Active Sulpho-belite cements. Hydration mechanisms and mechanical properties

Facultad de Ciencias Departamento de Química Inorgánica, Cristalografía y Mineralogía Active sulpho-belite cements. Hydration mechanisms and mechanical properties. Tesis Doctoral por Gema Álvarez Pinazo Málaga, 2015 AUTOR: Gema Álvarez Pinazo http://orcid.org/0000-0003-1396-2156 EDITA: Publicaciones y Divulgación Científica. Universidad de Málaga Esta obra está sujeta a una licencia Creative Commons: Reconocimiento - No comercial - SinObraDerivada (cc-by-nc-nd): Http://creativecommons.org/licences/by-nc-nd/3.0/es Cualquier parte de esta obra se puede reproducir sin autorización pero con el reconocimiento y atribución de los autores. No se puede hacer uso comercial de la obra y no se puede alterar, transformar o hacer obras derivadas. Esta Tesis Doctoral está depositada en el Repositorio Institucional de la Universidad de Málaga (RIUMA): riuma.uma.es Active sulpho-belite cements. Hydration mechanisms and mechanical properties MEMORIA presentada por la Ingeniera Química Dª Gema Álvarez Pinazo para optar al Grado de Doctora en Ciencias, Sección de Químicas, por la Universidad de Málaga. Fdo.: Gema Álvarez Pinazo Los Directores, Fdo.: Dr. Miguel Ángel García Aranda Catedrático de la Universidad Fdo.: Dra. Mª Ángeles Gómez de la Torre Profesora Titular de Universidad Fdo.: Dra. María Isabel Santacruz Cruz Investigadora Ramón y Cajal Dr. MIGUEL ÁNGEL GARCÍA ARANDA, Catedrático de la Universidad de Málaga, Dra. Mª ÁNGELES GÓMEZ DE LA TORRE, Profesora Titular de la Universidad de Málaga, y Dra. Mª ISABEL SANTACRUZ CRUZ, Investigadora Ramón y Cajal de la Universidad de Málaga, todos pertenecientes al Departamento de Química Inorgánica, Cristalografía y Mineralogía de la Facultad de Ciencias de la Universidad de Málaga, CERTIFICAN: Que la presente memoria realizada por Dª Gema Álvarez Pinazo, titulada: “ACTIVE SULPHO-BELITE CEMENTS. HYDRATION MECHANISMS AND MECHANICAL PROPERTIES”, ha sido realizada bajo nuestra dirección en el Departamento de Química Inorgánica, Cristalografía y Mineralogía de la Facultad de Ciencias de la Universidad de Málaga. Este trabajo reúne, a nuestro juicio, contenido científico suficiente y las condiciones necesarias para ser presentado y defendido ante el tribunal correspondiente para optar al Grado de Doctora. Málaga a 1 de Junio de 2015 Fdo.: Dr. Miguel Ángel García Aranda Catedrático de la Universidad Fdo.: Dra. Mª Ángeles Gómez de la Torre Profesora Titular de Universidad Fdo.: Dra. María Isabel Santacruz Cruz Investigadora Ramón y Cajal Dr. PEDRO JESÚS MAIRELES TORRES, Catedrático de la Universidad de Málaga y Director del Departamento de Química Inorgánica, Cristalografía y Mineralogía de la Facultad de Ciencias de la misma universidad. INFORMA: Que la presente memoria realizada por Dª Gema Álvarez Pinazo, titulada: “ACTIVE SULPHO-BELITE CEMENTS. HYDRATION MECHANISMS AND MECHANICAL PROPERTIES”, ha sido realizada bajo la dirección del Catedrático Miguel Ángel García Aranda, la Profesora Titular Dª Mª Ángeles Gómez De la Torre y la Investigadora Ramón y Cajal Dª. Mª Isabel Santacruz Cruz en el Departamento de Química Inorgánica, Cristalografía y Mineralogía de la Facultad de Ciencias de la Universidad de Málaga. Este trabajo constituye la Memoria de Tesis Doctoral de la interesada, cuya presentación autorizo en Málaga a 1 de Junio de 2015. Fdo.: Dr. Pedro Jesús Maireles Torres. i ACKNOWLEDGEMENTS My first thoughts are naturally given to the Department of Inorganic Chemistry, Crystallography and Mineralogy of University of Málaga to give me the opportunity to perform this Thesis, under the direction of Prof. Miguel Angel García Aranda, Dr. María de los Ángeles Gómez de la Torre, and Dr. María Isabel Santacruz Cruz. This memory is the result of efforts of many people who directly or indirectly have participated providing knowledge, giving their opinion, correcting and encouraging me, both in times of success and times of difficulty. Firstly, I would like to thank my PhD supervisors. I express my sincere gratitude for their assistance, suggestions, support, and friendly counselling in both research and life. Thanks for the time spent on my learning and transmit me their knowledge and values. Without all of that this memory would not have been possible. I also thank all the members of the Department of Inorganic Chemistry, Crystallography and Mineralogy. Countless memories accumulated during the last years are made by pieces of unforgettable experience with countless people; even every person that I crossed in my way has already earned his place in my esteem. Institute of Material Science of Madrid (ICMM-CSIC), especially Dr. Jesús Sanz Lázaro and Dr. Isabel Sobrados de la Plaza, and the two technicians of the NMR laboratory, Dr. Virginia Díez and Verónica López. Thanks for your hospitality and help during my stay. “Cementos Financiera y Minera (FYM)” cements factory, from Italcementi Group, in Málaga, is thanked. My special appreciation to my parents, my siblings and, in particular, to my husband for their love and support during my studies. Finally, I would like to acknowledge the financial contribution supported by Spanish MINECO through MAT2010-16213 research grant, which is cofounded by FEDER, and Junta de Andalucía through P11-FQM-07517. Index ix INDEX Acknowledgments i Agradecimientos iii Index ix Notations xiii Chemical Reactions xvii List of Tables xix List of Figures xxi ABSTRACT 3 RESUMEN 15 CHAPTER 1: INTRODUCTION 29 1.1. CLIMATE CHANGE. INDUSTRIAL CO2 EMISSIONS. 29 1.2. BELITE CEMENTS. 31 1.3. YE'ELIMITE -CONTAINING CEMENTS. 31 1.3.1. Classification. 32 1.3.2. Commercial ye'elimite -containing cements. 33 1.4. BELITE CALCIUM SULFOALUMINATE CEMENTS (BCSA). 35 1.4.1. Reduction of CO2 emissions. 36 1.4.2. Crystallochemistry of main anhydrous phases of BCSA cements. 38 1.4.3. Reactivity of principal constituents of BCSAF cements during hydration. 43 1.4.4. Crystallochemistry of main hydrated phases of BCSAF cements. 48 1.4.5. Properties of BCSA mortars and concretes. Durability. 52 1.4.5.1. Sulphate Resistance. 52 1.4.5.2. Corrosion Resistance. 52 1.4.5.3. Compressive Strength. 53 1.4.5.4. Dimensional Stability. 54 Index x 1.5. METHODOLOGY. 54 1.5.1. Cement (anhydrous and paste) characterisation. 54 1.5.1.1. LRPD and Rietveld method. 54 1.5.1.2. SXRPD (BL04 – MSPD, ALBA). 59 1.5.1.3. SEM. 59 1.5.1.4. Solid-state NMR spectroscopy. 60 1.5.1.5. Thermal measurements. 60 1.5.1.6. Isothermal conduction calorimetry. 61 1.5.1.7. MIP. 61 1.5.1.8. Rheological behaviour. 62 1.5.2. Mortars characterisation. 63 1.5.2.1. Compressive strength. 63 1.5.2.2. Shrinkage/Expansion properties (Length Changes). 64 1.5.2.3. Setting. 64 CHAPTER 2: OBJECTIVES 69 CHAPTER 3: ARTICLES SECTION 73 A#1. Rietveld quantitative phase analysis of Ye'elimite-containing cements. 75 A#2. In-situ early-age hydration study of sulfobelite cements by synchrotron powder diffraction. 89 A#3. Hydration reactions and mechanical strength developments of iron-rich sulfobelite eco-cements. 103 A#4. Rietveld quantitative phase analysis with molybdenum radiation. 125 A#5. Hydration of iron-rich BCSA cements with different calcium sulfate sources. 139 CHAPTER 4: RESULTS AND DISCUSSION 171 4.1. SYNTHESIS OF LABORATORY-PREPARED BCSAF CLINKERS: SCALE-UP. 173 4.1.1. Raw materials characterisation. 173 4.1.2. BCSAF clinkers preparation: optimisation of process. 174 Index xi 4.1.3. Characterisation of the scaled-up BCSAF clinkers. 177 4.1.3.1. RQPA (normalized to 100% of crystalline phases). 177 4.1.3.2. Selective dissolutions. 178 4.1.3.3. SEM-EDS characterisation. 181 4.2. PHASE ANALYSIS OF YE'ELIMITE-CONTAINING CLINKERS AND CEMENTS. 182 4.2.1. RQPA (normalized to 100% of crystalline phases). 183 4.2.2. ACn content determination. 185 4.3. HYDRATION STUDY OF BCSAF CEMENTS. 186 4.3.1. Hydration of BCSAF cements with gypsum as setting regulator. 187 4.3.1.1. In-situ early hydration behaviour (< 24 h). 188 4.3.1.2. Ex-situ hydration behaviour at late ages (> 24 h), with different amounts of gypsum. 190 4.3.1.3. Elemental composition of ACn. 194 4.3.1.4. A comparative study of MoKα1 and synchrotron radiations for selected samples. 201 4.3.2. Hydration of BCSAF cements with different calcium sulphate source. 204 4.3.2.1. Early hydration behaviour (< 24 h). 204 4.3.2.2. Hydration with different sulphate sources at later ages (> 24 h). 209 4.4. MECHANICAL PROPERTIES OF BCSAF MORTARS. 217 CHAPTER 5: CONCLUSIONS 227 CHAPTER 5: CONCLUSIONES (In Spanish) 233 CHAPTER 6: REFERENCES 239 ANNEX A: LICENSES AGREEMENTS 263 ANNEX B: OTHER PUBLICATIONS 269 xiii NOTATIONS A simplified notation is used when describing cement compounds or chemical formulations. The cement shorthand notation is found below: C = CaO A = Al2O3 S = SiO2 S = SO3 F = Fe2O3 B = B2O3 C = CO2 H = H2O K = K2O M = MgO T = TiO2 N = N2O P = P2O5 This leads to the following abbreviations for anhydrous and hydrates phases: Formula Oxides Cement nomenclature Name Ca 2 Mg(Si 2 O 7 ) 2CaO·MgO·2SiO 2 C 2 MS 2 Akermanite Ca 3 SiO 5 3CaO·SiO 2 C 3 S Alite 2Al(OH)3·nH2O Al2O3·(3+n)H2O AH3·nH Amorphous aluminium hidroxide CaSO 4 CaO·SO 3 C S Anhydrite CaSO 4 ·0.5H 2 O CaO·SO 3 ·0.5H 2 O C S H 0.5 Bassanite Ca 2 SiO 4 2CaO·SiO 2 C 2 S Belite CaCO3 CaO·CO2 CC Calcite or vaterite CaAl2O4 CaO·Al2O3 CA Calcium aluminate Ca2Al(OH)6[Al(OH)4·3H2O] 2CaO·Al2O3·8H2O C2AH8 Dicalcium aluminate hydrate CaMg(CO 3 ) 2 CaO·MgO·2CO 2 CM C 2 Dolomite Ca 6 Al 2 (OH) 12 (SO 4 ) 3 ·26H 2 O 6CaO·Al 2 O 3 ·3SO 3 ·32H 2 O C 6 A S 3 H 32 Ettringite (AFt) Ca 4 Al 2 Fe 2 O 10 4CaO·Al 2 O 3 ·Fe 2 O 3 C 4 AF F errite xiv Formula Oxides Cement nomenclature Name Ca 3 Al 2 (SiO 4 ) 3 3CaO·Al 2 O 3 ·3SiO 2 C 3 AS 3 Garnet Ca 2 Al 2 SiO 7 2CaO·Al 2 O 3 ·SiO 2 C 2 AS Gehlenite 2Al(OH) 3 Al 2 O 3 ·3H 2 O AH 3 Gibbsite CaSO 4 ·2H 2 O CaO·SO 3 ·2H 2 O C S H 2 Gypsum (Ca 4 Al 2 (OH) 12 )[OH(CO 3 ) 0. 5 (H2O)4] 4CaO·Al 2 O 3 ·10.5H 2 O· 0.5CO2 C4AH10.5C0.5 Hemicarbo - aluminate Ca3Al2(OH)12 3CaO·Al2O3·6H2O C3AH6 Hidrogarnet or katoite Ca 3 (Al 0.5 Fe 0.5 ) 2 (SiO 4 ) (OH)8 3CaO·0.5Al 2 O 3 ·0.5Fe 2 O 3 ·SiO2·4H2O C3A0.5F0.5SH4 Iron - hidrogarnet 2Fe(OH) 3 Fe 2 O 3 ·3H 2 O FH 3 Iron hydroxide Ca 12 Al 14 O 33 12CaO·7Al 2 O 3 C 12 A 7 Mayenite Ca 3 Mg (SiO 4 ) 2 3 CaO· MgO·2SiO 2 C 3 MS 2 M erwinite Ca4Al2(OH)12(SO4)·6H2O 4CaO·Al2O3·SO3·12H2O C4ASH12 Monosul ph ate (AFm) CaTiO 3 CaO·TiO 2 CT Perovskite Ca(OH)2 CaO·H2O CH Portlandite Ca 2 Al(OH) 6 [AlSiO 2 (OH) 4 · 3H2O] 2CaO·Al2O3·SiO2·8H2O C2ASH8 Stratlingite Ca 5 (SiO 4 ) 2 (SO 4 ) 5CaO ·2SiO 2 ·SO 3 C 5 S 2 S Ternesite Ca3Al2O6 3CaO·Al2O3 C3A Tricalcium aluminate Abbreviated names or initials have also been used to refer to some terms: • A#x: Article number x. • aBCSA: active Belite Calcium Sulpho-Aluminate. • ACn: Amorphous and Crystalline non-quantified. • ACSA: Alite Calcium Sulpho-Aluminate. • BACSA: Belite Alite Calcium Sulpho-Aluminate. • BCSA: Belite Calcium Sulpho-Aluminate. xv • BCSAA: Aluminum-rich Belite Calcium Sulpho-Aluminate. • BCSAF: Iron-rich Belite Calcium Sulpho-Aluminate. • CSA: Calcium Sulpho-Aluminate. • DIBt: Deutsches Institut für Bautechnik. • DTA: Differential Thermal Analysis. • EDS: Energy Dispersive X-ray Spectroscopy. • ETA: European Technical Approval. • FW: Free Water. • FWHM: Full-width at the half-maximun. • GHGs: Greenhouse gases. • ICSD: Inorganic Crystal Structure Database. • LoI: Loss of Ignition. • LXRPD: Laboratory X-Ray Powder Diffraction. • MIP: Mercury Intrusion Porosimetry. • MSPD: Materials Science and Powder Diffraction. • NMR: Nuclear Magnetic Resonance. • OPC: Ordinary Portland Cement. • PTFE: Polytetrafluoroethylene. • RMCO2: Raw material CO2. • RQPA: Rietveld Quantitative Phase Analysis. • SCMs: Supplementary Cementitious Materials. • SEM: Scanning Electron Microscopy. • SRM: Standard Reference Material. • SXRPD: Synchrotron X-ray Powder Diffraction. • TGA: Thermogravimetric Analysis. • w/c: water to cement ratio. • XRF: X-Ray Fluorescence. • XRPD: X-Ray Powder Diffraction. xvii CHEMICAL REACTIONS Ca4Al6O12(SO4) + 2CaSO4·xH2O + (38-2x) H2O  Ca6Al2(OH)12(SO4)3·26H2O + 4 Al(OH)3 [1.1] Ca4Al6O12(SO4) + (16+x) H2O  Ca4Al2(OH)12(SO4)·(H2O)(4+x) + 4 Al(OH)3 [1.2] Ca2SiO4 + (2+x-y) H2O  (CaO)y·SiO2·(H2O)x + (2-y) Ca(OH)2 [1.3] Ca2SiO4 + 2 Al(OH)3 + 5 H2O  Ca2Al(OH)6[AlSiO2(OH)4]·3H2O [1.4] Ca2SiO4 + x Ca4Al2Fe2O10 + 10x H2O  Ca3(Al1-xFex)2(SiO4)(OH)8 + (4x-1) Ca(OH)2 + (4x-2) Al(OH)3 [1.5] Ca2Al(OH)6[AlSiO2(OH)4·3H2O] + Ca(OH)2  Ca3Al2(SiO4)(OH)8 + 5 H2O [1.6] Ca4Al2Fe2O10 + 10 H2O  Ca3Al2(OH)12 + 2 Fe(OH)3 + Ca(OH)2 [1.7] Ca4Al2Fe2O10 + 10 H2O  4/3 Ca3(Al0.75Fe0.25)2(OH)12 + 4/3 Fe(OH)3 [1.8] Ca4Al2Fe2O10 + 3 CaSO4·2H2O + 30 H2O  Ca6Al2(OH)12(SO4)3·26H2O + 2 Fe(OH)3 + Ca(OH)2 [1.9] Ca6Al2(OH)12(SO4)3·26H2O  Ca4Al2(OH)12(SO4)·nH2O + 2[CaSO4·2H2O] + (22-n) H2O [1.10] Ca4Al2Fe2O10 + CaSO4·2H2O + (8+n) H2O  Ca4Al2(OH)12(SO4)·nH2O + 2 Fe(OH)3 + Ca(OH)2 [1.11] Ca2SiO4 + x Ca4Al2Fe2O10 + (1-2x)·2 Al(OH)3 + (10x+5) H2O  Ca2(Al1-xFex)(OH)6 [(Al1-xFex)SiO2(OH)4·3H2O] + 4x Ca(OH)2 [4.1] ABSTRACT Abstract 3 The threat of climate change is considered to be one of the major environmental challenges for our society, where carbon dioxide (CO2) is one of the major GHGs. Anthropogenic sources of CO2 are the combustion of fossil fuels, deforestation, unsustainable combustion of biomass, and decomposition of mineral sources such as limestone. Cement manufacturing is considered as one of the main sources of carbon dioxide emissions among industrial activities. Every ton of OPC produces about one ton of carbon dioxide. Consequently, OPC production accounts for 5-7% of the global CO2 emissions resulting from human activity, and for 4% of total global warming, making the cement industry an important sector for CO2-emission mitigation strategies. CO2 is emitted from the calcination process of limestone, from combustion of fuels in the kiln, as well as from power generation. Due to these environmental problems the industry of building materials is under increasing pressure to reduce the energy used in the production of Portland cement clinker and mainly the associated with GHGs emissions. Therefore, the design of new formulations of cements, such as belite or/and ye'elimite based cements, has many advantages and promises to be a viable solution. One of the most interesting alternatives is the BCSA cement. BCSA cement was first put into industrial production in China under the name “Third Cement Series” in the 1970s, and are considered environmentally friendly cements for several reasons, including the low amount of limestone required to achieve the desired composition. The substitution of alite (main phase in OPC) by belite reduces the limestone demand and temperature of cement manufacturing, thereby reducing CO2 emissions and energy consumption while maintaining satisfying long-term properties. However, that substitution compromises the early-age strength development because C2S reacts slower than C3S. This renders the high-C2S cement unusable in nearly any field related to structural applications. However, this is partially compensated by the presence of a fast reacting calcium sulphoaluminate phase (ye'elimite) in these BCSA cements. Thus, BCSA cements arise from clinkers containing belite (C2S) as main phase (40-50 wt%) and intermediate contents (20-30 wt%) of calcium sulphoaluminate, also called Klein’s salt, ye'elimite or tetracalcium trialuminate sulphate (C4A3S). These cements, also known as sulphobelite, are an emerging type of ye'elimite-containing cements due to their environmental benefits. The most common formulation of BCSA clinkers consists on β-C2S, orthorrombicC4A3S and C4AF. Due to the presence of the latter, these cements are usually called iron-rich BCSA (BCSAF). BCSAF cements might also contain minor phases such as calcium aluminates (C12A7 and CA), gehlenite (C2AS), and calcium sulphosilicate (C5S2S). The clinkering temperature is around 1250-1300°C, ∼200°C lower than that Abstract 4 for OPC production. The lower synthesis temperature not only reduces the energy consumption and CO2 emissions from cement manufacturing but also the resulting clinker is more friable (due to high porosity), which reduces the energy needed for grinding (indirect emissions). These cements also show rapid hardening, excellent durability, and, depending on the amount of gypsum added, self-stressing and volume stability. The properties and applications of this type of cements are strongly influenced by many factors: i) chemical and mineralogical composition of the clinker, ii) sulphate source (amount and type), iii) water to cement ratio (w/c); iv) the presence of other binders such as SCMs or even OPC. Carbon dioxide emissions in the cement industry can be classified in two main categories: those coming from raw materials and those from the operation processes. Raw materials and products obtained during the clinkering process are well known. On the one hand, the production of one ton of OPC clinker composed of 65 wt% of C3S, 15 wt% of C2S, 10 wt% of C3A and 10 wt% of C4AF released 0.54 tons of carbon dioxide. However one ton of BCSAF clinker composed by 50 wt% of C2S, 30 wt% of C4A3S, and 20 wt% of C4AF releases 0.39 tons of CO2. It implies a reduction of 0.15 tons of CO2 for BCSAF due to raw material decomposition. On the other hand, the reduction in CO2 emissions coming from operation processes are directly related to the type of processing equipment and the specific chosen fuel. Cement production is a high energy demanding process, so it is important to consider emissions from fuel consumption to achieve high clinkering temperatures in the kiln and thus quantify CO2 emissions. Several studies estimate an emission of ∼0.30 t of CO2 per ton of clinker produced assuming that good quality of bituminous coal is used and taking into account energy efficiency of modern kilns. The reduction in CO2 emissions coming from the burning of the fuel can be achieved by the reduction of clinkering temperature. This is the case of BCSAF clinker production where the operating temperature can be reduced in ∼200°C with a concomitant reduction of up to 0.04 t of CO2 