13.629 R*= 0.9248 3.05 1/IXxlO3) 3.1 3.15 3.2 Fig. 52 - Arrhenius plot for slag SS330 (crystalline) By applying Jander's model the E a (see Fig. 52) is 71 KJ/mole. The other two crystalline slags, SS340 and SS430 present a similar behavior to slag SS330. The crystalline forms were always less reactive than the amorphous ones (Fig. 53, 54, 55 and 56) producing less total heat and at longer times and with a higher apparent activation energy. Table X X shows the Knudsen estimates and the resulting activation energies, calculated between 60 and 70 °C, for slags SS340 and SS430 in both amorphous and crystalline form. Table X X - Values of Qi, tso and E a for synthetic slags SS340 and SS430 in amorphous and crystalline form Slag 50' C 60< 'C 70c C E a (KJ/mole) Qi(J/g) t » (hr) Qi (J/g) tso (hr) Qi (J/g) tso (hr) SS340 (amorph.) 124 11.32 165 5.45 163 3.2 50.56 SS340 (crystal) 64 9.86 81 4.48 85 2.15 69.71 SS430 (amorph.) 137 3.60 140 2.65 158 1.63 46.15 SS430 (crystal) 49 9.37 72 2.44 87 1.17 69.79 Chapter 5.2 - Hydration Study 125 Hydration of slag SS340 (quenched) at different temp, (no activator added) 0.030 T i m e ( h r ) Fig. 53 - Hydration of slag SS340 (quenched) at different temperatures Hydration of slag SS340 (crystalline) at different temp, (no activator added) 0.030 ~ 0.025 + E 10 o > tv tv 0.020 0.015 0.010 0.005 0.000 70° C ^ 60° C 5 0 o C 40° C 25° C 9 13 17 T i m e ( h r ) 25 Fig. 54 - Hydration of slag SS340 (crystalline) at different temperatures Chapter 5.2 - Hydration Study Hydration of slag SS430 (quenched) at different temp, (no activator added) 0.050 Time (hr) Fig. 55 - Hydration of slag SS430 (quenched) at different temperatures Hydration of slag SS430 (crystalline) at different temp, (no activator added) Fig. 56 - Hydration of slag SS430 (crystalline) at different temperatures Chapter 5.2 - Hydration Study 127 The t 5 0 values for crystalline slags SS340 and SS430 (Table X X ) are higher than for the amorphous ones, however, the difference can be regarded as not significant (~1 hr.). The E a for Portland cement is 32 KJ/mole. Compared to PC, the synthetic steel slags in amorphous form have a higher activation energy (~ 50 KJ/mole) and the E a is even higher for the crystalline form (~ 70 KJ/mole). The amorphous material is more reactive than the crystalline one and the final heat of hydration at 24 hr is almost 50 % higher in the amorphous material. Therefore, it is clear that steel slags, like blast furnace slags, would have to be quenched in order to be suitable for cement-making [4, 8, 9, 18, 29]. It must be noted that the amorphous (water quenched) samples in this study contained approximately 60 % of glassy phase which made them very reactive, even without the addition of activators. On the other hand, the amount of glassy phase in commonly granulated blast furnace slags is 67 % minimum and may reach 85-90 % [20, 56]. The mechanism of hydration for PC and the synthetic slags is a combination of nucleation and growth, and diffusion through the hydrated layer. The hydration mechanism in quenched and crystalline slags is also a combination of nucleation and growth and diffusion with a difference in the activation energy required for the reaction to take place. This is in agreement with previous reports dealing with the presence of a glassy phase [80, 87, 97]. Moreover, a combination of mechanisms might be supported by the fact that slag grains are multiphase and polysize materials and that in the early hydration period cementitious components present in the slag grains are both hydrating independently, and at different rates. There is therefore the possibility of having slag grains in different stages of hydration at any moment. In must be emphasized that the kinetic study was applied to early hydration, particularly in the acceleration and deceleration periods, where most of the hydration occurs. Exhibiting considerably lower heat release, and hence lower hydration rates, the decay period would certainly have to be controlled only by diffusion. Chapter 5.2 - Hydration Study 128 To illustrate the fact that species present in a polyphase material may hydrate differently, three calorimetric curves are added. They correspond to C4AF, C3A and C3S which are basic cementitious compounds found in cement clinker and to some extent in the steel slags. It can be seen that the C 4 A F and C 3 A curves (Fig. 57 and 58) do not follow the same pattern as the C3S curve (Fig. 59). Apart from showing different heat levels, C 4 A F and C 3 A do not display the classical induction period. Moreover, at any given time, each of this compounds will be in a different hydration phase and hence a different hydration mechanism will be dominant. Chapter 5.2 - Hydration Study 129 Hydration of C 3 A (crystalline) at different temp, (no activator added) 0 5 10 15 20 25 Time (hr) Fig. 57 - Calorimetric curve of C3A at different temperatures Hydration of C 4AF (crystalline) at different temp, (no activator added) 0.10 TT c" 0.08 -\ E i • 0 0 6 •{ 0 5 10 15 20 25 Time (hr) Fig. 58 - Calorimetric curve of C4AF at different temperatures Chapter 5.2 - Hydration Study Hydration of C 3S (crystalline) at different temp, (no activator added) 0.030 ~ 0.025 + I n 0) 0.020 + 0.015 + c o I 0.010 > tv 0.005 0.000 70° C 10 15 Time (hr) 25 Fig. 59 - Calorimetric curve for C3S at different temperatures Chapter 5.2 - Hydration Study 131 5.2.3 Hydration Behavior of Industrial Steel Slags Samples of oxidized, remelted and water quenched steel slags (Alta and Stelco) were also analyzed via calorimetry. Their chemistry is presented in Appendix I. The curves display two peaks as in the synthetic steel slags case. When in quenched form, at 24 hr, 70 °C and in the presence of 2 % CaCl2 total heat evolution is ~ 70 J/g. It should be pointed out that setting did occur and that samples were removed only with difficulty from the holder. When tested in annealed (crystalline) form, both slags exhibited much lower final heats (-19 J/g) and no setting was observed. One sample showed a tendency towards increased heat flows after 40 hr of testing, indicating the need for a higher "thermal activation" of the slag or for longer hydration times. The higher heats for the quenched industrial slags confirm the results obtained earlier for the synthetic slags, and again confirm that amorphous materials are more reactive than crystalline ones. It might be possible that the heat values