per ton of BCSAF clinker produced. Moreover, emissions derived from electricity consumption are about 0.09 t of CO2 per ton of OPC. The lower firing temperatures needed for BCSAF clinkering made it easier to be ground. Therefore, reduction on the electricity consumption yields a depletion of up to 0.02 t of CO2 emissions. Considering all emissions together, the production of one ton of OPC clinker releases a maximum of 0.97 tons of CO2; thus, the production of one ton of BCSAF clinker leads to a reduction of ∼22% CO2 emissions compared to OPC. That reduction depends on the composition. In the last few years, BCSAF cements have emerged with the initial aim to substitute OPC. An industrial trial, ∼2500 tons of BCSAF, was carried out in 2011 by Lafarge under the AETHERTM project (http://www.aether-cement.eu/). Clinkering was performed in a Portland industrial kiln but using lower operating temperatures (1225-1300°C), reducing CO2 emissions by 25-30% in comparison with CEM (I) type Abstract 5 OPC cement. The temperature in the clinkering zone resulted to be a key parameter, as too high temperatures may give kiln blockage, loss of grindability and C4A3S decomposition with high SO2 emissions, and too low temperatures gave under burnt binder with high free lime and C12A7 contents. The optimum clinkering temperature resulted much lower in NOx emissions than those for OPC, due to the lower burning temperature, and SO2 emissions resulted at the same level than for OPC production. As mentioned before, the main technological disadvantage of these cements is related to the low mechanical strengths developed at very-early ages due to the slow hydration of belite. However, this problem may be overcome by the activation of belite and the presence of high amounts of ye'elimite. The production of active BCSAF cements involves the stabilisation of highly reactive C2S polymorphs, i.e. βmodified form and α-forms, as they react faster with water. Active-BCSAF cements have been patented by Lafarge and they contain Klein’s salt and α-forms of belite as main phases. The latter was stabilised due to addition of minor elements, such as B2O3 and Na2O. Active-BCSAF with 2.0 wt% of B2O3, added as borax, developed comparable compressive strengths to those of OPC. For these reasons, BCSAF cements are considered, nowadays, as one of the most promising alternatives to OPC. This is supported by both the environmental benefits (lower CO2 emissions) and industrial interest. However, before to be implanted in Europe, all the steps evolved in the process need to be under control, including clinkering (activation/composition and temperature), hydration (rheological behaviour and phase assemblage), and final performances (mechanical strength and dimensional stability). This PhD Thesis is focused on the study and optimisation of those parameters to improve the final performances of BCSAF mortars. One of the main objectives of this Thesis has been to perform the "medium scale" synthesis (∼2 kg) of two BCSAF clinkers in our laboratory with the expected phase composition of 50 wt% of C2S, 30 wt% of C4A3S and 20 wt% of C4AF. One of the clinkers was “activated” through the addition of borax, 2 wt% expressed as B2O3, to the raw material mixture. The aim of the activation has been obtaining clinkers with different belite (β-C2S or α'H-C2S) and ye'elimite (orthorhombic or pseudo-cubic) polymorphs to understand the effect of the polymorphism on the paste hydration mechanism and mechanical performances of the corresponding mortars. These clinkers were named as BCSAF_B0 (non-active) and BCSAF_B2 (active), for boron-free and boron-containing clinkers, respectively. The optimised two–step clinkering process consisted on: heating at 900°C during 30 min and further heating to 1350°C during 30 min. The clinker was rapidly quenched by forced air flow. The two obtained BCSAF clinkers were characterised through LXRPD, including the analysis of selective dissolutions, and SEM-EDS. Abstract 6 The quantitative phase analysis of BCSAF clinkers and cement pastes is an essential part of this Thesis needed to better understand their hydration behaviour and thus how their performances can be modified for any given application. X-ray diffraction coupled with Rietveld analysis is a suitable methodology to obtain quantitative phase analysis of these materials in the laboratory. The application of RQPA to characterise clinkers/cements/pastes is not an easy task due to the presence of appreciable amounts of amorphous/sub-cooled and/or non-crystalline phases. However, the quantification of the amorphous phases is a very important issue to understand hydration mechanisms. The final phase assemblage obtained by LXRPD and Rietveld method for the two laboratory-scaled-up prepared clinkers confirmed that BCSAF_B0 clinker contains β-C2S and both orthorhombic and pseudocubic ye'elimite as main phases, meanwhile α'H-C2S and pseudo-cubic ye'elimite are stabilised in BCSAF_B2. The quantification of the ACn content was also performed. For that, the following two approaches have been used: i) external standard procedure (G-factor method) with reflection geometry; ii) internal standard procedure (spiking method with ZnO) with transmission geometry, using CuKα1,2 radiations. Several conclusions drawn from a comparative study using the two approaches were: on the one hand, the G-factor method allows the quantifications of both crystalline phases and ACn contents in these materials, where the latter reaches ∼25 wt% in BCSAF clinkers. On the other hand, the amount of ACn calculated showed the same trend independently on the used methodology. Moreover, the microstructural characterisation of both clinkers revealed that the average particle size of ye'elimite (with angular shaped particles) and belite (with typical rounded shape) particles present in BCSAF_B0 were smaller than those found in BCSAF_B2. In addition, ye'elimite particles in the active clinker contain small amounts of Si, Fe and Na. Other important objective of this Thesis has been to understand the influence of calcium sulphate (type and amount) on the hydration of active and nonactive BCSAF cements. BCSAF clinkers were mixed with different types and amounts of calcium sulphate sources (gypsum (G), anhydrite (A) and bassanite (B)). Cements prepared with gypsum are labelled hereafter as GgBx, where g stands for 5, 10 or 15 wt% of gypsum, and x for 0 or 2 (for non-active or active clinkers, respectively). Cements prepared with different sulphate sources are labelled hereafter as G10Bx, A10Bx and B10Bx, for 10 wt% of gypsum, anhydrite or bassanite, respectively. Bassanite, CSH0.5, and anhydrite, CS, were previously synthesized by heating commercial gypsum. All the cement pastes were prepared, at a w/c = 0.55, by mixing both BCSAF clinkers (BCSAF_B0 and BCSAF_B2) with the corresponding sulphate source. Two studies were carried out to better understand the hydration behaviour. Abstract 7 On the one hand, an in-situ SXRPD study for the first hours of hydration was carried out in the SXRPD station of MSPD beamline at ALBA synchrotron (Barcelona, Spain). On the other hand, ex-situ studies at later ages of hydration were performed to determine the influence of the amount and type of calcium sulphate added to BCSAF cements on hydration behaviour through LXRPD (CuKα1 radiation). In addition, a comparative study of MoKα1 and synchrotron radiations for selected hydrated samples was also performed. The in-situ study, in GgBx cements, showed important differences in the hydration process, such as different dissolution kinetic of gypsum and ye'elimite. For instance in G10B0, gypsum is completely dissolved after 5 h of hydration and ye'elimite dissolves at a higher pace than in active BCSAF. However, in G10B2, gypsum is dissolved after 11 h and ye'elimite is still present after 51 h of hydration. In addition, the crystallisation rate of AFt is also different in both cements. At 1 h of hydration G10B0 contains 14.2(2) wt% of AFt while at the same hydration time, only 1.9(1) wt% was quantified for G10B2. The fast sulphate consumption by crystalline ettringite precipitation in G10B0 paste is responsible of its higher pH value, 12.4 + 0.1, when compared to 10.3 + 0.1 for G10B2. The second most important difference between both hydration behaviours takes place after 1 day of hydration. On the one hand, in G10B0, the dissolution of β-C2S and C4AF starts after 24 h of hydration, with the consequent crystallisation of layered AFm type phases, such as stratlingite. On the other hand, for G10B2, α'H-C2S percentage remains constant up to 51 h of hydration and C4AF dissolves very slowly after 14 h. The difference in reactivity of both belite polymorphs is in disagreement with the general accepted idea in the cement field: α-forms of belite present faster hydration kinetics than β-forms. But under our studied experimental conditions, β-C2S reacts faster than α'H-C2S to yield stratlingite, and this behaviour may well be justified with the formation of high amounts of ettringite at early hours which implies a concomitant large quantity of amorphous aluminium hydroxide. The availability of amorphous AH3 promotes the precipitation of stratlingite, C2ASH8, from belite reaction. Then, the hydration behaviour of C2S is more dependent on the chemical environment than on its polymorphism. The influence of the amount of added gypsum in BCSAF cements at late ages of hydration (> 24 h) was also tested in GgBx cements. One of the main conclusions obtained in this part of the study was the astonishing behaviour of β-belite in nonactive clinkers since it reacts at a higher pace than α′H-belite in BCSAF_B2, as mentioned before. Moreover, ye'elimite reacts at a different pace for BCSAF_B0 and BCSAF_B2. Orthorhombic ye'elimite, in non-active cements, is completely hydrated after 3 days of hydration in GgB0, while pseudo-cubic ye'elimite in GgB2 reacts at slightly slower pace for the same gypsum content and age of hydration. This effect was also previously observed in the hydration study at early hours. Abstract 8 Other conclusion has been that ye'elimite reaction kinetics showed a small dependence on the amount of added gypsum, as there was a slight increase in hydration rate by increasing the gypsum content. This behaviour seems to be slightly more marked for the pseudo-cubic ye'elimite. Moreover, the final reaction degree of both polymorphs of dicalcium silicate is affected by the increment of gypsum. On the one hand, β-C2S reactivity (given by the degree of reaction, α) was enhanced by increasing the gypsum content (α rises from 65% to 75% by the addition from 5 to 15 wt% of gypsum). On the other hand, α′H-C2S reaction degree decreased from 62% to 42% for addition of 5 to 15 wt% of gypsum, respectively. Finally, the hydration of the ferrite phase was strongly retarded by increasing the gypsum content in both GgB0 and GgB2 cements. In all cases the main crystalline hydrated compounds were ettringite, stratlingite and katoite. The amount of crystallised ettringite in GgB2 cements resulted higher than that in GgB0 cements, irrespective of gypsum content. Moreover, the crystallisation process of stratlingite is strongly affected by the amount of added gypsum; in fact, the amount of stratlingite decreases by increasing the gypsum content. It is not only important to quantify the amorphous content, but also to try to characterise and estimate its elemental composition and correlate it with the cement hydration behaviour and mechanical properties. It is not possible to determine the chemical composition of ACn directly by LXRPD, but an attempt to find it out (sulphate, silicate, aluminate and iron-bearing groups) has been performed thought RQPA and Rietveld methodology combined with G-factor. For this purpose, the evolution of different ions-containing groups was studied with time. Since the amount of sulphates crystallised in GxB2 was higher than that in GxB0 pastes, it means that a bigger amount of sulphate groups were dissolved, but not crystallised in the last one; then, they were mainly incorporated into ACn phase(s) and/or in pore solution of the GxB0 cement pastes. The amount of crystallised silicate is higher in GxB0 than in GxB2, where higher amounts of stratlingite were found. Thus, a higher amount of hydrated silicate remains in the amorphous phase(s) for GxB2 cement paste. The amount of crystallised aluminium-bearing phases was higher in GgB0 than in GgB2 cement pastes, matching in some cases the maximum, whereas more than 30 wt% of the aluminate content remained in the ACn phase(s) for GgB2 pastes, and/or to a minor extent in pore solution. A hydration study of BCSAF cements with different calcium sulphate source [gypsum (G), bassanite (B) and anhydrite (A)] was also performed. First, an in-situ SXRPD study was carried out to determine the role of the type of calcium sulphate source (with 10 wt%) in the first hours of hydration. The reaction degree of both ye'elimite and ettringite in G10B2, A10B2 and B10B2 at early ages was determined. Abstract 9 The direct RQPA results were normalised taking into account the theoretical data of the sample at 0.0 h of hydration. Comparing the theoretical results with the direct RQPA data, we can conclude that, for G10B2 sample, the AFt crystallisation process is parallel to ye'elimite dissolution, and the gypsum dissolution is very fast. In addition, G10B2 presented an induction period close to 6 h, when the dissolution and crystallisation of phases become significant. For A10B2, the RQPA results showed that the dissolution kinetic of anhydrite is much slower than that for gypsum or bassanite, as expected. For A10B2 paste, the precipitation of ettringite is limited by CS dissolution, which starts to be significant up to 7 h. Using these results we can also state that the predicted reactivity of ye'elimite with water to form AFm as main hydrated phase has not taken place within the first 6 h of hydration. Our results state that ye'elimite dissolution yields ettringite in spite of the fact that anhydrite is not dissolved until 6 h. B10B2 paste was also analysed. Due to experimental requirements (sample loading in the capillaries, alignment and so on), it was not possible to measure the first ∼40 min of hydration. Since the dissolution of bassanite and precipitation of gypsum are very fast processes, bassanite was almost absent and gypsum had crystallised just after 1 h of hydration. Subsequent hydration reactions are similar to those already described for the gypsum-regulated cement, G10B2. The very fast dissolution of bassanite with precipitation of gypsum, as well as the low dissolution rate of anhydrite was quantified, confirming the accurateness of the methodology used. Since bassanite in contact with water suffers from a fast grain decay (intergranular attack) which produces an increasing of the surface area of the sulphate carrier, and as a consequence, a high water demand (and high viscosity). In addition, a primary gypsum precipitation occurs, which will also affect the rheological behaviour of the paste. Thus, both parameters the high water demand and the gypsum precipitation increase the viscosity of bassanite-pastes. Since our objective is to study the effect of the calcium sulphate source (including compressive strengths of the corresponding mortars) similar rheological behaviour, and in particular, similar viscosity values at very early hydration ages are desired. In this case a small amount of a commercial polycarboxylate-based superplasticizer (SP) (0.05 wt% of active matter referred to total solids content), was added to water to prepare bassanitecontaining pastes. It exhibited a considerable diminishing in viscosity and a similar rheological behaviour to those prepared with gypsum or anhydrite. The main conclusions obtained of this hydration study at late ages are below. In all cases, the sulphate source was consumed before 3 days of hydration to form ettringite as the main crystalline hydrated phase. AFm and stratlingite were also found in all the studied pastes but in variable amounts. Independently of the sulphate source, ettringite seems to be more stable in active cements (X10B2), which contained α'H-C2S and pseudo-cubic ye'elimite, as it is almost constant with time of Resumen 16 cementos que contienen ye'elimita, con beneficios ambientales. La formulación más común de un clínker BCSA consiste en β-C2S, C4A3S ortorrómbica y C4AF. Debido a la presencia de esta última fase, estos cementos se llaman generalmente cementos BCSA ricos en hierro (BCSAF). Los cementos BCSAF también pueden contener fases minoritarias tales como aluminatos de calcio (C12A7 y CA), gehlenita (C2AS) y sulfosilicato de calcio (C5S2S). La temperatura de clínkerización de estos clínkeres está en torno a 1250-1300°C, ∼200°C más baja que la de producción del OPC. Esta temperatura de síntesis inferior no sólo reduce el consumo de energía y las emisiones de CO2 procedentes de la fabricación de cemento, sino que también da lugar a un clínker más friable (debido a la alta porosidad), lo que reduce la energía necesaria para la molienda (emisiones indirectas). Estos cementos también muestran un endurecimiento rápido, una excelente durabilidad, y, dependiendo de la cantidad de yeso añadida, una buena estabilidad de volumen. Las propiedades y aplicaciones de este tipo de cementos están fuertemente influenciadas por muchos factores: i) la composición química y mineralógica del clínker, ii) la fuente de sulfato (cantidad y tipo), iii) la relación agua/cemento; iv) la presencia de otros aglutinantes tales como materiales suplementarios (SCMs, del inglés Supplementary Cementitious Materials) o incluso OPC. Las emisiones de dióxido de carbono en la industria del cemento se pueden clasificar en dos categorías principales: las que vienen de las materias primas y las de los procesos de operación. Las materias primas y los productos obtenidos durante el proceso de clínkerización se conocen bien. Por un lado, la producción de una tonelada de clínker OPC formado por un 65% en peso de C3S, 15% en peso de C2S, 10% en peso de C3A y 10% en peso de C4AF, libera 0.54 toneladas de dióxido de carbono. Sin embargo una tonelada de clínker BCSAF compuesta por 50% en peso de C2S, 30% en peso de C4A3S, y 20% en peso de C4AF libera 0.39 toneladas de CO2. Esto implica una reducción de 0.15 toneladas de CO2 para el BCSAF debido a la descomposición de materia prima. Por otro lado, la reducción en las emisiones de CO2 procedentes de los procesos de operación está directamente relacionada con el tipo de equipo de procesamiento y el combustible específico elegido. La producción de cemento es un proceso de alta demanda energética, por lo que es importante tener en cuenta las emisiones procedentes del consumo de combustible necesario para alcanzar las altas temperaturas de clínkerización en el horno, y así cuantificar las emisiones de CO2. Diversos estudios estiman una emisión de ∼0.30 t de CO2 por tonelada de clínker producido, asumiendo una buena calidad del carbón bituminoso usado y teniendo en cuenta la eficiencia energética de los hornos modernos. La reducción en las emisiones de CO2 procedentes de la quema del combustible se puede lograr mediante la reducción de la temperatura de clínkerización. Este es el caso de la producción de clínker BCSAF donde la temperatura de operación se puede reducir en ∼200°C con una reducción asociada de hasta 0.04 t de CO2 por tonelada de clínker BCSAF producido. Por otra parte, las emisiones derivadas del consumo de Resumen 17 electricidad son aproximadamente 0.09 toneladas de CO2 por tonelada de OPC. Las temperaturas de cocción más bajas necesarias para la clínkerización del BCSAF hacen que sea más fácil a moler. Por lo tanto, la reducción en el consumo de electricidad produce un descenso de las emisiones de CO2 de hasta 0.02 toneladas. Teniendo en cuenta todas las emisiones en conjunto, la producción de una tonelada de clínker OPC libera un máximo de 0.97 toneladas de CO2; sin embargo, la producción de una tonelada de clínker BCSAF conduce a una reducción de ∼22%, dependiendo de la composición. En los últimos años, los cementos BCSAF han surgido con el objetivo inicial de sustituir al OPC. Una prueba industrial, de ∼2500 toneladas de BCSAF, se llevó a cabo en 2011 por Lafarge en el marco del proyecto AETHERTM (http://www.aethercement.eu/). La clínkerización se llevó a cabo en un horno de Portland industrial pero utilizando temperaturas menores (1225-1300°C), reduciendo las emisiones de CO2 en un 25-30% en comparación con el cemento OPC tipo CEM (I). La temperatura en la zona de clínkerización resultó ser un parámetro clave, ya que temperaturas demasiado altas podían provocar el bloqueo del horno, la pérdida de la capacidad de molienda y la descomposición del C4A3S con altas emisiones de SO2, y temperaturas demasiado bajas daba un ligante poco quemado con altos contenidos de cal libre y C12A7. La temperatura óptima de clínkerización resultó dar menores emisiones de NOx que los OPC, debido a la temperatura de combustión inferior, y emisiones de SO2 del mismo nivel que para la producción de OPC. Como se mencionó anteriormente, la principal desventaja tecnológica de estos cementos está relacionada con sus bajas resistencias mecánicas a edades tempranas, debido a la lenta hidratación de la belita. Sin embargo, este problema puede superarse mediante la activación de la belita y la presencia de altas cantidades de ye'elimita. La producción de cementos BCSAF activados implica la estabilización de los polimorfos de C2S altamente reactivos, es decir, la forma β-modificada y las formas α, ya que estas reaccionan más rápidamente con el agua. Los cementos BCSAF activados han sido patentados por Lafarge y contienen sal de Klein y formas α de la belita como fases principales. Esta última fue estabilizada debido a la adición de elementos minoritarios, como B2O3 y Na2O. Cementos BCSAF activados con 2% en peso de B2O3, añadido como bórax, han desarrollado resistencias a la compresión comparables a los OPC. Por estas razones, los cementos BCSAF se consideran, hoy en día, como una de las alternativas más prometedoras a los OPC. Esto es apoyado por los beneficios ambientales (emisiones de CO2) y el interés industrial. Sin embargo, antes de ser implantados en Europa, deben de controlarse todos los pasos envueltos en el proceso, incluyendo la clínkerización (activación/composición y temperatura), la hidratación (comportamiento reológico y ensamblaje de fase), y las propiedades Resumen 18 finales (resistencias mecánicas y estabilidad dimensional). Esta tesis doctoral se centra en el estudio y optimización de estos parámetros para mejorar las propiedades finales de los morteros BCSAF. Uno de los principales objetivos de esta tesis ha sido llevar a cabo la síntesis a "mediana escala" (para obtener ∼2 kg) de dos clínkeres BCSAF en nuestro laboratorio con una composición de fases esperada de 50% en peso de C2S, 30% en peso de C4A3S y 20% en peso de C4AF. Uno de los clínkeres fue "activado" mediante la adición de bórax, 2% en peso expresado como B2O3, a la mezcla de materia prima. El objetivo de la activación fue obtener clínkeres con diferentes polimorfos de la belita (β-C2S o α'H-C2S) y de la ye'elimita (ortorrómbica o pseudo-cúbica), para comprender el efecto del polimorfismo en el mecanismo de hidratación de las pastas y en las propiedades mecánicas de los morteros correspondientes. Estos clínkeres se llamaron como BCSAF_B0 (no activo) y BCSAF_B2 (activo), para el libre de boro y el que contiene boro, respectivamente. El proceso de óptimo de clínkerización fue el siguiente: calentamiento hasta 900°C manteniendo 30 min a esa temperatura, y calentamiento posterior hasta 1350°C manteniendo durante 30 min. El clínker resultante fue enfriado rápidamente mediante un flujo de aire forzado. Los dos clínkeres BCSAF obtenidos se caracterizaron a través LXRPD, incluyendo el análisis de disoluciones selectivas y microscopía electrónica de barrido (SEM-EDS). El análisis cuantitativo de fases de los clínkeres BCSAF es una parte esencial de esta tesis necesaria para comprender mejor su comportamiento de hidratación y por tanto cómo sus propiedades mecánicas pueden modificarse para cualquier aplicación dada. La difracción de rayos X junto con el análisis de Rietveld es una metodología adecuada para obtener el análisis de fase cuantitativo de estos materiales en el laboratorio. La aplicación de RQPA para caracterizar clínkeres/cementos/pastas no es una tarea fácil debido a la presencia de fases amorfas y/o no cristalinas en las muestras. Por eso, la cuantificación de las fases amorfas es un tema muy importante para entender los mecanismos de hidratación. El ensamblaje final de fases obtenido por LXRPD y el método Rietveld para los dos clínkeres preparados en el laboratorio confirma que el clínker BCSAF_B0 contiene βC2S y ye'elimita (ortorrómbica y pseudo-cúbica) como fases principales, mientras que la α'H-C2S y la ye'elimita pseudo-cúbica se estabilizaron en el clínker BCSAF_B2. La cuantificación del contenido amorfo (ACn, del inglés Amorphous and Crystalline non-quantified) también se llevo a cabo. Para ello, dos métodos se han utilizado en esta Tesis doctoral: i) el procedimiento del estándar externo (Factor G) con la geometría de reflexión; ii) el procedimiento del estándar interno (“spiking” método con ZnO) con la geometría de transmisión, utilizando radiaciones CuKα1,2. Varias conclusiones se han extraído del estudio comparativo usando los dos métodos: por un lado, el método del factor G permite la cuantificación tanto de fases Resumen 19 cristalinas como del ACn en estos materiales, siendo este último de ∼25% en peso en los clínkeres BCSAF. Por otro lado, la cantidad de ACn calculado mostró la misma tendencia independientemente de la metodología utilizada. Además, una caracterización micro estructural de ambos clínkeres reveló que los tamaños medios de partícula para la ye'elimita (partículas con forma angular) y la belita (con típica forma redondeada) presentes en BCSAF_B0 eran ligeramente menores que en BCSAF_B2. Por otra parte, las partículas de ye'elimita en el clínker activado contienen pequeñas cantidades de Si, Fe y Na. Otro objetivo importante de esta Tesis ha sido comprender la influencia del sulfato de calcio (tipo y cantidad) en la hidratación de los cementos BCSAF (activos y no activos). Para ello, ambos clínkeres se mezclaron con diferentes tipos y cantidades de fuentes de sulfato de calcio [yeso (G), anhidrita (A) y basanita (B)]. Los cementos preparados con yeso se van a nombrar a partir de ahora como GgBx, donde g representa 5, 10 o 15% en peso de yeso, y x será 0 o 2 (para el clínker no activo o activo, respectivamente). Los cementos preparados con diferentes fuentes de sulfato se nombrarán en lo sucesivo como G10Bx, A10Bx y B10Bx, para el 10% en peso de yeso, anhidrita o basanita, respectivamente. La basanita, CSH0.5, y anhidrita, CS, fueron sintetizadas previamente por calentamiento del yeso comercial. Se prepararon todas las pastas de cemento, con una relación agua/cemento = 0.55, mezclando ambos clínkeres (BCSAF_B0 y BCSAF_B2) con la fuente de sulfato correspondiente. Se llevaron a cabo dos estudios para comprender mejor el comportamiento de hidratación. Por un lado, un estudio in-situ para las primeras horas de hidratación, se llevó a cabo mediante SXRPD en la línea MSPD del sincrotrón ALBA (Barcelona, España). Por otra parte, se realizaron estudios ex-situ a edades posteriores de hidratación, para determinar la influencia de la cantidad y el tipo de sulfato de calcio añadido al cemento BCSAF a través de LXRPD (radiación CuKα1). Además, también se ha llevado a cabo un estudio comparativo radiaciones de MoKα1 y sincrotrón para muestras hidratadas seleccionados. El estudio in-situ, de los cementos GgBx, mostró importantes diferencias durante la hidratación, como la diferente cinética de disolución del yeso y la ye'elimita. Por ejemplo, en G10B0, el yeso se disolvió completamente después de 5 h de hidratación y la ye'elimita se disolvió a un ritmo mayor que en el cemento BCSAF activo. Sin embargo, en G10B2, el yeso se disolvió después de 11 h y la ye'elimita todavía estaba presente después de 51h de hidratación. Además, la velocidad de cristalización del AFt también es diferente en ambos cementos. A 1 h de hidratación, G10B0 contiene 14.2(2)% en peso de AFt mientras que para el G10B2 solamente se cuantificó 1.9(1)% en peso de AFt. El rápido consumo del sulfato debido a la precipitación de etringita cristalina en la pasta de G10B0 es responsable del mayor valor de pH, 12.4 + 0.1, en comparación con 10.3 + 0.1 para la pasta G10B2. La Resumen 20 segunda diferencia más importante entre ambos comportamientos de hidratación tiene lugar después de 1 día. Por un lado, en G10B0, la disolución de β-C2S y C4AF comienza después de 24 h, con la consiguiente cristalización de fases tipo AFm, tales como la stratlingita. Mientras que para G10B2, el porcentaje de α'H-C2S permanece constante hasta las 51 h de hidratación y el C4AF se disuelve muy lentamente a partir de las 14 h. La diferencia en la reactividad de ambos polimorfos de la belita está en desacuerdo con la idea general aceptada en el campo del cemento: las formas α de la belita presentan cinéticas de hidratación más rápidas que las formas β. Aún así, bajo nuestras condiciones experimentales estudiadas, β-C2S reacciona más rápidamente que α'H-C2S, para producir stratlingita, y este comportamiento puede estar bien justificado con la formación de altas cantidades de etringita en las primeras horas lo que implica una gran cantidad de hidróxido de aluminio amorfo. La disponibilidad de AH3 amorfo promueve la precipitación de la stratlingita, C2ASH8, a partir de la reacción de la belita. En conclusión, el comportamiento de hidratación del C2S es más dependiente del ambiente químico que de su polimorfismo. La influencia de la cantidad de yeso añadida en los cementos BCSAF a edades tardías de hidratación (>24h) también fue estudiado. Una de las principales conclusiones obtenidas en esta parte del estudio fue de nuevo el sorprendente comportamiento de la β-belita en el clínker no activo, ya que reaccionó a un ritmo más alto que la α'H-belita en BCSAF_B2, como se mencionó antes. Además, la ye'elimita reaccionó a un ritmo diferente para BCSAF_B0 y para BCSAF_B2. La forma ortorrómbica de la ye'elimita, en los cementos no activos, se disuelve por completo después de 3 días de hidratación, mientras que la ye'elimita pseudo-cúbica presente en GgB2 se disuelve ligeramente más lenta para el mismo contenido de yeso y la misma edad de hidratación. Este efecto también se observó anteriormente en el estudio de hidratación en las primeras horas. Otra conclusión de este estudio fue que la cinética de reacción de la ye'elimita mostró una pequeña dependencia con la cantidad de yeso añadido, ya que la velocidad de hidratación aumentó ligeramente al aumentar el contenido de yeso. Este comportamiento parece un poco más marcado para la ye'elimita pseudo-cúbica. Por otro lado, el grado final de reacción de ambos polimorfos del silicato dicálcico se ve afectada por el incremento de yeso. Por un lado, la reactividad de β-C2S (dada por el grado de reacción, α) se mejoró al aumentar el contenido de yeso (α aumenta del 65% al 75% mediante la adición de 5 a 15% en peso de yeso). Por otro lado, el grado de reacción de α'H-C2S se redujo de 62% a 42% para la adición de 5 a 15% en peso de yeso, respectivamente. Por último, la hidratación de la fase de ferrita está fuertemente retardada por el aumento del contenido de yeso en ambos cementos (GgB0 y GgB2). Resumen 21 En todos los casos los principales compuestos cristalinos hidratados obtenidos fueron etringita, stratlingita y katoita. La cantidad de etringita cristalizada en los cementos GgB2 resultó más elevada que en los cementos GgB0, independientemente del contenido de yeso. Por otra parte, la cristalización de la stratlingita está fuertemente afectada por la cantidad de yeso añadido; de hecho, la cantidad de stratlingita disminuye al aumentar el contenido de yeso. No sólo es importante poder cuantificar el contenido amorfo de estos materiales, sino también tratar de caracterizar y estimar su composición elemental y correlacionarla con el comportamiento de hidratación del cemento y sus propiedades mecánicas. No es posible determinar la composición química de ACn directamente por LXRPD, pero se ha llevado a cabo un intento mediante RQPA y la metodología de Rietveld combinado con el factor G para descubrir la posible composición de la ACn (grupos sulfatos, silicatos, aluminatos y grupos que contienen hierro). Para este fin, se ha estudiado la evolución de los diferentes grupos de iones con el tiempo. Puesto que la cantidad de sulfatos cristalizadas en GxB2 fue mayor que en las pastas de GxB0, esto significaba que una mayor cantidad de grupos sulfatos se disolvieron, pero no cristalizaron en la última; entonces, se incorporaron principalmente en la fase (s) ACn y/o en los poros de las pastas de cemento GxB0. La cantidad de silicatos cristalizados fue mayor en GxB0 que en GxB2, donde se encontraron mayores cantidades de stratlingita. Por lo tanto, una mayor cantidad de silicatos hidratado permanecen en la fase amorfa (s) para las pastas de cementos GxB2. La cantidad de fases que contienen aluminio cristalizadas en GgB0 fue mayor que en las pastas de cemento GgB2, igualando en algunos casos el máximo, mientras que más del 30% en peso del contenido de aluminato permaneció en la fase de ACn (s) para pastas GgB2, y/o en menor medida en los poros. Un estudio de hidratación de cementos BCSAF con diferentes fuentes de sulfato de calcio [yeso (G), basanita (B) y anhidrita (A)] también se ha llevado a cabo. En primer lugar, un estudio in-situ con SXRPD, se realizó para determinar el papel del tipo de sulfato de calcio (con 10% en peso) en las primeras horas de hidratación. Se determinó el grado de reacción de la ye'elimita y la etringita en G10B2, A10B2 y B10B2 a edades tempranas. Los resultados directos de RQPA se normalizaron teniendo en cuenta los datos teóricos de la muestra a 0.0 h de hidratación. Comparando los resultados teóricos con los datos RQPA directos, se puede concluir que, para la muestra G10B2, el proceso de cristalización del AFt es paralelo a la disolución de la ye'elimita, y la disolución de yeso es muy rápida. Además, G10B2 presentó un período de inducción cercano a 6 h, a partir del cual la disolución y cristalización de las fases se vuelven significativas. Para A10B2, los resultados RQPA mostraron que la cinética de disolución de la anhidrita es mucho más lenta que la del yeso o la basanita, como se esperaba. Para A10B2, la precipitación de la etringita está limitada por la disolución de la CS, que empieza a ser significativo a partir de 7 h. Resumen 22 Usando estos resultados podemos afirmar que la reactividad prevista de la ye'elimita con agua para formar AFm como principal fase hidratada no ha tenido lugar dentro de las primeras 6 h de hidratación. Nuestros resultados muestran que la disolución de la ye'elimita da etringita a pesar del hecho de que la anhidrita no se disuelve hasta 6 h. También se analizó la pasta B10B2. Debido a los requisitos experimentales (relleno de la muestra en los capilares, alineación,…), no fue posible medir los primeros ∼40 min de hidratación. Debido a que la disolución de la basanita y la precipitación de yeso son procesos muy rápidos, la basanita está casi ausente y el yeso se ha cristalizado justo después de 1 h de hidratación. Las reacciones de hidratación posteriores son similares a las ya descritos para los cementos con yeso. La rápida disolución de la basanita con la precipitación de yeso, así como la lenta velocidad de disolución de la anhidrita fue cuantificada, confirmando la exactitud de la metodología utilizada. Debido a que la basanita en contacto con agua sufre un rápido deterioro de grano (ataque intergranular) se produce un aumento del área superficial de la fuente de sulfato, y como consecuencia, una alta demanda de agua (y una alta viscosidad). Además, la precipitación de yeso primario también afectará al comportamiento reológico de la pasta. Por lo tanto, ambos