obtained with these slags were affected by the addition of CaCl2 and that, in the presence of a different activator, the values could be higher. The detrimental effect of CaC^ on slags hydration is discussed in a separate section. In addition, a granulated blast furnace slag was also tested in the same test conditions as the oxidized-quenched industrial steel slag. The final heat value is comparable to the steel slag (66 vs. 70 J/g). Although the literature does not provide values for steel slag hydrolysis, a value of 56 J/g has been reported for PC + 90% blast furnace slag at 24 hr [89], which is in fair agreement with the present results. However, it should be noted that the relationship between the values found for blast furnace slag and steel slag is fortuitous since the chemistry and the degree of crystallinity of the two materials are markedly different. Chapter 5.2 - Hydration Study 132 Moreover, the steel slags studied in the present work are hydraulic without the need of adding an activator. This raises the possibility that steel slags could act as an activator for blast furnace slag and that steelmaking industry could manufacture its own cementitious material (steel slag + blast furnace slag and, possibly Na 2Si03 as an activator, without adding Portland cement). 5.2.4 Hydration of Mixtures of Portland Cement and Alta Steel Slag (quenched) This section deals with the results obtained in a series of tests using mixtures of Portland cement, Alta steel slag (oxidized and quenched) and gypsum ( C a S 0 4 - 2 H 2 0 ) . The mixtures contained 0, 10 and 25 % slag and 0, 2.7 and 5.4 % SO3, the latter being typical levels for ordinary Portland cement. o > u Hydration of Portland cement - Alta steel slag (quenched), at 25 °C and without activator 0.010 0.008 0.006 -: 0.004 4 I 0.002 4-J 0.000 i-0 % Alta slag Total heat (J/g) at 24 hr: - 228.28 Time (hr) •10% Alta slag -166.02 25% Alta slag -126.90 Fig. 60 - Calorimetric curves for mixtures of PC and Alta steel slag (quenched) Chapter 5.2 - Hydration Study 133 As expected, the addition of slag decreases the final heat at 24 hr and 25 °C. Fig. 60 indicates the heat evolution curve, as well as the final results. The shape of the curves do not change with slag addition and it appears that at these short times, the slag had not started to hydrate. Based on prior research [4, 61, 81], it can be inferred that the slag will react beyond the 24 hr time period, and hence contribute to the late strength development. Hydra t i on o f P o r t l a n d c e m e n t - A l ta s l a g m i x t u r e s (24 hr , 25 °C) Slag a d d i t i o n (%) Fig. 61 - Effect of slag and S 0 3 (gypsum) addition Table X X I - Heat released in mixtures of PC and Alta steel slag Slag addition (%) Heat released at 24 hr, (J/g) 0 % S O 3 2 . 7 % S 0 3 5 . 4 % S Q 3 0 228 - -10 166 164 166 25 127 142 149 Chapter 5.2 - Hydration Study 134 Fig. 61 and Table X X I indicate the final heat for different mixtures containing gypsum. It can be noted that the levels of gypsum used in the tests do not have a marked effect on the heat release process and do not retard the reaction. Overall, slag addition will decrease the heat release at early ages. 5.2.5 Hydration of Synthetic Steel Slags in the Presence of CaCl? In an attempt to accelerate the heat release and shift it to lower times, CaCl 2 was added to the synthetic slags as an activator. It should be noted that CaCl 2 is a common activator for PC and is employed when pouring concrete in very cold climates. Figs. 62, 63, 64, 65 and Table XXII show the results for CaCl 2 addition. Table XXII - Final heat release for slags SS330, SS340 and SS430 in the presence of CaCl 2 Temp. (°C) CaCl 2 (%) SS330 quenched (J/g) SS330 crystalline (J/g) SS340 quenched (J/g) SS340 crystalline (J/g) SS430 quenched (J/g) SS430 crystalline (J/g) 25 2 22 16 8 10 11 35 40 2 58 - 8 32 59 86 50 2 74 117 10 53 78 88 60 2 152 87 12 62 95 86 70 2 155 56 22 74 118 102 Because it is commonly employed as an activator for PC, one might expect a heat flow increase when using CaCl 2 in synthetic slags. However, the experimental results are consistent and indicate lower heat evolution with CaCl 2 than without. If a comparison is made between Chapter 5.2 - Hydration Study 135 amorphous and crystalline forms, the results are still unclear. For example, slag SS330 in amorphous form releases more heat than the crystalline form, at all temperatures with the exception of 40 °C. Slag SS430 gives similar values for both amorphous and crystalline forms, while slag SS340 is more reactive in crystalline form. In principle, the addition of CaCl2 "thermally activates' the hydration of the slag, raising the temperature of the paste. A calorimetric test done on CaC^ showed this material releases approx. 1300 J/g when in contact with water. If the heat inherent to the CaCh addition itself is subtracted from the final heat it appears that the synthetic slags are inert at temperatures below 40 °C, which is in contradiction with numerous tests done on synthetic slags without activator that showed significant heat release. In general, the results suggest that CaCb addition is more beneficial (in terms of accelerating the hydration process) for highly crystalline slags. The reason for this is not clear. In terms of overall results for CaC^ addition, the reduction in heat evolution (relative to slags without CaCb) could be due to the formation of a dense layer of products immediately after the CaCh rather violent reaction which could hinder further hydration. The dilemma might be solved by closer examination of the hydration products and the characteristics of the gel, nevertheless, from a practical point of view, CaCl 2 is not a good option for activating steel slags. Moreover, the literature reports that CaCl2 could act as either accelerator, or retarder (particularly for C3A and C4AF), depending on the amount and type of cementitious material [98, 99, 100]. This might also be a reason for low heat values in the high iron content steel slags. Chapter 5.2 - Hydration Study Hydration of slag SS330 (quenched) at different temp. (2%CaCI 2 added) 0.0100 0.0000 70° C 60° C 25° C 25 Fig. 62 - Hydration curve for slag SS330 (quenched) with 2 % CaCl 2 Hydration of slag SS330 (crystalline) at different temp. (2%CaCI 2 added) 0.0100 T Time (hr) Fig. 63 - Hydration curve for slag SS330 (crystalline) with 2 % CaCl 2 Chapter 5.2 - Hydration Study Hydration of slag SS430 (quenched) at different temp, (with 2%CaCI2) 0.0250 0.0000 9 13 17 Time (hr) 25 Fig. 64 - Hydration curve for slag SS430 (amorphous) with 2 % CaCl 2 c I 3 c o o > eo a> X Hydration of slag SS430 (crystalline) at different temp, (with 2%CaCl2) 0.0750 0.0500 0.0250 0.0000 70° C 25° C Fig. 65 - Hydration curve for slag SS430 (crystalline) with 2 % CaCl 2 Chapter 6 - CONCLUSIONS 138 C H A P T E R 6 CONCLUSIONS Isothermal calorimetric tests revealed the hydraulic nature of steel slags. An increased reactivity was observed for slags containing high amounts of glass when compared to the same slags in fully crystalline state. Moreover, it was confirmed that the reactivity will vary depending on the chemistry, as well as the glass presence. A summary of the main findings in this work is subsequently presented: 1. High iron bearing steel slags (-30 % Fe203) become hydraulic when in amorphous form without the need to add an admixture or an activator. 2. An apparent activation energy of hydration was quantified for both amorphous and crystalline steel slags. The amorphous form displays a lower E a compared to the crystalline one (47-52 KJ/mole vs. 66-70 KJ/mole). The difference might be explained by the fact that amorphous materials have a higher free energy of formation and hence the barrier for the reaction to occur is lower in this case. The apparent activation energy of hydration is sufficiently high in crystalline steel slags that for practical purposes these materials can be considered non-hydraulic. On the other hand, the lower activation energy in amorphous steel slags enhances their reactivity and renders them hydraulic. 3. The activation energy for hydrolysis of quenched steel slags is higher than for Portland cement. This parallels the case for granulated blast furnace slags, and it is attributed to diffusion being the rate controlling step. However, the hydration mechanism, for both amorphous and crystalline slags, seems to be a combination of nucleation and growth and Chapter 6 - CONCL USIONS 139 diffusion, as models corresponding to these mechanisms fit well the acceleration and deceleration periods of the heat release curves. 4. The activation energy and the final heat of hydration (at 24 hr and 60 °C) for amorphous steel slags are comparable to those for granulated blast furnace slags, in spite of 10 times higher iron levels and higher crystallinity degree in the quenched steel slags. This indicates the potential of using steel slags as a clinker substitute very much like the commonly used blast furnace slags. Chapter 7 - RECOMMENDATIONS 140 CHAPTER 7 RECOMMENDA TIONS The results in this work indicate the possibility of recycling steel slags as a Portland cement additive. However, additional studies are needed before moving to full scale industrial utilization. Some areas of research at both a fundamental and industrial level are suggested below: 1 . Find a technique for the identification of chain anions in silicate glasses accounting for the high hydration rate of glasses. Chain anions are unstable relative to the SiC>4 rings present in the crystalline silicates. The presence of chain anions might explain the high reactivity of glasses with water. 2. Explore a different route, i.e. the trimetylsilylation (TMS) method, to detect the polymerization of cementitious components in steel slag and/or steel slag-cement mixes. It is known that the C3S hydration occurs with the formation of polymeric anions which develop in parallel with the strength increase. Another explanation for the higher reactivity of glassy steel slags and accounting for the strength development could be the development of such polymer structures. 3. Continuing with a more fundamental approach, progress in the slag chemical assessment and the role of the glassy phase could be made by using other techniques such as: nuclear magnetic resonance (NMR) and ultraviolet spectroscopy (UV). These methods could determine the polymerization state and strength of the network bonding of Si and A l in the glass and pastes, and measure "slag basicity" or else said glassification potential. Chapter 7 - RECOMMENDATIONS . 141 4. The literature consulted does not report on the microstructure of steel slag or steel slag-cement pastes. A study on these lines might add to a further understanding of steel slag hydration mechanism by observing the shape, packing factors, etc. Moreover, by working with steel slag-cement mixes, the interaction between the two types of materials at different hydration periods could be observed. 5. Further investigations on different admixtures to be added to steel slag-cement in order to accelerate the hydration rate and broaden the range of application of steel slag-cements from slow setting (e.g. dams, massive constructions, etc.) to fast setting concrete applications. 6. Studies involving the conductivity and rheology of steel slag-cement pastes to better understand the hydration mechanism and flow characteristics. 7. The Bond Index (grindability test) of steel slag and steel slag-clinker mixes would have to be performed to estimate the energy consumption for the grinding process, and to establish the type of grinding (e.g. semi-autogenous), possible restrictions in the amount to be added to the mill or the possibility of grinding slag separately. 8. At an industrial level, the isothermal calorimetry could be used as a method of assessing the quality of slag prior to its mixing with clinker. A database could be developed for such quality control. Tests should be done beyond 60 °C. As pointed out in the present work, as well as by Conjeaud et al [9], a value of approximately 80 J/g at 24 hr and 60 °C would ensure that the slag is hydraulic in nature. Chapter 8 -REFERENCES 142 CHAPTER 8 REFERENCES 1. R. Coale, C. W. Wolhuter, P. R. Jochens and D. D. Howat, "Cementitious Properties of Metallurgical Slags", Cement and Concrete Research", Vol . 3, 1973, pp.81-92. 2. B. Tailing, J. Brandstetr, "Present State and Future of Alkali-Activated Slag Concrete", Mineral Admixtures in Cement and Concrete, Ed. by S. L . Sarkar, Trondhem, 1989. 3. R. W. Nurse, "Principal Phase Equilibria and Formation of Portland Cement Minerals", Proceedings of the 5 t h Intl. Symposium on the Chemistry of Cement, Tokyo, 1968, pp. 77-89. 4. J. N . Murphy, "Recycling Steel Slag as a Cement Additive", Master of Applied Science Thesis, Department of Metals and Materials Engineering, University of British Columbia, Aug., 1995. 