parámetros, la alta demanda de agua y la precipitación de yeso aumentan la viscosidad de las pastas con basanita. Dado que nuestro objetivo es estudiar el efecto de la fuente de sulfato de calcio (incluyendo resistencias a compresión de los correspondientes morteros) es necesario tener similares comportamientos reológicos, y, en particular, similares valores de viscosidad a edades muy tempranas de hidratación. En este caso una pequeña cantidad de un superplastificante comercial (SP), basado en policarboxilato, (0,05% en peso de materia activa, referida al total de contenido de sólidos), se añadió al agua para preparar pastas que contienen basanita. Se observó una disminución considerable de la viscosidad y un comportamiento reológico similar a los preparados con yeso o anhidrita. Las principales conclusiones obtenidas de este estudio de hidratación a edades más tardías se encuentran a continuación. En todos los casos, la fuente de sulfato se consumió antes de los 3 días de hidratación para formar etringita como principal fase cristalina hidratada. AFm y stratlingita también se encontraron en todas las pastas estudiadas pero en cantidades variables. Independientemente de la fuente de sulfato, la ettringita parece ser más estable en los cementos activos (X10B2), los cuales contienen α'H-C2S y ye'elimita pseudo-cúbica, y además es casi se mantiene casi constante con el tiempo de hidratación. Por otro lado, el contenido de AFt disminuye con el tiempo en los cementos X10B0 para dar AFm. En consecuencia, a altas edades la cantidad de AFt es mucho mayor en X10B2, lo que probablemente sea responsable de las mejores propiedades mecánicas encontradas en esta familia, en comparación con los datos obtenidos para X10B0. Centrándose en la reactividad de la belita, β-C2S Resumen 23 (presente en X10B0) se disuelve más rápido que α'H-C2S (presente en X10B2) durante los primeros 28 días, independientemente de la fuente de sulfato, como se ha descrito anteriormente. Por otra parte, se cuantifica stratlingita cristalina en X10B0 justo después de 3 días de hidratación, pero no se detecta hasta los 28 días en las pastas de X10B2. Los análisis térmicos (DTA y TG) para las pastas G10Bx, A10Bx y B10Bx a diferentes edades de hidratación también se llevaron a cabo. A modo de ejemplo, G10B2 presenta un mayor grado de reacción a 365 días de hidratación con una mayor pérdida de peso global (∼32% en peso) comparado con el valor correspondiente para G10B0 (∼26% en peso). Por lo tanto, las familias X10B2 mostraron menores cantidades de agua libre (FW, del inglés Free Water) a edades por encima de 28 días, lo que indica un mayor grado de reacción. El análisis de los datos, a altas edades, indica que el ensamblaje de fase es poco sensible a la fuente de sulfato inicial. Los contenidos de AFt, stratlingita, katoita y AFm, entre 120 y 365 días son muy similares para G10B2 y A10B2. Comparando los resultados para G10B0 y A10B0, se detectan algunas diferencias (menores), por ejemplo, se encontraron mayores cantidades de AFm en G10B0 que en A10B0. Un estudio SEM-EDS se realizó en las pastas X10B2 para caracterizar mejor la composición química de cada fase, especialmente para las fases amorfas/mal cristalizadas. Este tipo de estudio ayuda a estimar la composición elemental (especialmente para los átomos de Si, Al y Fe) de las nuevas fases cristalinas o amorfas que aparecen con el tiempo de hidratación, y que pueden afectar a las propiedades mecánicas. Los análisis de estas pastas hidratadas revelaron que la composición química de la fase amorfa(s) en G10B2 y B10B2 a los 120 días era muy similar, pero ligeramente más rica en silicio y hierro en A10B2. La evolución de la composición química con el tiempo puede generar información interesante. La representación de las relaciones atómicas, Al/Ca vs. Si/Ca, para la pasta A10B2 a 7 y 120 días de hidratación, muestra que partículas sin una forma definida, que podrían ser fases amorfas, son ricas en aluminio a 7 días, mientras que se enriquecen en silicio con el tiempo (120 días). El primer resultado está relacionado con los primeros productos de hidratación (AFt e hidróxido de aluminio amorfo) formados a partir de la disolución de la ye'elimita y el sulfato de calcio; el segundo con la reactividad de la belita. Por otro lado, los productos de hidratación que contienen hierro son difíciles de identificar por LXRPD. La relación atómica Al/Ca vs. Fe/Ca para la pasta A10B2 a diferentes edades (7 y 120 días) fue también estudiada. Analizando partículas con forma de aguja, que se identificaron como etringita, se encontró pequeñas cantidades de hierro. Estos resultados podrían indicar que el hierro se incorpora en la estructura cristalina de la ettringita, pero se necesita más investigación para poder confirmar este hecho. Además, a 7 días la muestra presenta algunas pequeñas partículas brillantes con una composición química similar a la de la stratlingita. Sin embargo, estas partículas parecen tener un bajo grado de cristalinidad, por lo que no fueron detectadas por LXRPD a esa edad de hidratación, pero mediante el análisis Resumen 24 térmico, el ATD de esta muestra mostró un pequeño hombro a ∼170°C, que está relacionado con la stratlingita. Las propiedades mecánicas de morteros estándar (de cementos activados y no activados) se prepararon con una relación cemento/arena/agua de 1/3/0.55. Para utilizar la mínima cantidad de muestra (estudios de laboratorio), se utilizaron moldes con dimensiones de 30 x 30 x 30 mm3. Para comparar los resultados con los obtenidos utilizando los moldes que cumplen la normativa, se utilizó un factor de conversión, y los resultados se multiplicaron por 1.78. El resultado más importante es que todos los morteros preparados con el cemento BCSAF activado desarrollaron mayores resistencias a la compresión que los morteros no activados, independientemente del tipo y la cantidad de fuente de sulfato usado. Además, aumentando el contenido de yeso de 5 a 10% en peso, la resistencia aumentó en ambos sistemas. La adición de 15% en peso de yeso produce una ligera disminución en la resistencia a la compresión, probablemente debido a la disminución de la velocidad de hidratación de la belita. Dentro de los morteros preparados con los cementos no activados, el A10B0 presentó los valores más altos. Este comportamiento puede explicarse/justificarse debido a su mayor valor de área específica y la mayor estabilidad del AFt presente en esa pasta en comparación con G10B0. Los cementos con basanita reaccionan muy rápidamente con el agua mostrando un tiempo de fraguado corto, esto conduce a morteros con bajo grado de homogeneidad. Debido a este hecho, no se midieron los valores de resistencia de los morteros B10B0. Para B10B2, aunque la adición de una pequeña cantidad de SP mejora la trabajabilidad del mortero, el retraso en el tiempo de fraguado no es suficiente para desarrollar valores de resistencia comparables con los morteros de yeso y de anhidrita. A los 120 días, el mortero G10B2 desarrolló el valor de resistencia mecánica más alta (68 ± 1 MPa), incluso cuando la cantidad de etringita en la pasta con anhidrita era ligeramente mayor que en la pasta con yeso. Este comportamiento nos obliga a concluir que el contenido amorfo está jugando un papel clave para el desarrollo de las resistencias a edades más tardías. Además, el cemento G10B2, presentó el mayor valor de área específica (1.8109 + 0.0241 m2/g) de todos los cementos estudiados y esto también puede justificar los resultados obtenidos. El grado de reacción de la α'H-C2S en G10B2 (74%) es ligeramente mayor que en A10B2 (65%), que podría ayudar en la mejora de las resistencias mecánicas. La porosidad de las tres pastas de cemento activado (X10B2) a 120 días de hidratación medidas mediante porosimetría de intrusión de mercurio (MIP, del inglés Mercury Intrusion Porosimetry), se pueden extrapolar a los morteros. La pasta con basanita mostró el mayor porcentaje de porosidad (16%), y la de yeso el valor más bajo (10%); este comportamiento ayuda a justificar las resistencias mecánicas medidas. En conclusión, Resumen 25 podemos decir que la cantidad y el tipo de sulfato óptimos en estos sistemas parece estar cercanos al 10% en peso de yeso, ya que es con los que se obtienen resistencias mecánicas más altas. Finalmente, también se midió la expansión/contracción de los morteros BCSAF a diferentes edades de hidratación. El primer resultado revelado de este estudio ha sido el efecto del contenido de yeso en la estabilidad dimensional de los morteros BCSAF. La adición de 15% en peso de yeso dio los valores de expansión más altos para ambos sistemas. Los morteros G5B2 y G10B2 mostraron una variación de longitud de entre -0.01% y 0.04% en el plazo de 180 días. En segundo lugar, también se estudió el efecto de la fuente de sulfato en la estabilidad dimensional. Los morteros preparados con anhidrita presentaron los valores más altos de expansión a edades tempranas, de forma similar a los morteros de sulfoaluminato de calcio (CSA, del inglés Calcium Sulpho-Aluminate). Sin embargo, el mortero B10B2 mostró valores de expansión muy bajos, similares a los morteros preparados con un 5% en peso de yeso, de acuerdo con los resultados obtenidos en el estudio de hidratación, en el que la basanita se disuelve completamente antes de los primeros 45 minutos de hidratación para dar la precipitación de yeso. En consecuencia, los morteros con basanita se comportan de manera similar a los realizados con yeso a edades muy tempranas. Por último, me gustaría resaltar que pertenezco a un grupo de trabajo que tiene una amplia experiencia en la caracterización de clínkeres y cementos anhidros por difracción de polvo de rayos X combinada con la metodología Rietveld, y en el procesamiento de materiales (cerámicos). En este punto me gustaría destacar mi contribución en la síntesis (a mayor escala) y la caracterización de clínkeres BCSAF y pastas de cemento, incluyendo la cuantificación del contenido ACn, y la medida de las propiedades mecánicas (resistencia a la compresión y cambio longitudinal) de los correspondientes morteros preparados. Chapter 1 32 1.3.1. Classification. Although ye'elimite-containing cements are very promising, their use is limited in Europe by the few standards concerning special cements derived from nonPortland clinkers. At the present state of European standard regulations, binders based on ye'elimite-containing cements cannot be used in structural concrete according to the EN 206–1; only three formulations of CSA cements, produced by Buzzi Unicemin Trino (Italy), obtained in June 2013 a CE mark based on an ETA procedure, released by DIBt, allowing their use for structural applications (Paul et al., 2015). There are very wide ranges of phase assemblages in ye'elimite-containing cements (Zhang et al., 1999; Quillin, 2001). These eco-cements can be classified according to the content of their main crystalline phase. Aranda and De la Torre (2013) unified the terminology used for these cements, and gathered them in three mains groups: i) CSA: CSA cements are prepared from clinkers containing a high amount of C4A3S (50-80 wt%) (Sahu and Majling, 1993; Zhang et al., 1999; Older, 2000; Glasser and Zhang, 2001). These clinkers may also have minor phases such as C2S, CT, C4AF, CS and others. The calcium sulphate addition is very important as it may strongly affect the properties of the resulting binder (Winnefeld and Barlag, 2010; Marchi and Costa, 2011; Berger et al., 2011a; Chen et al., 2012; Bizzozero et al., 2014; GarcíaMaté et al., 2015a). Due to the high amount of expensive aluminium source needed in their productions, CSA clinkers cannot replace OPC in massive constructions. Therefore, CSA clinkers can be prepared or partially substituted by different industrial by-products or waste materials as source of aluminium, calcium and silica (Viani and Gualtieri, 2013; García-Maté et al., 2013; Shen et al., 2014) to reduce cost but maintaining their performances. CSA clinkers show a reduction of ∼37% of the CO2 emission footprint when compared to OPC. ii) BCSA: this term is reserved to the cements arising from clinkers containing C2S as main phase (40-50 wt%) and intermediate C4A3S contents (20-30 wt%). These cements, also known as sulphobelite are a new emerging type of ye'elimitecontaining cements due to their environmental benefits. The most common formulation of BCSA clinkers consists on β-C2S, C4A3S and C4AF (Janotka and Krajci, 1999; Janotka et al., 2007; Adolfsson et al., 2007). These cements are called BCSAF. The clinkering temperature of these clinkers is 1250-1300°C, ∼200°C lower than that for OPC. In addition, BCSAF clinkers are porous, because of that can be easily ground. These cements also show a rapid hardening, excellent durability, self-stressing and volume stability, depending on the amount of gypsum added (Pera and Ambroise, 2004). These cements are being studied with the final aim of replacing OPC as the Introduction 33 aluminium demand is much smaller when compared to that of CSA. In addition, the production of BCSAF clinker shows a relative reduction of ∼22% of the CO2 footprint when compared to OPC manufacturing. Other type of BCSA cements are the BCSAA. They have C2S, C4A3S, C12A7 and CA as main phases (Martín-Sedeño et al., 2010) and they are prepared in order to further enhance mechanical strengths at very early ages. Nevertheless, the clinkering temperature should be increased (around ∼100°C higher than that of BCSAF) and moreover, higher amounts of expensive bauxite (or another aluminium-rich source) are needed. Recently, a new formulation of BCSA cement which contains alite jointly with ye'elimite has been published (Liu et al., 2013). These cements are known as BACSA cements. Their manufacture may produce 15% less CO2 than OPC. The reaction of alite and ye'elimite with water will develop cements with higher mechanical strengths at early ages, while belite will contribute to later values (Liu and Li, 2005; Lili et al., 2009; Liu et al., 2002; 2009; 2013). BACSA cements may contain ∼40 wt% of belite, ∼20 wt% of ye'elimite and ∼20 wt% of alite. BACSA would overcome the problems of BCSA cements since the basicity of the pastes should be higher due to the presence of alite, and therefore, the pozzolanic effect with fly ash or slag will be promoted. iii) ACSA: ACSA cements are characterized by the simultaneous presence of C3S and C4A3S phases. In this special case, ye'elimite phase content may be even higher than that of alite (Li et al., 2007b). Other phases which may appear in these clinkers, in smaller amounts, are C2S, C4AF and C3A (Abdul-Maula and Odler, 1992; Odler and Zhang, 1996; Zhang and Odler, 1996). However, there are inherent difficulties in the production of this type of clinker due to the differences between the optimum formation temperatures of the main phases. On the one hand, alite formation is favoured by the presence of melted phases (De la Torre et al., 2007) and at least a temperature of 1350°C is required. On the other hand, decomposition or melting of ye'elimite takes place above 1350°C (De la Torre et al., 2011a, b). However, the addition of minor quantities of fluorite (Ma et al., 2006) and other minor elements such as Mg (Liu and Li, 2005), Cu (Ma et al., 2006), Mn (Lili et al., 2009), Ti (Liu et al., 2009) or Zn (Pérez-Bravo et al., 2014) to raw materials will yield to the coexistence of these two phases by using clinkering temperatures of ∼1300°C. Recently, a new (different) strategy to produce cements with alite and ye'elimite has been published (Ma et al., 2013), which consists on a two-step clinkering cycle, one to form alite at 1450°C and a second one to re-crystallise ye'elimite at 1250°C. 1.3.2. Commercial ye'elimite-containing cements. Nowadays, there are some commercial ye'elimite-containing clinkers/cements being marketed and used for special applications in Europe, e.g. Chapter 1 34 S.A. Cement from Buzzi-Unicem, ALIPRE® 2009 from Italcementi Group, or CSA cement (model number 62.5, 72.5, 82.5, 92.5) from Tangshan Polar Bear Building Materials, China, among others. An industrial trial of low energy belite-based cements was reported by Popescu et al. (2003), highlighting environmental profits of belite-rich materials and even concluded that these cements developed higher mechanical strengths than OPC at very large hydration ages (after 90 days). However, mechanical strengths at early ages resulted much lower than those of a typical OPC due to the low C4A3S content (∼12 wt%). In the last few years, BCSAF cements have emerged with the initial aim to substitute OPC. An industrial trial, ∼2500 tons of BCSAF, was carried out in 2011 by Lafarge under the AETHERTM project (http://www.aether-cement.eu/) (Walenta and Comparet, 2011). Clinkering was performed in a Portland industrial kiln but using lower operating temperatures (1225-1300°C), reducing CO2 emissions by 25-30% in comparison with CEM (I) type OPC cement. The temperature in the clinkering zone resulted to be a key parameter, as too high temperatures may give kiln blockage, loss of grindability and C4A3S decomposition with high SO2 emissions, and too low temperatures gave under burnt binder with high free lime and C12A7 contents. The optimum clinkering temperature resulted much lower in NOx emissions than those for OPC, due to the lower burning temperature, and SO2 emissions resulted at the same level than for OPC production. The main technological disadvantage of these cements is related to their low mechanical strengths developed at very-early ages due to the slow hydration of belite. However, this problem is being overcome by the activation of belite and the presence of high amounts of ye'elimite (Gartner and Li, 2006; Cuberos et al., 2010; Morin et al., 2011). The production of aBCSAF cements involves the stabilization of highly reactive C2S polymorphs, i.e. β-modified form and α-forms, as they react faster with water. aBCSAF cements have been patented by Lafarge (Gartner and Li, 2006; Morin et al., 2011). They show ∼20-30 wt% of Klein’s salt and α-forms of belite, the latter due to addition of minor