5. D. Ionescu, "High Iron Content Glasses-An Alternative in the Use of Electric Arc Furnace Dust", Master of Applied Science Thesis, Department of Metals and Materials Engineering, University of British Columbia, Oct., 1995. 6. D. Meyer, T. R. Meadowcroft, P. Barr, "Recycling Steel Slag in Portland Cement", Internal Report, Department of Metals and Materials Engineering, University of British Columbia, July, 1996. 7. C. D. Hills, C. J. Sollars, R. Perry, "Ordinary Portland Cement Based Solidification of Toxic Wastes: The Role of OPC Reviewed", Cement and Concrete Research, Vol . 23. 1993, pp. 196-212. 8. D. P. Bentz, P. A . Stutzman, E. J. Garboczi, "Experimental and Simulation studies of the Interfacial Zone in Concrete", Cement and Concrete Research, Vol . 22, 1992, pp. 891-902. 9. M . Conjeaud, C. M . George and F. P. Sorrrentino, " A New Steel Slag for Cement Manufacture: Mineralogy and Hydraulicity", Cement and Concrete Research, Vol . 11, 1981, pp. 85-102. 10. D. M . Roy, "Instructional Modules in Cement Science", The Pennsylvania State University, 1985, pp. 95-111. 11. H . F. W. Taylor, "The Chemistry of Cements". Vol . 1, Academic Press Inc, 1964, pp. 37-69. Chapter 8-REFERENCES 143 REFERENCES (cont.) 12. G. C. Bye, "Portland Cement - Composition, Production and Properties", Pergamon Press, Great Britain, 1983, pp. 83-104. 13. E. S. Dana, "Dana's Minerals and How to Study Them". Wiley, New York, 1998, pp. 162-279. 14. D. Barthelmy, "Minerals arranged by the Strunz Classification of Minerals", GEOLIB -Internet Scientific database, http://web.wt.net/~daba/Mineral/strunz.html, Norway, 18 May, 1999. 15. P. Perroud, "Athena Mineralogy: Mineral Databases", Internet Scientific Database, http://un2sg4.unige.ch/athena/mineral/minppcl8.html, Geneve University, Dept. of Mineralogy, Switzerland, 18 May, 1999. 16. R. W. Nurse, "Slag Cements", The Chemistry of Cements, edited by H . F. W. Taylor, Vol . 2, 1964, pp. 42-48. 17. I. Teoreanu, "Bazele Tehnologiei Liantilor Anorganici", Ministerul Invatamintului, Bucarest, 1993, pp. 46. 18. F. M . Lea, "The Chemistry of Cement and Concrete", Chemical Publishing Company Inc., 1975, pp. 89. 19. J. G. D. Steyn, "Some Notes on the Influence of Lime and Magnesia on the Soundness of Iscor Blast Furnace Slag", Symposium on Iron and Steelmaking, South African Institute of Mining and Metallurgy, Johannesburg, Sept., 1965, pp. 67. 20. T. W. Parker and R. W. Nurse, "Investigations on Granulated Blastfurnace Slags for the Manufacture of Portland Blastfurnace Cement" National Building Studies, Dept. of Scientific and Industrial Research, London, 1949, pp. 1-20. 21. H. F. W. Taylor, Discussion of Ref. 3, Cement and Concrete Research, Vol . 7, 1977, pp. 465. 22. G. W. Swarts, L. M . Cook, "Photosensitivity of Reduced Silicate Glasses Containing Iron", 7 t h International Congress on Glass, Jun 26-July 3, 1965. 23. J. C. Bowker, C. H. P. Lupis, P. A . Flinn, "Structural Studies of Slags by Mossbauer Spectroscopy", Canadian Metallurgical Quarterly, Vol . 20, No. 1, 1981, pp. 69-78. 24. A . Monaco and W-K. Wu, "The Effect of Cooling on the Mineralogical Characterization of Steel Slag", Proceedings of the International Symposium on Resource Conservation and Environmental Technologies, 1994, pp. 99-106. Chapter 8 -REFERENCES 144 R E F E R E N C E S (cont.) 25. L. R. Nelson, "Determination of the Possible Causes of a Foaming Slag by its Mineralogical and Textural Characterization", Extraction and Processing for the Treatment and Minimization of Wastes. 1993, pp. 1041-1062. 26. I. Dragoi, S. A Todinca, and S. Bajenaru, "The Study of Hydraulic Potential of Melted Phase form Portland Cement Clinker", Materiale de Constructii, Vol . 25(1), 1995, pp. 12-14. 27. J. Neubauer, R. Sieber, H-J. Kuzel and M . Ecker, "Investigations on Introducing Si and Mg into Brownmillerite - A Rietveld Refinement", Cement and Concrete Research. Vol . 26, N o . l , 1996, pp. 77-82. 28. D. G. Montgomery and G. Wang, "Preliminary Study of Steel Slag for Blended Cement Manufacture", Materials Forum. Vol . 15, 1991, pp. 374-382. 29. Y . Wang and D. Lin, "The Steel Slag Blended Cement", Silicates Industrielles. Vol . 6, 1983, pp. 121-126. 30. S. S. Sun and Y . Yuan, "Study of Steel Slag Cement", Silicates Industrielles. Vol . 2, 1983, pp. 31-34. 31. T. Idemetsu, S. Takayama and A . Watanabe, "Utilization of Converter Slag as a Constituent of Slag Cement", Trans. Japan Concr. Inst., Vol . 3, pp. 33-38. 32. M . Ktihn, P. Drissen and Jurgen Geiseler, " A New BOF Slag Treatment Technology", EOSC'97: 2 n d European Oxygen Steelmaking Congress. Taranto, Italy, 13-15 Oct.,1997, pp. 445-453. 33. P. Drissen and M . Kiihn, "Liquid Slag Treatment Guarantees High product Quality of Steel Slag", Seminar on Economic Aspects of Clean Technologies. Energy and Waste Management in the Steel Industry, Linz, Austria, 22-24 April, 1998. 34. M . Tiifekgi, A . Demirbas and H . Gene:, "Evaluation of Steel Furnace Slags as Cement Additives", Cement and Concrete Research, Vol . 27, 1997, pp. 1713-1717. 35. A . Duda, "Hydraulic Reactions of L D Steelwork Slags", Cement and Concrete Research, Vol . 19, 1989, pp. 793-801. 36. W. Yu-Ji and L . Da-Li, "The Steel Slag Blended Cement", Silicates Industriels, Vol . 6, 1983, pp. 121-126. 37. F. Wachsmuth, J. Geiseler, W. Fix, K. Koch and K. Schwerdtfeger, "Contribution to the Structure of BOF-Slags and its Influence on their Volume Stability, Canadian Metallurgical Quarterly, Vol . 20, No.3, 1981, pp. 279-284. Chapter 8 -REFERENCES 145 REFERENCES (cont.) 38. S. Shushan, W. Jianhua, Z. Guilin and L . Yongjun, "Steel Slag Cement", Proceedings of the Mc. Master Symposia. 1995, pp. 253-260. 39. S. Mindess, "Mechanical Performance of Cementitious Systems", Structure and Performance of Cements, edited by P. Barnes, 1983, pp. 319-363. 40. P. Barnes, "Structure and Performance of Cements". Applied Science Publishers Ltd, 1983, pp. 1-69. 41. J. H . E. Jeffes, "The Relationship of Structure and Thermodynamic Properties of Polyanionic Compounds", Canadian Metallurgical Quarterly, Vol . 20, No. 1, 1981, pp. 37-50. 42. G. Mascolo, "Hydration Products of Synthetic Glasses Similar to Blast Furnace Slags", Cement and Concrete Research, Vol . 3, 1973, pp. 207-213. 43. P. S. Silva and F. P. Glasser, "Hydration of Cements Based on Metakaolin: thermochemistry", Advances in Cement Research, 1990, Vol . 3, No. 12, pp. 167-177. 44. M . Wilson, J. G. Cabrera, Y . Zou, "The Process and Mechanism of Alkali-Silica Reaction Using Fused Silica as the Reactive Aggregate", Advances in Cement Research, Vol . 6, No. 23, 1994, pp.117-125. 45. P. W. Brown, "Kinetics of Tricalcium Aluminate and Tetracalcium Aluminoferrite Hydration in the Presence of Calcium Sulfate", Journal of American Ceramics Society, Vol . 76, No. 12, 1993, 2971-2976. 46. S. Brunauer, M . Yudenfreund, I. Odler, J. Skalny, " Hardened Portland Cement Pastes of Low Porosity", Cement and Concrete Research, Vol . 3, 1973, pp. 129-147. 47. I. Soroka, "Portland Cement Paste and Concrete", The Macmillan Press, London, 1979, pp. 28-45. 48. J. A . Forrester, " A Conduction Calorimeter for the Study of Cement Hydration", Cement Technology, 1970, pp. 95-99. 49. S. Diamond, "Hydraulic Cement Pastes: Their Structure and Properties, Proceedings of Conference at Tapton Hall, University of Sheffield, 8-9 April, 1976. 50. D. Viehland, J. F. L i , L. J. Yuan, Z. Xu, "Mesostructure of Calcium Silicate Hydrate (C-S-H) Gels in Portland Cement Paste: Short-Range Ordering, Nanocrystallinity, and Local Compositional Order", Journal of American Ceramic Society, Vol . 79, 1996, pp. 1731-1744. Chapters -REFERENCES 146 REFERENCES (cont.) 51. D. E. Macphee, E. E. Lachowski, F. P. Glasser, "Polymerization Effects in C-S-H: Implications for Portland Cement Hydration, Advances in Cement Research, Vol . 1, No. 3, 1988, pp.131-137. 52. L. J. Parrot, "An examination of the Silicate Structure of Tricalcium Silicate Hydrated at Elevated Temperature", Cement and Concrete Research, Vol . 11, 1981, pp. 415-420. 53. A. Emanuelson, S. Hansen, "Distribution of Iron Among Ferrite Hydrates", Cement and Concrete Research. Vol . 27, No. 8, pp. 1167-1177. 54. J. M . Gaidis, E. M . Gartner, "Hydration Mechanisms JJ", Materials Science of Concrete JJ, Ed. J. Skalny and S. Mindess, American Ceramic Society, 1986, pp. 9-39. 55. F. J. Tang, E. M . Gartner, "Influence of Sulfate Source on Portland Cement Hydration", Advances in Cement Research, Vol . 1, 1988, pp. 67-74. 56. V . M . Malhotra and P. K. Mehta, "Pozzolanic and Cementitious Materials", Advances in Concrete Technology Series, Gordon and Breach Publishers, 1996, pp. 169-171. 57. I. G. Richardson, C. R. Wilding, M . J. Dickson, "The Hydration of Blast Furnace Slag cement", Advances in Cement Research, Vol . 2, No. 8, 1989, pp. 147-157. 58. J. Bijen, H. Pietersen, "Mineral Admixtures: Reactions: Micro-Structure and Macro-Properties", Advances in Cement and Concrete, Proceedings of an Engineering Foundation Conference, Durham, N H , July 24-29, 1994, pp. 292. 59. H . S. Pietersen, "Proceedings of the Fourth International Conference on Fly Ash, Silica Fume, Slag and Natural Pozzolans in Concrete", Ed. V . M . Malhotra, Istanbul, Turkey, 1992, Vol . l ,pp. 795-812. 60. F. P. Glasser, "Chemical, Mineralogical and Microstructural Changes Occuring in Hydrated Slag-cement Blends". Ed. J. Skalny and S. Mindess, 1985, pp. 41-79. 61. A . Fernandez-Jimenez, F. Puertas, "Alkali-Activated Slag Cements: Kinetic Studies", Cement and Concrete Research. Vol . 27, No. 3, 1997, pp. 359-368. 62. H. Mori and K. Minegishi, "Effect of the Temperature on the Early Hydration of the System 3CaOAl 2 0 3 -CaS0 4 -2H20-Ca(OH) 2 -H 2 0", Proceedings of the 5 t h Intl. Symposium on the Chemistry of Cement, Tokyo, Oct., 1968, pp. 349-361. 63. P. Yan and I. Odler, "An Ettringite Cement Based on a C-A-S Glass and Gypsum", Advances in Cement Research, Vol . 7, No. 27, 1995, pp. 125-128. Chapter 8 -REFERENCES 147 REFERENCES (cont.) 64. K. Nakagawa, I. Terashima, K. Asaga and M . Daimon, "Influence of Ca(OH)2 and CaS04 2H2O on Hydration Reaction of Amorphous Calcium Aluminate", Cement and Concrete Research, Vol . 20, 1990, pp. 824-832. 65. I. Odler and P. Yan, "Investigations on Ettringite Cements", Advances in Cement Research, Vol . 6, No. 24, 1994, pp. 165-171. 66. J. MacDowell and F. Sorrentino, "Hydration Mechanism of Gehlenite Glass Cements", Advances in Cement Research. Vol . 3, No. 12, 1990, pp. 143-152. 67. D. Menetrier-Sorrentino, A . Capmas and F. Sorrentino, "Hydration of Amorphous and Crystallized Compounds Based on Monocalcium Aluminate", Proceedings of the UNITECR'91, Ashen, Germany, 1991. 68. O. Hisashi, N . Tsuyuki, T. Hisanomi, O. Machnaga, J. Kasaia," The Hydration Behaviors of Glassy 12CaO-7Al 20 3 in the Solution of CaS0 4 -2H 2 0 or Ca(OH) 2", Ceramics Society of Japan. Vol . 87, 1979, pp. 400-404. 69. K. Nakagawa, I. Terashima, K. Asaga and M . Daimon, " A Study of Hydration of Amorphous Calcium Aluminate by Selective Dissolution Analysis", Cement and Concrete Research. Vol . 20, 1990, pp. 655-661. 70. R. Kondo and S. Ueda, "Kinetics and Mechanisms of the Hydrothermal Reaction of Granulated Blast Furnace Slag", Proceedings 5 t h Intl. Symposium on the Chemistry of Cement, Tokyo, Vol . 2, 1968, pp. 232-248. 71. K. C. Narang and S. K. Chopra, "Studies on Alkaline Activation of BF, Steel and Alloy Slags", Silicates Industriels. Vol . 9, 1983, pp. 175-180. 72. C. Shi and R. L. Day, "Selectivity of Alkaline Activators for the Activation of slags", Cement. Concrete and Aggregates. CCAGDP, Vol.18, N o . l , June 1996, pp. 8-14. 73. X . Wu, W. Jiang and D. M . Roy, "Early Activation and Properties of Slag Cement", Cement and Concrete Research, Vol . 20, 1990, pp. 961-974. 74. C. Shi and R. L . Day, "Acceleration of the Reactivity of Fly Ash by Chemical Activation", Cement and Concrete Research, Vol . 25, N o . l , pp. 15-21. 75. M . R. Rixom, "Chemical Admixtures for Concrete". 1978, Halsted Press, New York, pp. 145-175. 76. A . K. Suryavanshi, J. D. Scantlebury, S. B. Lyon, " The Binding of Chloride Ions by Sulphate Resistant Portland Cement", Cement and Concrete Research, Vol . 25, No.3, 1995, pp. 581-592. Chapter 8 -REFERENCES 148 REFERENCES (cont.) 77. D. L . Griffiths, A . N . F. Al-Qaser, R. J. Mangabhai, "Calorimetric Studies on High Alumina Cement in the Presence of Chloride, Sulphate and Seawater Solutions", Calcium Aluminate Cements-Proceedings of the Intl. Symposium, University of London, 1990, pp. 167-177. 78. C. Shi, R. L. Day, "Acceleration of the Reactivity of Fly Ash by Chemical Activation", Cement and Concrete Research, Vol . 25, N o . l , 1995, pp. 15-21. 79. W. L. De Keyser, N . Tenoutasse, "The Hydration of the Ferrite Phase of Cements", Proceedings of the Intl. Symposium on the Chemistry of Cement, Tokyo, Part JJ, 1968, pp. 379-386. 