elements, such as B2O3 and Na2O. These minor elements promoted the stabilisation of α'-forms of belite and the distortion of βform. aBCSAF with 2.0 wt% of B2O3, added as borax, developed comparable compressive strengths to those of OPC. The role of belite polymorphs is explained in detail in section 1.4.3. This material developed mechanical strengths of 25 MPa at one hydration day and the strengths were even better than those developed by type I 52.5 OPC after 7 days (Walenta and Comparet, 2011). Introduction 35 1.4. BELITE CALCIUM SULPHOALUMINATE CEMENTS (BCSA). Belite calcium sulphoaluminate cements are environmentally friendly cements for several reasons, including the low amount of limestone required to achieve the desired composition. Table 1.1 shows the lime contents of the main phases present in BCSAF clinkers and C3S, present in OPC. C3S shows the highest lime amount and requires the highest formation temperature. The substitution of alite by belite reduces the limestone demand and temperature of cement manufacturing, thereby reducing CO2 emissions and energy consumption while maintaining satisfying long-term properties. However, that substitution compromises the early-age strength development because C2S reacts slower than C3S. This renders the high-C2S cement unusable in nearly any field related to structural applications. However, this is partially compensated by the presence of a fast reacting calcium sulphoaluminate phase (ye'elimite) in these BCSA cements (Mehta, 1980). This phase, that was described by Alexander Klein in 1963 as an additive to Portland cement to make expansive cements, has the lowest lime content of any of the cement phases reported in Table 1.1. Table 1.1. Lime content for different cement phases (Mehta, 1980). Cement Phase Lime Content (wt%) C2S 65.1 C4A3S 36.7 C4AF 46.2 C3S 73.7 Raw meal formulations for BCSA clinkers have been discussed extensively in the literature (Gartner and Li, 2006; Idrissi et al., 2010, 2012; Martín-Sedeño et al., 2010; Cuberos et al., 2010). However, due to environmental and cost concerns, BCSA cements are mostly produced by combining natural and industrial waste materials to provide the necessary CaO, SiO2, Al2O3 and SO3 amounts required for each phase formation. The addition of waste materials in the raw mixtures to produce BCSA clinkers also has been investigated (Arjunan et al., 1999; Katsioti et al., 2006; Phair, 2006; Seluck et al., 2010; Yang et al., 2013; Ma et al., 2014a). For example, Yang et al. (2013) studied the use of the phosphate fertilizer industry waste products as raw materials to prepare BCSAF cements with similar performances to those obtained by using common grade raw materials. This material is not only an iron source but also supplies silicon, aluminium and sulphur to reduce burning temperature. Chapter 1 36 Other authors (Strigac et al., 2000) studied the composition of ferrite phase in BCSA cement and showed that this phase formed a solid solution and accommodated some SiO2. It is also known that the actual composition of the ferrite phase solid solution can range from C4AF to C6AF2. C4A3S accommodated some SiO2 and up to 2% of Fe2O3. C2S can accommodate minor elements such as sulphur or alkaline (Ghosh et al., 1979) and on the other hand, CS forms more homogenous compositions with minor substitute ions. In the last decade, the research in the preparation of BCSAF clinkers has increased (De la Torre et al., 2011a, b; Chen et al., 2011). A recent study (Bullerjahn et al., 2014) proved that iron-rich BCSA contains ye'elimite with significant amounts of iron stabilizing the cubic form and enhancing its hydration rates. Moreover, they stated that ternesite (C5S2S) was formed under certain clinkering conditions and it was found to be hydraulically active. In addition, stoichiometric C4A3S is orthorhombic at room temperature (RT) (Calos et al., 1995; Cuesta et al., 2013). However, it can form solid solution with Na+, B3+, Si4+ and/or Fe3+, among other elements, to restore the pseudo-cubic symmetry (Saalfeld and Depmeier, 1972; Cuesta et al., 2014c). Sokol et al. (2014) have recently reported natural ye'elimite-larnite rocks found in the Hatrurim formation of Negev Desert, Israel, which mineralogical compositions are similar to BCSA clinkers. Their mineralogy consists on 35-50 wt% βC2S, 15-20 wt% C4A3S, 7-15 wt% ferrites, and 15–20 wt% fluorapatite and/or fluorellestadite. This study concludes that chalky and/or marly sediments with randomly distributed clay, phosphorite, and gypsum may be used as cheap naturally homogenised and pulverised mixtures for industrial production of BCSA cement clinkers. 1.4.1. Reduction of CO2 emissions. Carbon dioxide emissions in the cement industry can be classified in two main categories: those coming from raw materials and those from the operation processes. Figure 1.2 depicts in detail the CO2 emissions released during the manufacturing of both a typical BCSAF clinker (50 wt% of C2S, 30 wt% of C4A3S, and 20 wt% C4AF) and OPC (65 wt% of C3S, 15 wt% of C2S, 10 wt% of C3A and 10 wt% of C4AF). Considering all emissions together, the production of one ton of OPC clinker releases a maximum of 0.97 t of CO2; however the production of one ton of BCSAF clinker leads to a reduction of ∼22%, which will depend on the composition. On the one hand, the generation of CO2 from decarbonation of raw materials for different minerals was calculated by Gartner (2004). For example, RMCO2 values for C4A3S, C2S and C4AF resulted to be 0.216, 0.511 and 0.362, respectively. However, Introduction 37 the value for C3S (major phase of OPC), is much higher, 0.578 (Gartner, 2004). According to these values, CO2 emissions released during clinkering of 1 ton of BCSAF and OPC are 0.54 t and 0.39 t, respectively (figure 1.2). It implies a reduction of 0.15 tons of CO2 for BCSAF due to raw material decomposition. On the other hand, CO2 emissions derived from operation processes are directly related to the type of processing equipment and the specific chosen fuel. Gartner in 2004 estimated 0.30 t of CO2 per ton of clinker produced assuming that good quality of bituminous coal is used and taking into account energy efficiency of modern kilns. The reduction in CO2 emissions coming from fuel burning can be achieved by different strategies (Gartner, 2004; Juenger et al., 2011) including the reduction of clinkering temperature. This is the case of BCSAF clinkers where the operating temperature can be reduced down to 150°C with a concomitant reduction of up to 0.04 t of CO2 per ton of BCSAF clinker produced. Moreover, emissions derived from electricity consumption are about 0.09 t of CO2 per ton of OPC (McCaffrey, 2002). The lower firing temperatures needed for BCSAF clinkering make it easier to be ground, hence it yields a depletion of up to 0.02 t of CO2 emissions. OPC Raw materials 78 wt% calcite 10 wt% kaolin 10 wt% quartz 2 wt% iron oxide OPC clinker Phases 65 wt% C 3 S 15 wt% C 2 S 10 wt% C 4 AF 10 wt% C 3 A BCSAF clinker Phases 50 wt% C 2 S 30 wt% C 4 A 3 S 20 wt% C 4 AF 0.09 0.34 0.54 Electricity Fuel Raw materials 0.30 0.39 OPC BCSA 0.07 BCSAF Raw materials 59 wt% calcite 9 wt% kaolin 7 wt% quartz 17 wt% red bauxite 8 wt% gypsum ~0.97 t ~0.76 t Clinkering kiln 1.00 t t CO 2 / t clinker Al 2 O 3 approximate contents of the clinkers, wt%, just above ~17% ~6% Figure 1.2. Comparison of OPC and BCSAF production CO2 footprint. Chapter 1 38 1.4.2. Crystallochemistry of main anhydrous phases of BCSAF cements. The crystallography of the main anhydrous phases present in BCSAF cements (belite, ye’elimite and ferrite) is described. i) Belite or dicalcium silicate [Ca2SiO4, C2S]. Belite is of great interest in BCSAA and BCSAF cements as it is their main component (∼60 wt%). Stoichiometric dicalcium silicate presents five forms (Mumme et al., 1996) γ, β, α'L, α'H and α. On the other hand, element substitutions stabilize different structures of dicalcium silicate and β-C2S is the form that commonly prevails in OPC, BCSAF and BCSAA without any activation. All the structures are built from Ca2+ and SiO44ions. The arrangements of these ions are closely similar in α, α'L, α'H and β polymorphs, but that in γ-C2S is somewhat different. γ-C2S is much less dense than the other polymorphs. These polymorphic transformations, as a function of temperature, are shown in Figure 1.3. The physical and chemical properties of these phases can be altered by introducing defects or strains in their crystalline structures. Moreover, these defects can even stabilize high-temperature forms of C2S at room temperature (Ghosh et al., 1979; Nettleship et al., 1992). The different type of defects can be produced by the addition of foreign elements to form solid solutions (Jelenic et al., 1978) or by specific thermal treatments (Fukuda and Ito, 1999). There are many studies concerning the chemical-stabilization of β-C2S by foreign ions such as SO3, B2O3, Cr2O3, Na2O, K2O, BaO, MnO2 and Al2O3 (Pritts and Daugherty, 1976; Kantro and Weise, 1979; Matkovic et al., 1981; Fierens and Tirlocq, 1983; Ziemer et al., 1984; Benarchid et al., 2004; Cuesta et al., 2014b). Figure 1.3. Dicalcium silicate polymorphic transformations with temperature. These investigations showed that the hydration reactivity of stabilized β-C2S depends on the preparation parameters that influence particle and crystallite size. These parameters include temperature, type and amount of stabilizer, and fineness of the final ground. The stabilization of α'-forms through the introduction of foreign Introduction 39 oxides, such as MgO, P2O5, K2O, BaO, Na2O, B2O and SO3 has also been studied (Bensted, 1979; Fukuda et al., 2001; Park, 2001; Morsli et al., 2007a, b; Li et al., 2007a; Wesselsky and Jensen, 2009; Cuberos et al., 2010; Morin et al., 2011; Cuesta et al., 2012). These works stated that hydraulic properties were increased when compared to the materials without foreign ions. Table 1.2 shows the crystallographic information of dicalcium silicate polymorphs. The crystal structures of α'H, α'L and β phases are derived from that α form by progressive decreases in symmetry, which arise from changes in the orientations of the SiO44tetrahedral and small movements of Ca2+ ions (Taylor, 1997). Such structures belong to the family of the glaserite, K3Na(SO4)2 (Moore, 1973). However, the γ-C2S framework belongs to the olivine type structure (O'Daniel and Tescheischwili, 1942). Tabla 1.2. Crystallographic data of dicalcium silicate polymorphs. Polymorph Space Group Unit cell parameters V/Z T(°C)/ a (Å) b (Å) c (Å) β ββ β (º) (Å3) Stabilizer ICSD α αα α P63/mmc a 5.420 5.420 7.027 90.0 89.4 - 81099 P63/mmc b 5.532(9) 5.532(9) 7.327(11) 90.0 97.1 1545/- 82998 P-3m1 b 5.532(9) 5.532(9) 7.327(11) 90.0 97.1 1545/- 82999 α αα α'H Pnmaa 6.7673(4) 5.5191(4) 9.3031(6) 90.0 86.9 -/5% (molar) Ca 3 (PO 4 ) 2 81097 Pnma b 6.871(0) 5.601(0) 9.556(1) 90.0 92.0 1250/- 82997 Pmnbc 5.647(1) 7.037(1) 9.644(2) 90.0 95.8 - /Ca 1.8 Sr 0.2 SiO 4 49662 Pnmad 6.8263(3) 5.4684(3) 9.2658(4) 90.0 86.5 RT/ B2O3 and Na 2 O - α αα α'L Pna21 b 20.527(2) 9.496(1) 5.590(1) 90.0 90.8 1060/- 82996 Pna21d 20.863(2) 9.5000(8) 5.6005(5) 90.0 92.5 -/ Ca 0.84 Sr 1.16 SiO 4 39203 P21cn e 5.566 9.355 20.569 90.0 89.3 - 39100 β ββ β P21/na 5.512(0) 6.758(0) 9.314(0) 94.6 86.5 -/0.5 wt% Cr 2 O 3 81096 P21/n f 5.48(2) 6.76(2) 9.28(2) 85.5 85.7 - 24640 P21/n g 5.502(1) 6.745(1) 9.297(1) 94.6 86.0 - 963 P21/n i 5.5127(1) 6.7586(1) 9.3266(2) 94.5 86.6 RT/Al2O3 - γ γγ γ Pbnm a 5.082(0) 11.224(0) 6.764(0) 90.0 96.5 - 81095 Pbnm h 5.081(2) 11.224(5) 6.778(10) 90.0 96.6 - 200707 (a) Mumme et al., 1995; (b) Mumme et al., 1996; (c) Catti et al., 1984; (d) Cuesta et al., 2012; (e) Il´inets and Bikbau, 1990; (f) Udagawa et al., 1979; (g) Midgley, 1952; (h) Jost et al., 1977 ; (i) Cuesta et al., 2014b; (j) Udagawa et al., 1980. Chapter 1 40 Figure 1.4 shows the simulated diffractograms of different C2S polymorphs. The presence of ions in the structure stabilizes the high temperature forms at RT. The stabilization of the high temperature polymorph is due to both ionic substitutions and quenching. It has been found the existence of α and α'H forms in clinkers (Regourd and Guinier, 1974; Morsli et al., 2007a, b). Figure 1.4. Range from 29 to 35° (2θ) of the theoretical diffractograms. (a) α-C2S at RT (Mumme et al., 1995); (b) α'H-C2S at RT (Cuesta et al., 2012); (c) β-C2S at RT (Mumme et al., 1995) and (d) γ-C2S at RT (Udagawa et al., 1980). ii) Ye'elimite, also named Klein´s salt or calcium sulphoaluminate, [Ca4Al6O12(SO4), C4A3S]. Ye'elimite is the major mineralogical compound in CSA cements. It is liable for high mechanical strengths at early hydration ages. Stoichiometric calcium sulphoaluminate or ye'elimite, can be described as a sodalite (M4[T6O12]X) where M=Ca, T=Al, and X=SO4, and crystallizes as a tectoaluminosilicate sodalite structure Introduction 41 (Cuesta et al., 2013). This structure was first analyzed by Hanstead and Moore (1962) using X-ray powder diffraction. Wang et al. (1990) and Saalfeld and Depmeier (1972) reported atomic parameters for a cubic crystal structure with space group I-43m and a=9.195 Å. In 1995, Calos published an orthorhombic crystal structure, Pcc2 space group, which has been revised by joint neutron and X-ray powder diffraction, Rietveld refinement and atomistic calculations (Cuesta et al., 2013). Recently, the disordered crystal structure of cubic stoichiometric ye'elimite at 800°C has been satisfactorily studied in the I-43m space group using a split-atom model (Kurokawa et al., 2014). In a recent study (Bullerjahn et al., 2014) the existence of two polymorphs of ye'elimite (pseudo-cubic and orthorhombic) was observed in BCSA cements. Published crystallographic data for the Klein´s salt are shown in Table 1.3. Fe-Sidoped ye'elimite shows a pseudo-cubic structure at room temperature and suffers a phase transition on heating (Cuesta et al., 2014c). Table 1.3. Crystallographic data of Klein´s salt polymorphs. Polymorph Space Group Unit cell parameters V/Z Ref. a (Å) b (Å) C (Å) β ββ β (º) (Å 3 ) bibl. ICSD Cubic I-43m 9.205 9.205 9.205 90.0 390 (a) 9560 Orthorhombic Pcc2 13.028 13.037 9.161 90.0 389 (b) 80361 Tetragonal P-4c2 13.031 13.031 9.163 90.0 389 (c) - Cubic I-43m 9.197 9.197 9.197 90.0 389 (d) - Orthorhombic Pcc2 13.036 13.035 9.168 90.0 346 (e) - Cubic at 1073 K I-43m 9.253 9.253 9.253 90.0 396 (d) - Cubic at 1073 K I-43m 9.243 9.243 9.243 90.0 395 (f) - (a) Saalfeld y Depmeier, 1972; (b) Calos et al., 1995; (c) Zhang et al., 1992; (d) Cuesta et al., 2014c; (e) Cuesta et al., 2013; (f) Kurokawa et al., 2014. Figure 1.5 shows the simulated patterns of two of the C4A3S polymorphs (orthorhombic and pseudo-cubic). As seen in this figure, all the peaks of the pseudocubic structure are also present in the orthorhombic form. We highlight that the RT pseudo-cubic form is not a strict ye'elimite polymorph as it required the presence of dopants within the crystal structure. Chapter 1 48 The gypsum released can react with any remaining C4AF to form further AFm (C4ASHn) with overall stoichiometry: C4AF + CSH2 + (n+2) H → C4ASHn + FH3 + CH [1.11] However, the hydration process involving iron-containing phases may be more complex, as the formation of solid solutions between Feand Al-containing hydrates may stabilize mixed solids, such as Fe-AFt (Möschner et al., 2009) and FeAFm (Dilnesa et al., 2012). 1.4.4. Crystallochemistry of main hydrated phases of BCSAF cements. i) Ettringite or AFt phase. AFt is the abbreviation for "alumina, ferric oxide, tri-sulfate" or (Al2O3-Fe2O3tri). It represents a group of calcium sulfoaluminate hydrates and has the general formula [Ca3(Al,Fe)(OH)6·12H2O]2·X3·nH2O, where X represents a doubly charged anion or, sometimes, two singly charged anions. Ettringite is the most common and important member of the AFt group and, in this case, X denoting sulfate (X=SO42-). The crystal structure of ettringite can be described as compact columns of [Ca3Al(OH)6·24H2O]3+, running parallel to the c-axis, with 3SO42and H2O molecules in the intervening channels. Figure 1.7 shows a revised structure model of ettringite presented by Goetz-Neunhoeffer and Neubauer (2006). Figure 1.7. Hexagonal ettringite unit cell of the revised structure projected down c. (Goetz-Neunhoeffer and Neubauer, 2006). Introduction 49 ii) AFm phases. Phases in this family (Al2O3-Fe2O3-mono) are comprised of a layer structure with the general formula [Ca2Al(OH)6]X·xH2O where X denotes one formula unit of a singly charged anion, (for instance OHor [SiAlO2(OH)4]- in the case of stratlingite, or half a formula unit of a doubly charged anion (for instance SO4 2-) placed in the interlayer space jointly with water molecules. These compounds crystallize in hexagonal planes and are very relevant to cement hydration. AFm are formed from aluminium rich phases, such as C4A3S, when the sulphate source (gypsum, bassanite or anhydrite) is depleted and there is enough free water available (Winnefeld and Lothenbach, 2010). The presence of AFm-phases with different layer spacing is justified twofold: i) by the partial anion replacement OH-/SO4 2within the layers; and ii) by the progressive release of the water molecules as a consequence of the hydration of other phases. Stratlingite (Rinaldi et al., 1990), kuzelite (Allmann, 1977) or C2AH8 are AFm-type phases. There are other AFm-type phases such as monocarbonates (François et al., 1998) or monochlorides (Renaudin et al., 1999) that usually appear in chemically aggressive environments. Stratlingite: Ca4Al2(OH)12[AlSiO2(OH)4]2·2H2O, is an AFm phase which appears as hydration product of aluminium-rich cements, such as calcium aluminate, calcium sulphoaluminate and also belite calcium sulphoaluminate cements. The structure of stratlingite is known from single crystal studies of mineral fragments from Mayern and Montalto di Castro (Rinaldi et al., 1990). It is formed by a principal octahedral or brucite-type layer, [Ca2Al(OH)6·2H2O]+ with a full occupancy, and a double tetrahedral layer, [(T,□)4(OH,O)8·0.25H2O]-, with a 45% of vacancies. The symmetry of stratlingite is R3m. The