80. W. Jiang, D. M . Roy, "Hydrothermal Processing of New Fly Ash Cement", Ceramic Bulletin, Vol . 71. No. 4, 1992. 81. R. D. Hooton, "The Reactivity and Hydration Products of Blast Furnace Slag", Suplementary Cementing Materials for Concrete, Ed. V . M . Malhotra, 1987, pp. 247-290. 82. M . Regourd, "Structure and Behaviour of Slag Portland Cement Hydrates", Proceedings of the 7 t h International Congress on the Chemistry of Cements, Paris, 1980, pp. 10-25. 83. J. H . Taplin, "On the Hydration Kinetics of Hydraulic Cements", Proceedings of the 5 t h International Congress on the Chemistry of Cements, Tokyo, Oct., 1968, pp. 337-348. 84. T. Knudsen, "On Particle Size Distribution in Cement Hydration", Proceedings of the 7 t h International Symposium on the Chemistry of Cement, Paris, 1980, Vol . II, pp. 170-175. 85. X . Wu, D. M . Roy, C. A . Langton, "Early Stage Hydration of Slag-Cement", Cement and Concrete Research, Vol . 13, 1983, pp. 277-286. 86. G. De Schutter, L. Taerwe, "General Hydration Model for Portland Cement and Blast Furnace Slag Cement", Cement and Concrete Research, Vol . 25, No. 3, 1995, pp. 593-604. 87. D. M . Roy, G. M . Idorn, "Hydration, Structure and Properties of Blast Furnace Slag Cements, Mortars and Concrete, ACI Journal, Nov.- Dec, 1982, pp. 444-457. 88. L. D. Adams, "The Measurement of Very Early Hydration Reactions of Portland Cement Clinker by a Thermoelectric Conduction Calorimeter", Cement and Concrete Research, Vol . 6, 1976, pp. 293-308. 89. D. C. Killoh, " A Comparison of Conduction Calorimeter and Heat of Solution Methods for Measurement of the Heat of Hydration of Cement", Advances in Cement Research, Vol . 1, No. 3, July, 1988, pp. 180-185. Chapter 8 -REFERENCES 149 REFERENCES (cont.) 90. Z. Huanhai, W. Xuequan, X . Zhongzi and T. Mingshu, "Kinetic Study on Hydration of Alkali-Activated Slag", Cement and Concrete Research. Vol . 23, 1993, pp. 1253-1258. 91. M . Regourd, B. Mortureux, E. Gautier, H . Hornain and J. Volant, "Characterization and Thermal Activation of Slag Cements, "Proceedings of the 7 t h Intl. Congress on the Chemistry of Cement Paris, 1980, pp. Ill-105-111. 92. P. N . Aukett, J. Bensted, "Application of Heat Calorimetry to the Study of Oilwell Cements", BP Research, Sudbury Research Center, Middlesex, England, Internal Report, 1992. 93. S. Zeng, N . R. Short and C. L. Page, "Early-age Hydration Kinetics of Polymer-modified Cement", Advances in Cement Research. 1996, Vol . 8, No. 29, pp. 1-9. 94. P. Fierens and J. Tirlocq, "Effects of Synthesis Temperature and Cooling Conditions of Beta-Dicalcium Silicate on Its Hydration Rate", Cement and Concrete Research, Vol . 13, 1983, pp. 41-48. 95. P. Fierens, Y . Kabuema and J. Tirlocq, "Influence du Milieu de Trempe sur la Cinetique D'hydratation du Silicate Tricalcique", Cement and Concrete Research, Vol . 12, 1982, pp. 191-198. 96. H. E. Avery, "Basic Reaction Kinetics and Mechanisms", The Macmillan Press, 1983, pp. 10. 97. R. Magnan, "Application of Microcalorimetry to Studies of the Hydration of Cements", The American Ceramic Bulletin, Vol . 49, No. 3, 1970, pp. 314-316. 98. H. Akhter, F. K. Cartledge, A . Roy and M . E. Tittlebaum, " A Study of the Effects of Nickel Chloride and Calcium Chloride on Hydration of Portland Cement", Cement and Concrete Research, Vol . 23, 1993, pp. 833-842. 99. S. Tsivilis, G. Kakali, K . Haldeou and G. Parissakis, " A Mathematical Model for the Control of Cement Setting Using Calcium Chloride as Accelerator", Cement and Concrete Research, Vol . 25, No. 5, 1995, pp. 948-954. 100. W. L. De Keyser and N . Tenoutasse, "The Hydration of the Ferrite Phase of Cements", Proceedings of the Fifth Intl Symposium of Cement, Tokyo, 1968, pp. 379-386. 101. H. Uchikawa, "Effect of Character of Glass Phase in Blending Components on Their Reactivity in Calcium Hydroxide Mixture", 8 t h Intl Congress on the Chemistry of Cement, Rio de Janeiro, 1986, Vol . 4, pp. 245-250. Chapter 8 -REFERENCES 150 R E F E R E N C E S (cont.) 102. P. F. G. Banfill, "Superplasticizers for Cement Fondu. Part 2: Effects of Temperature on the Hydration Reactions", Advances in Cement Research, 1995, Vol . 7, No. 28, Oct., pp. 151-157. 103. A . Beziak, "On the Determination of Rate Constants for Hydration Processes in Cement Pastes", Cement and Concrete Research, Vol . 10, 1980, pp. 553-563. 104. L. S. Dent Glasser, E. E. Lachowski, K. Mohan and H . F. W. Taylor, " A Multi-method Study of C 3 S Hydration", Cement and Concrete Research. Vol . 8, 1978, pp. 733-740. 105. V . S. Ramachandran and M . S. Lowery, "Effect of a Phosphonate-based Compound on the Hydration of Cement and Cement Components", Fourth C A N M E T / A C I Intl. Conf. On Superplasticizers and Other Chemical Admixtures in Concrete, Montreal, Canada, 1995, pp. 131-151. 106. A. R. Ramachandran and M . W. Grutzeck, "Effect of pH on the Hydration of Tricalcium Silicate", Journal of American Ceramic Society, Vol . 76, No. 1, 1993, pp. 72-80. 107. K. Fujii and W. Kondo, "Hydration of Tricalcium Silicate in a Very Early Stage", "Proceedings of the 7 t h Intl. Congress on the Chemistry of Cement, Paris, 1980, paper II-93, pp. 362-371. 108. L. E. Copeland and D. L. Kantro, "Hydration of Portland Cement", Proceedings of the 7 t h Intl. Congress on the Chemistry of Cement, Paris, 1980, pp. 387-417. 109. I. Jelenic, "Hydration of B20s-Stabilized a'- and P-Modifications of Dicalcium Silicate, Cement and Concrete Research, Vol . 8, 1978, pp. 173-180. 110. F. Akbari and C A . Pickles, " A Review of the Utilization and Processing of Steelmaking Slags", Proceedings of the 3 r d Intl. Symposium on Waste processing and Recycling in Mineral and Metallurgical Industries, 37' Annual Conference of Metallurgists, Calgary, Aug. 16-19, 1998, pp. 123-179. 111. Cement and Concrete Terminology, Reported by ACI Committee 116, Publication SP-19(90), American Concrete Institute, Detroit, Mich., 1990. 112. A S T M - Standard Test Method for Heat of Hydration of Hydraulic Cement, C 186-94, pp. 144-149. 