structure analysis (Rinaldi et al., 1990; Kwan et al., 1995) indicates that the octahedral layer shows an ordered scheme where each Al-octahedron is linked to 6 edge-sharing CaVII polyhedra (2 out 3 positions are then occupied by the seven coordinated Ca-type cation). Figure 1.8 shows the crystal structure of stratlingite where selected atoms are labelled (Santacruz et al., 2015). It is known that the structural and microstructural models have important implications for a correct quantitative phase analysis of stratlingite in cement pastes and it is highly dependent on the hydration conditions (for instance, in pastes of BCSA cements) (Santacruz et al., 2015). Chapter 1 50 Figure 1.8. Layered crystal structure of stratlingite (Santacruz et al., 2015). iii) Katoite. Katoite has a structure related to the grossular or garnet. Garnet minerals have a cubic structure with the general formula X3Y2(SiO4)3. The X site is usually occupied by divalent cations (Ca2+, Mg2+ and Fe2+) and the Y site by trivalent cations (Al3+, Fe3+ and Cr3+) in an octahedral/tetrahedral framework with [SiO4]4− occupying the tetrahedral positions (see Figure 1.9). The anhydrous end members of the Ca3(Al,Fe)2(SiO4)3 series are grossular (Ca3Al2(SiO4)3) and andradite (Ca3Fe2(SiO4)3). Hydrogarnet (Ca3(Al,Fe)2(SiO4)y(OH)4(3−y); 0 < y < 3) includes a group of minerals where the [SiO4]4− tetrahedra are partially or completely replaced by OH−. The Al-containing hydrogarnet includes hydrogrossular (Ca3Al2(SiO4)y(OH)4(3−y); 0 < y < 3) with the end member katoite (Ca3Al2(OH)12 or C3AH6). The nomenclature of minerals of the hydrogarnet group Ca3(AlxFe1−x)2(SiO4)y(OH)4(3−y) as recommended by Passaglia and Rinaldi (1984) is given in Figure 1.10. The formation of siliceous hydrogarnet was reported for cements hydrated at high temperatures (Collier et al., 2009; Le Saout et al., 2006; Lothenbach et al., 2008; Neuville et al., 2009) or in the presence of excess of Fe(OH)3 (Collier et al., 2006). Introduction 51 Figure 1.9. Octahedral and tetrahedral connections of hydrogarnet structure (Dilnesa et al., 2014). Figure 1.10. Nomenclature of minerals of the hydrogarnet group Ca3(AlxFe1−x)2(SiO4)y(OH)4(3−y) (Passaglia and Rinaldi, 1984). Chapter 1 52 1.4.5. Properties of BCSA mortars and concretes. Durability. Ye'elimite-containing cements and their corresponding mortars and concretes are considered as durable binders. Nevertheless, degradation processes should be taken into account, depending on the environment on which these mortars or concretes are serving. There are different ways of cement degradation e.g. carbonation, sulphate attack or chlorine diffusion. All these effects need ions diffusion through the porous microstructure. In CSA and BCSAF cements, the resistance to diffusion is enhanced due to the quick water consumption during the hydration process. Moreover, porosity decreases during hydration due to the large amount of hydration products generated at very early ages (Bernardo et al., 2006). 1.4.5.1. Sulphate Resistance. Cementitious building materials are damaged by sulphate attack at moderate temperatures due to delayed ettringite formation. In OPC, when insufficient gypsum is added, ettringite reacts with the aluminium containing phases (C3A and C4AF) to form AFm (C4ASH12) (Mindess et al., 2003). Calcium monosulphoaluminate is subject to transform back to ettringite in the presence of environmental sulphate, causing cracking due to the enlargement of volume. However, BCSA cement generally shows good sulphate resistance (Quillin, 2001; Dan and Janotka, 2003) even for insufficient amounts of calcium sulphate in the system. The durability of these materials is usually studied through XRPD (Paglia et al., 2001), microscopy, linear expansion and compressive strengths. 1.4.5.2. Corrosion Resistance. Atmospheric carbon dioxide can be dissolved in the pore solution of cement pastes and react with the hydration products increasing the carbonate concentration. This phenomenon will cause three effects: i) lowering pH value, ii) precipitation of CaCO3 by the reaction of carbonate ions with Ca2+ ions and iii) possible damaging of ettringite. Porosity and water/cement (w/c) ratio determine the degree of resistance to carbonation. Carbonation happens quicker in CSA than in Portland concretes, leading to partial decomposition of ettringite, which may cause a moderate strength loss (Sharp et al., 1999; Mechling et al., 2013). In addition, decreasing w/c ratio (highstrength concrete) increases resistance to carbonation as there is not free water available to dissolve CO2. The pH of the pore solution in BCSA cements is generally lower than that in OPC due to the lower amount of CH formed in BCSA cements (Kalogridis et al., 2000). Furthermore, ettringite, the main hydration product in BCSA cements, is susceptible to carbonation, which further reduces the pore solution pH at later ages (Quillin, Introduction 53 2001). In steel reinforced concrete, a passivated oxide layer is formed on the reinforcement steel surface and protects it from corrosion when the pH of the surrounding pore solution is high (Jones, 1996). However, this passivated layer breaks down when the surrounding pore solution pH is below ∼11.5. The low pH of the pore solution in BCSA cement might not be high enough to protect the reinforcement steel, leading to corrosion in the reinforcement steel if cracking is developed in the concrete. 1.4.5.3. Compressive Strength. BCSA cements can be produced with a wide range of properties from rapid to slow hardening and good to poor durability depending on their phase assemblage (Mehta, 1980), as it is shown in Figure 1.11. The hydration reactions of the fastreacting C4A3S and C4AF with CS to form ettringite and AH3 resulted in rapid setting and contributed to the high early-age strength development, while the hydration reaction of the slow-reacting C2S to form C-S-H contributed to the long-term strength development. However, mortars made from BCSAF cements that contain 10 wt% C4A3S, 50-65 wt% C2S, 10 wt% CS and 15-30 wt% C4AF with a water-to-cement ratio of 0.4 show slow hardening, poor early-age and long-term compressive strengths likely due to their low C4A3S, C4AF and CS contents and high C2S contents. Sulphates (C4A3S + CS) C4AFC2S Figure 1.11. Schematic phase diagram of ternary system representing BCSA cement; slow, normal and rapid refer to hardening rates (adapted from Mehta, 1980). Chapter 1 54 1.4.5.4. Dimensional Stability. Other problem related to durability is the expansion of concrete during hydration. Differences in dimensional stability are observed in BCSA systems by different factors: i) C4A3S content (Kasselouri et al., 1995; Beretka et al., 1996; Janotka et al., 2007; Chen et al., 2012); ii) amount of gypsum available in the system (Glasser and Zhang, 2001; Bizzozero et al., 2014); iii) w/c ratio and curing conditions (humidity degree) (Beretka et al., 1996; Odler and Colan-Subauste, 1999; Chen et al., 2012). The sulphate diffusion into the mortar matrix may delay the ettringite formation and cause expansion (Chen et al., 2012). When most of the ettringite is formed before the paste hardens, non-expansive and rapid hardening BCSA cements can be achieved. However, the formation of a significant amount of ettringite after hardening may cause expansion and cracking (Ogawa and Roy, 1982). However, after some days of hydration water may be exhausted and, unless there is another deterioration mechanism (as leaching) (Berger et al., 2011b), the expansion due to the delayed formation of ettringite is not significant (Glasser and Zhang, 2001). For the reasons discussed above, BCSAF cements are considered, nowadays, as one of the most promising alternatives to OPC. This is supported by the environmental benefits (lower CO2 emissions) and industrial interest. However, before been implanted in Europe, all the steps involved in the process need to be under control, including synthesis (activation, clinkering conditions and composition), hydration (rheological behaviour and phase assemblage), and final performances (mechanical strength, durability, and dimensional stability). This PhD Thesis is focused on the study and optimisation of most of those parameters to improve the final performances of BCSAF mortars. 1.5. METHODOLOGY. The most important characterisation techniques used in this work are described in this section. They are gathered in two mains groups: those used for cements and those for pastes and mortars. 1.5.1. Cement (anhydrous and paste) characterisation. 1.5.1.1.LXRPD and Rietveld method. Introduction 55 XRPD can reveal the mineralogical nature of phases present in the solid. Principle is that a beam of X-ray penetrates the sample, and a constructive interference occurs between the X-rays and the crystals in the sample. The angle and the intensity of diffraction are characteristic of a crystalline structure. All lines of diffraction are used to identify the nature of crystals. LXRPD studies can be performed using different radiations (Mo and Cu) and setups: i) Transmission geometry (Debye-Scherrer), and ii) Reflection geometry (Bragg-Brentano). The irradiated volume in diffraction is a key issue since higher volume yields enhanced particle statistics. For a given sample, several methods can be used to increase the number of crystallites contributing to the diffraction pattern, including: i) rotate the sample about the normal to the sample surface for a flat plate sample or the sample axis for a capillary sample-holder; ii) oscillate the sample about the incident angle axis, this motion removes the exact Bragg-Brentano theta/2theta relationship between sample and receiving slit and may lead to aberrations in the peak intensities; iii) repack the sample, recollect and reanalyze the diffraction data, averaging the results from each analysis will produce more meaningful parameter values, iv) reduce the average crystallite size(s) by milling, however, caution must be exercised in the choice of mill since many grinding techniques introduce peak broadening (amorphization or inducing crystal strains) and some phases can undergo solid-solid phase transitions or dehydration during grinding; and v) enhancing particle statistic by the increasing the diffracting volume. The irradiated volume for Moradiation (transmission with flat sample) is higher than the volume for Cu-radiation (reflection). This larger irradiated volume for Mo-radiation should show several benefits that will be demonstrated in the results section. The experimental set-up details for different diffractometers used for this Thesis were the following: - X'Pert MPD PRO diffractometer (PANalytical B.V.) with strictly monochromatic CuKα1 radiation (λ=1.54059 Å) and primary monochromator Ge (111). The optics configuration was a fixed divergence slit (1/2°), a fixed incident antiscatter slit (1°), a fixed diffracted antiscatter slit (1/2°) and X'Celerator RTMS (Real Time Multiple Strip) detector, working in scanning mode with maximum active length. Data were collected from 5° to 70° (2θ) for ∼2h. The samples were rotated during data collection at 16 rpm in order to enhance particle statistics. - EMPYREAN diffractometer (PANalytical B.V.): used to recorded in flatsample transmission geometry the patterns studied by the internal standard method. It has equipped with a θ/θ goniometer, CuKα1,2 radiation (λ=1.542 Å) and a focusing Chapter 1 56 mirror. This PreFIX optical component is capable of converting the divergent beam into a convergent radiation focused on the goniometer circle. The EMPYREAN diffractometer was equipped with fixed incident and diffracted beam anti-scatter slits of ¼° and 5 mm, respectively. The detector was PIXCEL 3D RTMS, which comprises more than 65,000 pixels, each 55×55 μm in size; each having its own circuitry. The overall measurement time was ∼3h per pattern to have very good statistic over the 2θ range of 5–70° with 0.0131° step size (2θ). The samples were spun at 16 rpm. - D8 ADVANCE DaVinci diffractometer (Bruker AXS), (250 mm of diameter) with Mo radiation equipped with a primary Johansson monochromator Ge (220), which gives a strictly monochromatic radiation (λ = 0.7093 Å), MoKα1. The X-ray tube worked at 50 kV and 45 mA. The optics configuration was a fixed divergence slit (2°) and a fixed diffracted anti-scatter slit (9°) and the energy-dispersive linear detector LYNXEYE XE 500 μm, specific for high energetic radiation, was used with the maximum opening angle. Using this conditions the samples were measured between 3-30º (2θ) with a step size of 0.009º and with a measurement total time of 2 hours and 30 minutes. The samples were spun at 10 rpm. Hugo Rietveld in the late sixties (Rietveld, 1967; 1969) devised “The Rietveld method” for the deeper characterisation of polycrystalline compounds. It consisted in the use of measured powder pattern intensities instead of reflection (peak) intensities, and supposed a conceptual breakthrough. This methodology, together with the coming of new technologies (computers), allowed to properly dealing with strongly overlapping reflections. Rietveld method is currently the most effective procedure for analysing powder diffraction data. This method can be used to carry out the QPA in crystalline samples. To obtain a successful RQPA, a properly prepared sample, a well aligned and maintained diffractometer and good structural descriptions for each crystalline phase are needed. Under these pre-requisites, a good powder diffraction pattern may be collected and RQPA can be carried out. Furthermore, every crystalline phase in the sample should be identified. This is easy to say but sometimes quite complex to fulfil because sometimes there are strong peak overlapping in the diffraction patterns and does not allow to conclusively determining all present phases. Finally, the RQPA needs to be to carry out with the appropriate software. In our case GSAS (Larson and Von Dreele, 2000; Toby, 2001) was the software package used. In addition to the raw data, any Rietveld program needs a control file to execute the refinements. In this control file, the crystal structures of the different components must be included. The fit is carried out by optimising all appropriate variables such as: i) scale factor of every crystalline phase; ii) background parameters for the chosen function; iii) unit cell parameters for every crystalline phase; iv) peak shape parameters for every Introduction 57 computed phase; and finally, v) correction parameters which may be phasedependent (such as preferred orientation, extinction, etc.) or pattern-dependent (zero-shift, absorption correction when working in transmission geometry, etc.). Usually, for RQPA, the structural descriptions (atomic positional parameters, atomic displacement parameters and occupation factors) are not optimised but kept as reported in bibliography. The application of RQPA to clinkers/cements/pastes is not an easy task for the following reasons (Aranda et al., 2012; Aranda et al., 2015): i) there are many phases, usually more than five, which increases the diffraction peak overlapping and so the correlations; ii) each phase has its own mass absorption coefficient which may yield the micro-absorption problem; iii) the small irradiated volume (∼2 mm3) for Cu Kα, which may lead to poor particle statistics; iv) some phases, for instance alite or gypsum, crystallise as plaques which show preferred orientation, increasing the errors; v) phases can crystallise as several polymorphs that must be identified a priori; vi) the diffraction peak broadening for some phases may be anisotropic and it must be properly modelled; and vii) the atomic impurities inside each phase are not known and their scale factors are usually computed for ideal/stoichiometric phases. In any case, RQPA shows several advantages over other methods based on powder diffraction and other technologies (microscopy, thermal analysis, etc.). As stated above, conventional RQPA requires all crystal structures to be known. Aranda et al (2012) reported the main hydrated and anhydrous cement standard phases. There are alternative whole-pattern quantitative phase analysis methods for crystalline phases with unknown structures (Smith et al., 1987; Taylor and Zhu, 1992; Scarlett and Madsen, 2006; Snellings et al., 2014). Currently, three ways to derive the phase content, Wα, from the Rietveld refined scale factor, Sα, can be used to determine the RQPA (Madsen et al., 2011; Gualtieri et al., 2014): i) Normalization to full crystalline phase content method. The simplest approach is the approximation that the sample is composed only of crystalline phases with known structures. This method normalises the sum of the analysed weight fractions to 1.0. Thus, if the sample contains amorphous phases, and/or some amounts of unaccounted crystalline phases, the analysed weight fractions will be overestimated. This approach is by far the most widely used method in RQPA. However, it must be highlighted that the resulting weight fractions are only accurate if the amount of unaccounted crystalline phases and amorphous content are very small (negligible) which may not be the case in anhydrous cements and for sure, is not the case in cement pastes. Chapter 1 64 the degree of hydration of the clinker phases, the phase composition and the microstructure of the hardened paste, which in turn determine its physical properties, including strength. 1.5.2.2. Shrinkage/Expansion properties (Length Changes). During the cement hydration, length changes can occur. The expansion (or shrinkage) experienced in mortars is mainly related to the nature of the pore structure, which affects the mobility of ions and the available space to accommodate new phases. Both adequate deformability and strength values are required to allow expansion without the formation of cracks. The dimensional stability of all the mortars prepared was measured with an analogic length comparator (mod. E077, MATEST) in standard prismatic samples (40× 40×160 mm) by applying the equation ΔL(%) = [(Lf − L0)/L0] × 100. In this equation, Lf is the measured length at a given time and L0 is the initial length (taken just after demolding and prior to immersion in water). The hydration of OPC cement causes a reduction in the absolute volume of the paste. In CSA and BCSA cement systems the expansive behaviour associated with ettringite formation has been widely studied (Mehta, 1973; Cohen, 1983a, b; Andac et al., 1999; Scherer, 1999; Scherer, 2004; Flatt and Scherer, 2008; Winnefeld and Lothenbach, 2010; Chen et al., 2012) but the mechanisms of expansion are still not fully understood. Calcium, sulphate or hydroxyl ions promote, when present in elevated amounts, the formation of fine ettringite crystals on the surface of C4A3S particles; in addition, they can combine, if their concentration is relatively low, with enough Al(OH)4ions to produce large ettringite crystals in the bulk solution. Various mechanisms implicated in external sulphate attack of cementitious materials were well reviewed by Brown and Taylor (1999). Most of their arguments have a general validity to systems in which expansion is related to ettringite formation (Bizzozero et al., 2014). The two head theories about the origin of expansion associated with ettringite are: the “crystal growth” and “swelling”. First one concludes that expansion is caused by the growth of ettringite crystals and the related crystallization pressure (Bentur and Ish-Shalom, 1974; Ogawa and Roy, 1981; Herrick et al., 1992; Deng and Tang, 1994); but according to the last theory (Mehta, 1973; Mehta and Hu, 1975; Mehta and Wang, 1982), the expansion is due to the adsorption of water molecules on ettringite, resulting in interparticle repulsion and swelling. 1.5.2.3. Setting. The setting and hardening of cement is a continuous process, but two points are distinguished for test purposes: Introduction 65 i) The initial setting time, which is the interval between the mixing of the cement with water and the time when the mix has lost plasticity, stiffening to a certain degree. It marks roughly the end of the period when the wet mix can be moulded into shape. It is determined by monitoring the repeated penetration of a needle into a fresh cement paste of standard consistence (using the Vicat apparatus), and the time of initial setting is taken when the distance between the needle and the base plate is of 6±3 mm according to UNE-EN 196-3. ii) The final setting time, which is the point at which the set cement has acquired a sufficient firmness to resist a certain defined pressure and according to UNE-EN 196-3 corresponds to a penetration of the needle of only 0.5 mm in the cement paste specimen. There are several parameters which affects initial and final setting times such as water/solid ratio, fineness, sulphate source, solubility and so on. For example Pelletier-Chaignat et al. (2012) measured the setting times of samples containing calcium sulphoaluminate clinker and gypsum combined with quartz filler or limestone filler. The results show that the use of limestone filler instead of quartz filler accelerates the early hydration of the cement, thus shortening the initial setting time. In this Thesis, paste setting times were determined using the Vicat test method according to UNE-EN 196-3, in VICATRONIC continuous penetration equipment (mod. E044N, MATEST). CHAPTER 2 OBJETIVES Objectives 69 2. OBJECTIVES The general aim of this PhD Thesis is to better understand the hydration mechanism of BCSAF cements to correlate it with the mechanical properties of the corresponding mortars. The specific objectives established in this work are described below: i) To scale up the synthesis of two iron-rich belite sulphoaluminate clinkers (BCSAF) in the laboratory at "medium scale" (∼2 kg) that contain different polymorphs of belite and ye'elimite. ii) To perform an accurate full mineralogical analysis of anhydrous BCSAF clinker and other related materials, including the quantification of amorphous and crystalline non-quantified (ACn) contents by LXRPD and Rietveld method. iii) To understand the hydration of BCSAF cement pastes at different curing ages through the quantification of the phase assemblage including ACn contents. To estimate the elemental composition of ACn phase(s) through LXRPD (indirectly) and SEM-EDS studies (directly). iv) To determine the effect of different parameters in the phase assemblage of the cement during hydration, such as the type of polymorphs present in the clinker composition, and the amount and type of sulphate source in the cement. v) To correlate the phase assemblage of BCSAF cement pastes with the mechanical properties, mainly compressive strength values and shrinkage/expansion of the corresponding mortars. CHAPTER 3 ARTICLES SECTION ARTICLE A#1 ARTICLE A#3 103 ARTICLE A#4 125 136 ARTICLE A#5 Chapter 4 174 4.1.2. BCSAF clinkers preparation: optimisation of process. Two BCSAF clinkers with the expected phase composition of 50 wt% of C2S, 30 wt% of C4A3S and 20 wt% of C4AF (Cuberos et al., 2010) were prepared. One of the clinkers was “actived” through the addition of borax, 2 wt% expressed as B2O3, to the raw material mixture. The aim of the activation has been obtaining clinkers with different belite (β-C2S or α'H-C2S) and ye'elimite (orthorhombic or pseudo-cubic) polymorphs to understand the effect of the polymorphs on the pastes hydration mechanism and mechanical performances. Hereafter, these clinkers will be named as BCSAF_B0 (non-active) and BCSAF_B2 (active), for boron-free and boron-containing clinker, respectively. Table 4.2 depicts dosage of raw materials used to prepare both clinkers and the corresponding elemental composition of the raw mixtures. The clinkering process was optimised to obtain about 2 kg of clinker. Table 4.2. Raw materials dosages in wt% and elemental composition of raw mixtures except for water or CO2 (expressed as oxide in wt%). BCSAF_B0 BCSAF_B2 Limestone 59.2 56.9 Bauxite 17.1 16.5 Kaolin 9.3 8.9 Gypsum 7.5 7.2 Marl 6.9 6.6 Borax - 3.9 CaO 52.7 51.3 SiO 2 17.5 17.0 Al 2 O 3 19.2 18.6 Fe 2 O 3 6.6 6.4 SO 3 4.0 3.8 B 2 O 3 - 2.0 Na 2 O - 0.9 In both cases, the raw mixture (∼3 kg) was pre-homogenized in plastic bags for a few minutes. Then, the sample was mixed for 15 min in a micro-Deval machine (A0655, Proeti S.A., Spain) at 100+5 rpm with steel balls (9 balls of 30 mm of diameter, 21 balls of 18 mm of diameter and balls of 10 mm of diameter up to a total ball weight of 2500+4 g). A small amount of grinding additive was used to prevent the sticking of the material to the balls. The resulting mixture was pressed into pellets of about 40 g (55 mm of diameter and ∼5 mm of height, see figure 4.1.a). Pellets were placed in a large Pt/Rh crucible of 325 cm3 of volume (figure 4.1.b). Results and discussion 175 (a) (b) Figure 4.1. (a) BCSAF raw mixture pellet; (b) Pt/Rh crucible use for clinkering. A two-step clinkering cycle was previously optimised for 5 g of BCSAF clinkers (dye-pressed samples with 10 mm of diameter) with different boron contents, (Cuberos et al., 2010). In that study, the best results were found when samples were heated at 900°C during 30 min (heating rate of 5 °C/min) and further heated at 1300°C for 15 min; all samples were quenched with air flow. The same clinkering process was followed to start our study but keeping the sample at 1300°C for 15 min resulted not enough for the scaling up; hence, the process had to be optimised. Table 4.3 gives RQPA for BCSAF_B0 clinker prepared under different thermal cycles. Two main parameters were controlled to determine both the optimum temperature and time of dwelling at high temperature: i) the targeted phase assemblage and ii) the minimum free lime content. It can be seen that 1350°C is the temperature with the lowest percentage of free lime (< 1 wt%) indicating the success of the clinkering process. In addition the content of nonhydraulically active phases, such as gehlenite (C2AS) and γ-C2S, is the lowest when heating the samples at this temperature. Figure 4.2 shows, as an example, a selected region of three patterns of BCSAF_B0 clinker after different heating cycles. From all these results, we concluded that 1350°C (and 30 min of dwelling) was the optimum clinkering cycle for our two scaled-up clinkers. Chapter 4 176 Table 4.3. Direct RQPA for BCSAF_B0 clinker prepared following different clinkering cycles. 1300°C 1325°C 1350°C 15 min 45 min 30 min 45 min 30 min β -C 2 S 41.5(4) 38.3(4) 46.7(3) 45.6(3) 46.8(3) o - C 4 A 3 S 16.2(2) 17.1(2) 13.6(6) 16.1(6) 13.8(5) c - C 4 A 3 S - - 13.0(6) 11.3(6) 12.8(5) C 4 AF 9.9(3) 10.1(3) 11.0(2) 11.7(3) 15.9(2) C 2 AS 12.2(2) 18.3(2) 6.4(2) 6.5(2) 4.9(2) C 3 A - - 4.6(3) 4.3(2) 3.3(2) γ -C 2 S 1.7(1) 2.2(2) 2.9(2) 3.1(2) 1.8(1) C S 2.9(1) 1.0(1) - - - CaO 15.6(1) 13.1(1) 1.8(1) 1.3(1) 0.8(1) Figure 4.2. Selected region of patterns of BCSAF_B0 clinker after heating at (a) 1300°C for 15 min , (b) 1325°C for 30 min, and (c) 1350°C for 30 min; and  highlights free lime (CaO). 15 20 25 30 35 40 45 50 (a) (b) (c) 2θ Results and discussion 177 Once the heating cycle was optimised, ∼3 kg of raw material was clinkered to obtain ∼2 kg of each clinker (BCSAF_B0 and BCSAF_B2). For that, six pellets (∼40 g each) were placed into the Pt/Rh crucible, heated (900°C/30 min and 1350°C/30 min with heating rates of 5 °C/min) and quenched with air flow. Since only ∼200 g of raw mixture can be clinkered at the same time, the process was repeated several times. The as-clinkered pellets were grinded in a micro-Deval mill at 100 rpm for 1 h in batches of ∼700 g; the obtained powder was sieved through a 250 μm mesh and characterised to verify the viability of the methodology. 4.1.3. Characterisation of the scaled-up BCSAF clinkers. Once the clinkering process was optimised, the two obtained clinkers were characterised through LXRPD, including the analysis of selective dissolutions, and SEM-EDS. 4.1.3.1. RQPA (normalised to 100% of crystalline phases). Both scaled-up BCSAF_B0 and BCSAF_B2 clinkers were studied by LXRPD (CuKα1 radiation) to identify, characterise and quantify the crystalline phases (by the Rietveld method), as described in A#1. Table 4.4 reports the direct RQPA results (wt%), normalised to 100% of crystalline phases, obtained for these two clinkers. Table 4.4. Direct RQPA results (wt%) for the two BCSAF clinkers normalised to 100% of crystalline phases. Numbers between brackets are the standard deviations of three independent measurements. Rietveld disagreement factors are also given. Phase BCSAF_B0 BCSAF_B2 α H ’-C 2 S - 57(2) βC2S 48.7(6) - γ-C 2 S 2.6(5) - o-C4A3S 15(1) - c-C4A3S 14(1) 31(2) C4AF 14.9(2) 10.1(6) C2AS 4.4(2) CT 1.3(2) 2.1(1) RWP/% 3.7 4.3 Chapter 4 178 Figure 4.3 shows a selected range of Rietveld plots for both clinkers, where the main phases are labelled. The final phase assemblage obtained for the two laboratory-scaled-up prepared clinkers confirms that BCSAF_B0 clinker contains β-C2S and both orthorhombic and pseudo-cubic ye'elimite as main phases, meanwhile α'HC2S and pseudo-cubic ye'elimite are stabilised in BCSAF_B2. 0.0 0.5 1.0 1.5 * * * * ** I (u.a.) α’H-C2S c-C4A3S C4AF * (b) 10.0 20.0 30.0 40.0 0.0 2θ 2θ θθ θ Lambda 1.5406 A, L-S cycle 216 Obsd. and Diff. Profiles 0.0 0.5 1.0 I (u.a.) β-C2S o-C4A3S c-C4A3S C2AS C4AF (a) 10.0 20.0 30.0 40.0 Figure 4.3. Selected range of Rietveld plots for (a) BCSAF_B0, and (b) BCSAF_B2 clinkers. Peaks due to a given phase are labelled. 4.1.3.2. Selective dissolutions. Selective dissolutions of BCSAF_B0 and BCSAF_B2 clinkers were carried out for a better characterisation (see A#1). Since main peaks of CT, C3A and merwinite, Results and discussion 179 Ca3Mg(SiO4)2, are overlapped, their identification/quantification by LXRPD and RQPA is a very rough task. Figures 4.4 and 4.5 show a small selected region of the Rietveld plots for the aluminate and silicate residues of BCSAF_B0 and BCSAF_B2 clinkers, respectively. It should be noted that the silicate residue is obtained when aluminate phases are removed. In similar way, the aluminate residue is achieved when the silicate phases were dissolved. (a) (b) β-C2S β-C2S β-C2S β-C2S γ-C2S C2AS o-C4A3S CT C2AS CT o-C4A3S C4AF o-C4A3S, C4AF Figure 4.4. Selected range of the Rietveld plots for: (a) BCSAF_B0 aluminate residue and (b) BCSAF_B0 silicate residue. In Table 4.5 is reported the direct RQPA results (wt%), normalised to 100% of crystalline phases, obtained for both aluminate and silicate residues of BCSAF_B0 and BCSAF_B2 clinkers. The results of this table show that aluminium containing phases have not been completely removed from the silicate residue (e.g. little amounts of C4A3S and C4AF is still present in BCSAF_B2 clinker). Moreover, it also can be Chapter 4 180 observed the existence of a small amount of C2AS in both aluminate and silicate residue of BCSAF_B2, that it was not quantified in the original clinker. Table 4.5. Direct RQPA results (wt%) for the aluminate and silicate residues of two BCSAF clinkers normalised to 100% of crystalline phases. Rietveld disagreement factors are also given. Phase Silicate residue Aluminate residue BCSAF_B0 BCSAF_B2 BCSAF_B0 BCSAF_B2 α’ H -C 2 S - 82.8(1) - - βC2S 78.1(1) - - - γ-C 2 S 4.9(2) - - - o-C4A3S 3.5(1) - 20.1(6) - c-C4A3S - 7.1(1) 26.8(5) 66.4(1) C4AF - 3.1(1) 33.9(3) 22.2(3) C2AS 10.6(2) 2.8(2) 15.6(3) 4.6(4) CT 2.8(1) 4.1(1) 3.7(2) 6.8(2) RWP/% 5.5 4.8 4.3 3.9 The Rietveld plot for the silicate residue of BCSAF_B0 is very informative as the diffraction peaks from C4AF disappear but the diffraction peak at ∼33.3° (2θ) is still present (arrows in Figure 4.4 and 4.5). Hence, this phase may be perovskite or merwinite but not C3A. The Rietveld refinements of the silicate residue indicated that the fit with perovskite was better (lower R-factors) than that with merwinite. A deep analysis of the Rietveld plots of the residues indicates that the peak widths in the BCSAF_B2 are narrower than those in BCSAF_B0. For instance, the diffraction peaks from CT and C4A3S in BCSAF_B2 aluminate fraction are narrower than those in the BCSAF_B0 aluminate fraction; see Figures 4.5.a and 4.4.a, respectively. This behaviour is likely due to a better particle growth when borax is added. In fact, scanning electron microscopy data (figure 4.6 in the next section) indicate that the average particle sizes for BCSAF_B2 are larger than those of BCSAF_B0. Results and discussion 181 (a) (b) C2AS CT c-C4A3S C4AF c-C4A3S , C4AF α’H-C2S C2AS C4AF α’H-C2Sα’H-C2S α’H-C2S CT ,α’H-C2S c-C4A3S, C4AF Figure 4.5. Selected range of the Rietveld plots for: (a) BCSAF_B2 aluminate residue and (b) BCSAF_B2 silicate residue. 4.1.3.3. SEM-EDS characterisation. Size, morphology and composition of particles of both clinkers were studied through SEM-EDS (in A#2). Figures 4.6.a and 4.6.b show SEM micrographs of the polished anhydrous BCSAF_B0 clinker at different magnifications. Figures 4.6.c and 4.6.d show the equivalent micrographs for the polished anhydrous BCSAF_B2 clinker. Figures 4.6.a and 4.6.c show ye'elimite angular shaped particles, as previously described (Pérez-Bravo et al., 2014). The identification of these particles was confirmed by EDS analysis, showing semi-quantitative average Al/Ca and S/Ca ratios of 1.4 and 0.10, respectively, which are relatively close to the theoretical values, 1.5 and 0.25, respectively. Moreover, this study revealed that the average particle size of ye'elimite in BCSAF_B0 was slightly smaller than that in BCSAF_B2; this is marked with arrows in Figures 4.6.a and 4.6.c. In addition, belite particle size was also studied Chapter 4 182 by SEM, being also smaller in BCSAF_B0 than in BCSAF_B2 clinker; in both cases these particles show a typical rounded shape. Figure 4.6. SEM micrographs of (a) and (b) anhydrous non-active-BCSAF clinker (BCSAF_B0), and (c) and (d) anhydrous active-BCSAF clinker (BCSAF_B2) (Figure 4 in A#2). 4.2. PHASE ANALYSIS OF DIFFERENT YE'ELIMITE-CONTAINING CLINKERS AND CEMENTS. In general, ye'elimite-containing cements are complex materials due to the presence of many crystalline phases, some of them also displaying polymorphism. LXRPD is the most appropriate technique to identify, characterise and quantify the crystalline phases within these samples. Therefore, if the mixture has an appreciable amount of amorphous and/or non-crystalline phases, this method may be considered as semi-quantitative. To overcome this problem, two approaches have been used in this PhD Thesis, the internal and the external standard methods (see Introduction section). In this section we discuss the most relevant results described in A#1 concerning the characterisation of these anhydrous materials through RQPA. The Results and discussion 183 research was aimed to: i) identify and quantify the crystalline phases present in several ye'elimite-containing clinkers and cements and ii) quantify their ACn content through two methodologies (external standard in reflection geometry and internal standard in transmission geometry). In order to do so, three commercial CSA clinkers (ALIPRE®, BELITH_CS10 and S.A.cement), one CSA cement (CSA_trial) and two laboratory-prepared BCSAF clinkers (BCSAF_B0 and BCSAF_B2) were analysed through LXRPD. 