151 APPENDIX J APPENDIX I 152 Composition of Slags The purpose of the experimental work was to determine the influence of glass presence on the hydration reactivity of steel slags. The study focused on three synthetic slags blended from laboratory grade powders resulting in chemically altered steel slags with the aim to obtain increased amounts of calcium silicates and glassy phase, embracing the calcium silicate regions of the CaO-Si02-Fe203-MgO-Al 203 system (see Fig. 1 page 11). Synthetic blends were used in order to eliminate the compositional variability and extraneous impurities associated with commercial slags. It is well known that a considerable degree of variability exists in commercial slags and this variability may be of consequence in obtaining a cementitious material. However, this study concentrated on the glassification (crystallinity) level, the presence of iron in a trivalent form, and a high binary basicity (CaO/Si02). In addition, samples of industrial steel slags partially oxidized and quenched to obtain increased amount of trivalent iron and glass, but without altering their binary basicity, were also characterized. The overall compositions of the synthetic slags fabricated in this work vs. the industrial ones are given in Table I. Table I - Synthetic Slags Compositions (Weight %) - Chemical Analysis f l , 41 Constituent SS330 SS340 SS430 Alia Steel Slag (oxidized and quenched) Stelco Steel Slag (oxidized and quenched) CaO 40.29 34.21 43.15 32.34 34.90 S i0 2 14.17 12.72 11.49 13.75 17.60 CaO/Si0 2 2.84 2.69 3.75 2.35 1.98 (CaO+MgO/ S i0 2 +Al 2 0 3 2.66 1.78 3.20 1.72 2.19 F e 2 0 3 31.41 33.17 32.11 18.72 25.8 total iron (*) MgO 8.63 7.48 7.86 7.94 11.60 A1 2 0 3 4.91 10.65 4.43 9.65 3.60 MnO - - - 9.77 6.50 T i 0 2 - - - 0.43 0.63-P2O5 - - - 0.23 0.02-FeO - - - 9.10 (*) FeO amount n/a, probably < 10% 153 Fabrication of Synthetic Slags Due to the corrosive nature of the slags and their different basicities, a certain difficulty was experienced in selecting an adequate crucible. In most steelmaking BOF operations the refractory of choice is magnesia. High grade, dense, 300 cc magnesia crucibles were procured to simulate the refractory used in steelmaking. However, these crucibles could not handle the thermal stresses produced when placed in the induction furnace at 1500 °C, and subsequently they were preheated at 900 °C contained in a composite crucible made of silicon-carbide silicon-nitride bonded Crystalon. The space between the two crucibles was packed with alumina powder. Once preheated, the crucibles were introduced into the hot chamber of the furnace at 1000 °C and the temperature was then gradually increased to 1530 °C. Argon was flushed between the carbon susceptor and the crucible to prevent the formation of a reducing atmosphere that could reduce the trivalent iron to divalent iron. Fig. 1, 2 and 3 show the furnace set-up. After 30-40 min in the furnace, slags were removed and poured into a water bath at approximately 7 °C. Then the granulated slag was collected and dried. Samples were taken for microscopic analysis while the remainder was ground in a ceramic mill and sized to minus 325 mesh for x-ray diffraction and hydration tests. This particle size was chosen in order to match the Tilbury Portland cement sample used as a reference in this study. The data sheet supplied by the cement manufacturer indicated that the average fineness of several of their batches was 3.8% +325 mesh [4]. In addition, two other sources [56, 100] suggest this slag fineness to be suitable for hydration studies. The crystalline forms of these synthetic slags were obtained by crystallizing (ageing) the granulated slags at 1100 °C for 2 hours in a Leco furnace heated by four Super Kanthal elements. Fig. 1 - View of furnace with top cover carrying the thermocouple Fig. 2 - Taking out top cover of furnace prior to crucible removal Fig. 3 - Removal of crucible at 1530 °C. 155 For microscopic examination, thin sections (typical procedure for mineralogical observations on opaque glassy type materials) were prepared. These specimens were then carbon coated and analyzed using S E M , E D X and Image Analysis to determine the microstructure and composition of phases. As for the industrial steel slag samples, they were previously oxidized at 900 °C in an attempt to form higher oxides of iron, then melted at 1530 °C and either water quenched or slow cooled in air to obtain glassy and crystalline forms. 156 APPENDIX II The Isothermal Conduction Calorimetry Method In general, the hydration of a pozzolan or cement is accompanied by a transformation from a higher to a lower energy state; excess energy is evolved as sensible heat, or enthalpy. Observations on the heat of hydration can reveal information on the nature and sequence of the chemical reactions which occur. For most cementing formulations, a complex series of reactions occurs after mixing the solid components with water. The interpretation of these reactions is not straightforward. High surface area solids such as cement clinker exhibit an exothermic surface reaction upon wetting, and the initial release of soluble ions causes structural changes to the liquid phase which are also reflected in the enthalpy change. Once cement gel develops, it adsorbs large amounts of water, so the total heat evolved includes not only the chemical heat of hydration but also the heat of adsorption of water which is incorporated into or on the hydrated material. Two methods have been widely used to determine enthalpy changes: heat of solution and dynamic conduction calorimetry [48]. Conduction calorimetry is more precise in determining the course of heat evolution in cementitious systems. It has the advantage of giving not only the total heat evolved, but also the instantaneous rate of heat evolution at any point in time. Thus, conduction calorimetry is a non-intrusive technique which allows the rate of heat evolution to be continuously monitored, giving valuable information on both the rate and the nature of the chemical reactions taking place. The value of this approach has been illustrated by widespread use of the Wexham calorimeter [10] in cement hydration studies. However, this type of calorimeter has some limitations when it comes to monitoring heat evolution at early stages, 157 because it is difficult to mix water and cement in-situ, hence water and cement are mixed outside the calorimeter with the result that the initial peak cannot be determined. Although not critical to the overall assessment of the reactivity of the cement, this peak provides valuable information on the reactions which are taking place. When using a Thermonetics SEC type, model C-12-45-2-E this peak can be easily detected, as water and cement come in contact inside