4.2.1. RQPA (normalized to 100% of crystalline phases). Firstly, direct RQPA of these samples, normalised to 100% of crystalline phases, was performed, and they are shown in Tables 4.4 and 4.6. These values were obtained from the approach described in section 1.5.1.1., and hence, the presence of an ACn fraction was not considered. Note that standard deviations in these tables are those derived from three independent measurements (they are not the mathematical errors from the Rietveld refinements). These three analyses were carried out to different portions of the samples for better averaging (i.e. not recording three patterns for the same sample). Figures 1 to 6, in A#1, show a selected range of the Rietveld plots for the six studied ye'elimite-containing materials, showing the good fit of the analysis. The major peaks for each phase are labelled. Several conclusions can be obtained from this analysis, as described below: Table 4.6. Direct RQPA results (wt%) for the ye'elimite-containing cements normalized to 100% of crystalline phases. Numbers between brackets are the standard deviations of three independent measurements. o-C4A3S c-C4A3S α αα α’H-C2S β ββ β-C2S C4AF CT M C5S2S CSH2 CS C3S ALIPRE®a 51.0(7) 18.5(6) 9.4(3) 7.7(1) 3.5(1) 0.52(2) 9.0(4) BELITHCS10b 40.1(9) 25.5(6) 16.0(2) 2.4(1) 9.3(1) 2.2(2) S.A.cementc 27.5(5) 28.7(6) 21.4(9) 9.7(4) 3.5(4) 1.1(1) 6.3(1) CSA_triald 17(1) 23.6(7) 9.0(9) 4.8(2) 16.2(5) 13.7(4) 8.5(2) 5.9(5) aAlso contains 0.5(1) wt% of Na2Si2O5. bAlso contains 4.6(1) wt% of akermanite. cAlso contains 1.9(1) wt% of CA. dAlso contains 1.8(7) wt% of dolomite. i) The good accuracy of the analyses was confirmed by comparing the amount of magnesium oxide obtained by RQPA and the obtained by XRF (Tables 2 and 5 in A#1). Magnesium oxide contents were chosen because magnesium is little soluble in the ye'elimite or belite structures. For example, RQPA showed the highest amount of periclase (MgO) for BELITH_CS10, 2.2(2) wt%, and this is in full agreement with elemental analysis determined by XRF. Furthermore, S.A.cement (the second Chapter 4 190 The second most important difference between both hydration behaviours takes place after 1 day of hydration. On the one hand, in G10B0, the dissolution of βC2S and C4AF starts after 24 h of hydration, with the consequent crystallisation of layered AFm type phases, such as stratlingite (see Figure 4.8.a). On the other hand, for G10B2, α'H-C2S percentage remains constant up to 51 h of hydration and C4AF dissolves very slowly after 14 h. The difference in reactivity of both belite polymorphs is astonishing, i.e. β-C2S reacts faster than α'H-C2S. This behaviour is in disagreement with the general accepted idea in the cement field: α-forms of belite are reported to have faster hydration kinetics than β-forms (Jelenic et al., 1978; Chatterjee, 1996). However, under our studied experimental conditions, β-C2S reacts faster than α'HC2S, to yield stratlingite, see equation [1.4] (introduction section). This behaviour may well be justified with the formation of high amounts of ettringite at early hours which implies a concomitant large quantity of amorphous aluminium hydroxide. The availability of amorphous AH3 promotes the precipitation of stratlingite, C2ASH8, from belite reaction (equation [1.4]). In conclusion, the hydration behaviour of C2S is likely more dependent on the chemical environment than on its polymorphism. 4.3.1.2. Ex-situ hydration behaviour at late ages (> 24 h), with different amounts of gypsum. The influence of the amount of gypsum in BCSAF cements at late ages of hydration (> 24 h) was determined through LXRPD and published in A#3. Three different amounts of gypsum were tested: 5, 10 and 15 wt%. The hydrated pastes, after stopping hydration, were characterised by LXRPD and Rietveld methodology at 3, 7, 28 and 120 days. The G-factor approach was used to obtain a full mineralogical phase assemblage including ACn contents. The FW content was also determined by the difference between the added water and the combined water determined by DTA-TGA. Figure 4.9 shows the degree of reaction of belite polymorphs, orthorhombic and pseudo-cubic ye'elimites, and ferrite as a function of hydration time and amount of added gypsum. Several conclusions can be drawn of this study: - Effect of the activation of clinkers on hydration. The most surprising result of the previous study at early ages was that β-belite in non-active clinker reacts at a higher pace than α′H-belite in BCSAF_B2. That behaviour has been corroborated in this study at later ages, Figure 4.9.a. Thus, the reactivity of β-C2S seems to be enhanced by an oversaturation of aluminate ions at early hours to yield stratlingite (Martín-Sedeño et al., 2010; Winnefeld and Lothenbach, 2010;). Moreover, ye'elimite reacts at a different pace for BCSA_B0 and BCSA_B2. Orthorhombic ye'elimite, in non-active cements, is completely hydrated after 3 days of hydration in GgB0, while pseudo-cubic ye'elimite in GgB2 reacts at slightly slower pace for the same gypsum Results and discussion 191 content and age of hydration, Figure 4.9.b. This effect was previously observed in the hydration study at early hours (A#2). G5B0 G10B0 G15B0 G5B2 G10B2 G15B2 0 10 20 30 100 120 0 10 20 30 40 50 60 70 80 90 Degree of reaction (%) (a) C2S 0 10 20 30 100 120 0 10 85 90 95 100 Degree of reaction (%) (b) C4A3S 0 10 20 30 100 120 0 10 20 30 40 50 60 70 80 90 100 Degree of reaction (%) (c) C4AF t (d) Figure 4.9. Hydration degree, α (%), for GgB0 and GgB2 pastes as a function of curing time and amount of gypsum added of (a) C2S; (b) C4A3S; and (c) C4AF. Solid lines are just guide3 to the eyes. Chapter 4 192 It must be underlined that although the phases in BCSAF_B0 react faster than in BCSAF_B2, the non-active cement develop much smaller mechanical strength, which clearly indicate the importance of the hydration chemistry/environment, as discussed previously. - Effect of the amount of gypsum during hydration. Ye'elimite reaction kinetics show a small dependence on the amount of added gypsum, as there is a slight increase in hydration rate by increasing the gypsum content. This behaviour seems to be slightly more marked in pseudo-cubic ye'elimite. The final reaction degree of both polymorphs of dicalcium silicate is more affected by the addition of gypsum. On the one hand, β-C2S reactivity (α) was enhanced by increasing the gypsum content (α rises from 65% to 75% by the addition from 5 to 15 wt% of gypsum). On the other hand, α′H-C2S reaction degree decreased from 62% to 42% for addition of 5 to 15 wt% of gypsum, respectively, Figure 4.9.a. The hydration of the ferrite phase was strongly retarded by increasing the gypsum content in both GgB0 and GgB2 cements, Figure 4.9.c, in agreement with previous studies (Wang, 2010). - Concerning the hydration products. The main crystalline hydrated compounds were ettringite, stratlingite and katoite. Figure 4.10 shows, as a representative example, a selected range of Rietveld plots for two samples, G15B0 and G15B2, hydrated for 120 days. The main peaks of each phase are labelled. The amount of crystallised ettringite in GgB2 cements is higher than that in GgB0 cements, irrespective of gypsum content (see Tables 1 and 2 in A#3). Moreover, the crystallisation process of stratlingite is strongly affected by the amount of added gypsum; in fact, the amount of stratlingite decreases by increasing the gypsum content, Tables 1 and 2 in A#3. The crystallisation of stratlingite was confirmed by DTA measurements. DTA curves for G15B0, at all the studied ages, were plotted (see Figure 3 in A#3). The endothermic signal at ∼235°C, corresponding to the dehydration of AH3, is present at 3 and 7 days. After 28 and 120 days, this signal disappears and two endothermic peaks related to stratlingite precipitation at 160−175°C and ∼200°C appear (arrows in figure 3 in A#3), thus confirming stratlingite formation by consumption of AH3, according to reaction 1.4 given in the introduction. Results and discussion 193 10.0 15.0 20.0 25.0 30.0 35.0 -2.0 0.0 2.0 4.0 * 10.0 15.0 20.0 25.0 30.0 35.0 -2.0 0.0 2.0 4.0 * (a) Ettringite Stratlingite Katoite β-C2S CC γ-C2S*C2AS Ettringite Stratlingite Katoite α’H-C2S CC *C2AS (b) 2θ θθ θ Figure 4.10. Selected range of Rietveld plots for two pastes, (a) G15B0 and (b) G15B2, cured for 120 days. In addition, the amount of precipitated crystalline katoite seems to depend on C4AF hydration, as significant amounts of this siliceous-hydrogarnet phase crystallises just after the dissolution of C4AF (Tables 1 and 2 in A#3). This supports the hypothesis that katoite might contain a significant amount of iron as a substitute for aluminium. Consequently, the katoite solid solution, also known as hydrogarnet, may be expressed as C3A1−xFxSH4. The x values were determined by unit cell parameter inspection. Refined a unit cell parameters obtained in cubic katoites, for all cement pastes, were compiled. Then, according to the equation a = 0.16x + 12.29, obtained by assuming a linear variation of the unit cell (Taylor, 1997) the Al/Fe ratio in C3A1−xFxSH4 can be estimated. Figure 4.11 shows the theoretical values of a as a function of x (dotted line), and the corresponding values obtained for the different cement pastes. The average value of the refined unit cell parameters at different hydration ages was used in the cement pastes for which the standard deviation (σn−1) Chapter 4 194 was lower than 0.009. When σn−1 was higher, the average was not calculated. For G10B2 and G15B2 pastes two values of refined unit cell (a) were calculated. One at early hydration ages (3 and 7 days), marked with an asterisk in figure 4.11, and other at later ages of hydration (28 and 120 days). 1.228 1.230 1.232 1.234 1.236 1.238 1.240 1.242 1.244 1.246 0 0.2 0.4 0.6 0.8 1 x (moles Fe2O3in formula) a (nm) G5B0 G5B2 G10B0 G10B2 G15B0 G15B2 * * Figure 4.11. Values of the cell parameter (a) for katoite vs values of x (moles of Fe2O3 in formula C3A1-xFxSH4) for different GgBx pastes. (Dotted line stands for theoretical dependence of a as function of x). 4.3.1.3. Elemental composition of ACn. It is not only important to quantify the amorphous content, also try to characterise and estimate its elemental composition to correlate it with the cement hydration behaviour and mechanical properties. As it was discussed in the introduction, it is not possible to determine the chemical composition of ACn directly by LXRPD. An attempt to find out the possible composition of the ACn (sulphate, silicate, aluminate and iron-bearing groups) was performed through RQPA combined with G-factor, which was published in A#3. For this purpose, the evolution of different ions-containing groups was studied with time: i) Sulphate groups. The amount of reacted/dissolved sulphate was determined from the disappearance of ye'elimite and gypsum; and the amount of crystallised sulphate was calculated from the formation of ettringite. Figure 4.12 shows the evolution of sulphate content with time for G10B0 and G10B2 pastes as representative examples. The dotted line represents the maximum sulphate that can be hydrated and/or crystallised. As mentioned before, we can observe that the Results and discussion 195 crystallisation of AFt in BCSA_B2 is higher than that in BCSA_B0 pastes: ∼76 wt% of the hydrated sulphates had crystallised for G10B2 against ∼57 wt% for G10B0, meaning that, in this case, ∼43 wt% of dissolved, but not crystallised, sulphate groups (Figure 4.12) were mainly incorporated into ACn phase(s) and/or in pore solution (Winnefeld and Lothenbach, 2010). 0 20 40 60 80 100 120 0 1 2 3 4 5 6 7 SO24 from remaining AFt SO24 from crystallized ye'elimite and gypsum Sulphate group (wt%) t (days) ~43% (a) G10B0 0 20 40 60 80 100 120 0 1 2 3 4 5 6 7 SO24 from remainig AFt SO24 from crystallized ye'elimite and gypsum Sulphate group (wt%) t (days) ~24% (b) G10B2 Figure 4.12. Hydration and crystallization rate of sulphate groups for (a) G10B0 and (b) G10B2 pastes. Dashed lines represent the maximum values of dissolved sulphate group. ii) Silicate groups. The evolution of silicate contents (both dissolved and crystallised) in G10B0 and G10B2 pastes (as representative examples) during hydration is shown in Figure 4.13. On the one hand, hydrated silicate contents stand Chapter 4 196 for the amount of silicon-bearing phases that have reacted with water (calculated from belite consumption, Tables 1 and 2 in A#3). On the other hand, crystallised silicate contents mean the amount of silicon-containing crystalline hydrated phases that have precipitated and been quantified, that is, stratlingite and katoite (from Tables 1 and 2 in A#3). Dotted lines in Figure 4.13 stand for the maximum hydrated silicate content for G10Bx cement pastes. The amount of crystallised silicate is higher in G10B0 than in G10B2, where higher amounts of stratlingite were found. Thus, a higher amount of hydrated silicate remains in the amorphous phase(s) for G10B2 cement paste. 0 20 40 60 80 100 120 0 2 4 6 8 10 12 14 SiO44 from β-C2S remainig SiO44 from crystallized Katoite SiO44 from crystallized Stratlingite Total SiO44 crystallized Silicate group (wt%) t (days) ~50% (b) G10B2 0 20 40 60 80 100 120 0 2 4 6 8 10 12 14 SiO44 from β - C2S remaining SiO44 from crystallized Katoite SiO44 from crystallized Stratlingite Total SiO44 crystallized Silicate group (wt%) t (days) ~7% (a) G10B0 Figure 4.13. Hydration and crystallisation rate of silicate groups for (a) G10B0 and (b) G10B2 pastes. Dashed lines represent the maximum values of dissolved silicate group. Results and discussion 197 iii) Aluminate groups. The evolution of aluminium was also studied through the same methodology. Hydrated aluminate contents were obtained from C4A3S and C4AF consumption, and crystallised aluminate amounts were calculated from the precipitated AFt, katoite, and stratlingite. Evolution of aluminate with hydration time, for G10Bx, G5Bx and G15Bx pastes, is given in Figures 4, S4 and S5 (Supporting Information) of A#3. The amount of crystallised aluminium-bearing phases was higher in GgB0 than in GgB2 cement pastes, matching in some cases the maximum, whereas more than 30 wt% of the aluminate content remained in the ACn phase(s) for GgB2 cement pastes, and/or to a minor extent in pore solution. - NMR study: A NMR study was also performed on these samples (G10Bx) during my research stage at Instituto de Ciencias de los Materiales (ICMM-CSIC) in Madrid. Here, the complex heterogeneous molecular compositions and structures of anhydrous and hydrated cements have been followed by solid-state NMR. Onedimensional (1D) single-pulse NMR measurements are generally quantitative, allowing resolved species to be evaluated and their relative populations monitored over time, as the cement hydrates and solidifies. 1D single-pulse 29Si and 27Al NMR have been used to characterise the species in anhydrous and hydrated G10Bx cement. It has been established that the primary silicate constituents of cements are so-called Q0 29Si silicate species that compose dicalcium silicate (C2S) phase. (“Qn” refers to 29Si atoms that are covalently bonded via bridging oxygen atoms to n < 4 other Si atoms; “Qn(mAl)” similarly refers to a 29Si atom bonded via bridging oxygen atoms to n other silicon or aluminium atoms, m of which are Al). During the hydration process, more condensed silicate species form disordered products, such as calcium-silicate hydrates or amorphous phases, are characterised by Q1 and Q2 29Si moieties with broad distributions of local environments that are typically not resolved by single-pulse 29Si MAS NMR or other methods. In cements hydrated for longer times, Q3 and Q4 29Si silicate species are also prevalent, which indicates a more densely cross-linked silicate network, as expected. 29Si NMR MAS spectra for G10B0 and G10B2 (anhydrous and hydrated) are shown in Figures 4.14 and 4.15, respectively. Similarly, solid-state 1D 27Al NMR measurements show that aluminate species in anhydrous cements have principally 4-fold coordination (AlIV). The initial anhydrous cement phases, containing aluminate, are C4A3S and C4AF, and they are composed principally of AlIV species. The 27Al NMR signals are often extensively broadened by ferroor paramagnetic iron components in close proximity. The iron components can be present in crystalline or disordered regions as cations, oxides, and/or hydroxides (e.g., Fe2+, Fe3+, Fe2O3, Fe3O4, Fe(OH)3, referred to broadly as “ferrites” within the cement literature) with indefinite compositions and distributions. During hydration, AlIV species are thought to react with the calcium Chapter 4 198 silicate species and be partially incorporated into the amorphous phase, with some AlIV species replacing four-coordinated 29Si atoms. For these reasons, the quantification of the intensities in all the 27Al MAS NMR spectra becomes very difficult. (a) Anh (b) 7d (c) 28d (d) 180d β-C2S γ-C2S Gehl Stratl β-C2S γ-C2S Gehl Stratl β-C2S γ-C2S Gehl Stratl β-C2S γ-C2S Gehl δ δδ δ(ppm) -65 -70 -75 -80 -85 -90 Figure 4.14. 29 Si NMR MAS spectra for G10B0 cement, (a) anhydrous, and hydrated at different curing times: (b) 7 days, (c) 28 days, and (d) 180 days. The 29 Si signals, that are clearly resolved at -70.3, -71.4 and -73.5 ppm in Figures 4.14 and 4.15, are assigned to α, β and γ belite polymorphs, respectively. Other 29 Si signal was observed in two figures (Figures 4.14 and 4.15) at -72.5 ppm Results and discussion 199 related to the gehlenite phase. Hydrated cements are substantially more complicated than the anhydrous starting materials due to the presence of species with local environments witout long-range order which exacerbates the 29Si signals overlapping. 29Si signals intensity over the range -67 to -76 ppm, decrease slightly with the hydration time, which indicates that the anhydrous silicate species have reacted to form hydration products. Additionally, at early hydration time (7 days) of G10B0, broad upfield 29Si signal is observed at aprox. -86.5 ppm, (Figure 4.14.b), which is attributed to Q2(1Al) silicate species. This signal indicates that the stratlingite is formed in this paste (Santacruz et al., 2015) and it increases with the hydration time (Figures 4.14.c and d). Moreover the 29Si signal intensity over -80 ppm increases with the hydration time, indicating that a condensed Q2(2Al) silicate species are being formed, probably in the amorphous phase. In G10B2 the 29Si signal corresponding to the formation of stratlingite is not observed until 28 days of hydration (Figure 4.15.c). In general the 29Si signal for G10B2 is wider than the corresponding for G10B0, likely due to the higher amount of ACn content present in the G10B2. These results seem to be in accordance with that obtained in previous studies about the evolution of silicate groups in G10B0 and G10B2 samples. The quantification of 29Si NMR measurements is possible, but it is not an easy task. The lack of information about the 29Si signals corresponding to different hydrated cement phases and the presence of an amorphous fraction, made difficult to perform an accurate quantification.