the calorimeter and at the same temperature. Isothermal Conduction Calorimetry Set-up The set-up used in the present study consisted of a recirculating water bath with temperature control between -15-90 °C, a Thermonetics SEC isothermal conduction calorimeter, a LabNote type data acquisition system, and a computer. Figs. 1,2,3 and 4 show the actual set-up. Fig. 1 - The Isothermal Calorimeter Set-Up, including all components 158 - Temperature control via the recirculating water bath 159 Fig. 4 - Calorimeter, connection to the water bath and digital thermometer Fig. 5 - Inside the calorimeter, view of sample holder Fig. 5 shows the Thermonetics calorimeter. It is designed in such a way that all the heat generated within the calorimeter must pass through its walls. The walls consist of special thermoelectric transducers which relate thermal flux to a DC voltage output signal. The 160 transducers consist of thermopiles with uniquely spaced junctions. The "cold" junctions are on one side of the wall, and the "hot" junctions are on the other side [48]. There are 1440 junction sets (2880 individual junctions) uniformly spaced in the six sensor plates. The envelope is maintained at a constant temperature by passing water or air through the outer jacket of the calorimeter. As the heat flows through the wall or envelope, a small temperature difference is established. This temperature difference is directly proportional to the heat flow. Because of the presence of a large number of thermopiles, extreme sensitivity to minute heat flows can be obtained. Experimental Procedures and Data Analysis Samples of up to 2 g cement and/or slags were placed in the sample holder inside the calorimeter. The calorimeter was preheated to the desired starting temperature and maintained constant by means of the regulator attached to the recirculating water bath. Once temperature equilibrium was reached, distilled water, with temperature equal to the sample holder, was injected inside the sample holder, the ratio of water/cement being 0.5. Mixing was done inside the calorimeter when the signal was stable so that the entire enthalpy of mixing could be recorded. In addition to observing a stable signal on the monitor, equilibrium was checked by measuring the temperature inside the sample holder with a digital thermometer inserted through an opening provided in the equipment. Data acquisition was started at the moment of mixing and readings were taken every minute for up to 24 hours. The analysis of data requires both a blank measurement and a reference sample. For each temperature, a blank measurement was carried out. This blank data was later substracted from the experimental data to correct for the baseline shift arising from the difference in heat capacity between sample to be tested and reference sample. The blank data is very important, particularly 161 for temperature programmed experiments. It is desirable to match the thermal capacity of the reference and actual samples, and for the blank measurement to match the actual experiment as closely as possible. Since in these experiments the initial components are consumed to form new materials, the heat capacity will change with time. Only the extremes are acceptable as reference systems, i.e. dry cement and final set cement. In this case, dry cement was used as reference. Generally, a correction is only required for the first 2-3 hours of the experiment. During this time the cement will remain as slurry, chemically and physically intermediate between the two possible reference systems. A large difference in heat capacity between cement and the water/cement system can occur due to possible water evaporation at each temperature. In view of these, the blank correction for each temperature was established by running a test with the sample holder containing set cement and calculating the average mV over a 1 hour period, giving the system correction value. This approach has been used successfully in other research [48]. A baseline correction was calculated using the formula given in A S T M standard C 186-94 [112]. Corrected results were then compared to dry cement results, with good agreement. Also, a comparison was made with heat values found in the literature. It is roughly estimated that the calorimeter used in the present research can give up to a max 20-25% positive error when compared to results reported in the literature. However, this value is not very precise and may be lower, since heat values used for comparison refer to various hydration times, uncertain temperatures usually bordering room temperatures, probably distinct types of Portland cement and different calorimetric set-ups, consequently making the error assessment difficult. A short Fortran program was written to allow for calculating the area under the curve, representing the total heat release, as well as the heat at any given time and the time at which a maximum heat release takes place. This program was designed to allow for both blank and sample weight adjustments. 162 The following figure shows several synthetic slags samples removed from the calorimeter after 24 hr of hydration (Fig. 6). Samples hardened and were easily removed from the holder indicating that setting has occurred. Fig. 6 - Samples of cement (light gray), slags (dark gray) and sample holder Calibration of Calorimeter From the total experimental error, 8 to 9% corresponds to the equipment itself, as calculated in the initial calibration of the calorimeter by using a calibration heater of 48.1 Q and a calibration constant of 0.00906 cal/sec millivolt, both provided by the manufacturer. To calibrate the calorimeter the heater is inserted with its leads coming through the hypodermic needle hole in the calorimeter lid. The lid is then closed and a constant A C or DC voltage is applied to the heater, at a power of 0.5 to 1.0 watts. The voltage and current are monitored with laboratory volt and amper meters. At steady state, the calorimeter millivolt output is measured and recorded with a recorder, from which the calibration constant is established (ratio of heat added to millivolt output). The calibration results are presented in Table I. 163 Table I - Calibration of the calorimeter I P V Calculated Recorder Error (mA) (w) (mV) output (mV) output (mV) 0.1247 0.7506 5.99 19.73 18.00 8.77% • 0.1173 0.6621 5.64 17.46 16.10 7.82% 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items