Preparation and hydration of model ecocement phases. Characterization by diffraction and cognate methods

FACULTAD DE CIENCIAS DEPARTAMENTO DE QUÍMICA INORGÁNICA, CRISTALOGRAFÍA Y MINERALOGÍA TESIS DOCTORAL “Preparation and hydration of model ecocement phases. Characterization by diffraction and cognate methods” Ana María Cuesta García Málaga, 2015 AUTOR: Ana María Cuesta García http://orcid.org/0000-0002-8634-2241 EDITA: Publicaciones y Divulgación Científica. Universidad de Málaga Esta obra está bajo una licencia de Creative Commons Reconocimiento-NoComercialSinObraDerivada 4.0 Internacional: http://creativecommons.org/licenses/by-nc-nd/4.0/legalcode 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 Preparation and hydration of model ecocement phases. Characterization by diffraction and cognate methods MEMORIA presentada por la Licenciada en Química Dª Ana María Cuesta García para aspirar al grado de Doctora en Ciencias, Sección de Químicas, con la mención de “Doctorado Internacional“ Fdo.: Ana María Cuesta García 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.: Dr. Enrique Ramírez Losilla Profesor Titular de Universidad D. MIGUEL ÁNGEL GARCÍA ARANDA, Catedrático de la Universidad de Málaga, Dª Mª ÁNGELES GÓMEZ DE LA TORRE, Profesora Titular de la Universidad de Málaga, y D. ENRIQUE RAMIREZ LOSILLA, Profesor Titular de Universidad, 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 Ana María Cuesta García, titulada: “Preparation and hydration of model ecocement phases. Characterization by diffraction and cognate methods”, 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 28 de Julio de 2015 Fdo.: Dr. Miguel Ángel García Aranda Catedrático de Universidad Fdo.: Dra. Mª Ángeles Gómez de la Torre Profesora Titular de Universidad Fdo.: Dr. Enrique Ramírez Losilla Profesor Titular de Universidad D. PEDRO JESÚS MAIRELES TORRES, Catedrático de la Universidad y Director del Departamento de Química Inorgánica, Cristalografía y Mineralogía de la Facultad de Ciencias de la Universidad de Málaga. Informa: Que la presente memoria realizada por Dª Ana María Cuesta García, titulada: “Preparation and hydration of model ecocement phases. Characterization by diffraction and cognate methods”, ha sido realizada bajo la dirección del Catedrático D. Miguel Ángel García Aranda, la Profesora Titular Dª Mª Ángeles Gómez De la Torre y el profesor Titular D. Enrique Ramírez Losilla 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 28 de Julio de 2015. Fdo.: D. Pedro Jesús Maireles Torres. A mi madre. Table of contents Nomenclature ..................................................................................................................... 1 Chemical reactions ............................................................................................................. 5 Abstract .............................................................................................................................. 9 Resumen ........................................................................................................................... 21 1. Introduction to cements ................................................................................................. 35 1.1. Ye'elimite-containing cements .............................................................................. 35 1.2. Pure phases of interest in calcium sulfoaluminate cements .................................... 41 1.2.1. Dicalcium silicate (Ca2SiO4, C2S) ................................................................ 41 1.2.1.1. Dicalcium silicate chemical compositions and structures ................... 42 1.2.1.2. Hydration mechanisms of dicalcium silicate ...................................... 48 1.2.2. Ye'elimite (Ca4[Al6O12]SO4, C4A3S)............................................................. 50 1.2.2.1. Ye'elimite chemical compositions and structures ............................... 51 1.2.2.2. Ye'elimite hydration mechanisms ...................................................... 54 1.2.3. Tetracalcium aluminoferrite (Ca4Al2Fe2O10, C4AF) ..................................... 56 1.2.3.1. C4AF chemical compositions and structures ...................................... 56 1.2.3.2. C4AF hydration mechanisms ............................................................. 58 1.2.4. Ettringite or AFt (Ca6Al2(OH)12(SO4)3·26H2O C6AS3H32) ........................... 60 1.2.5. AFm-type phases ......................................................................................... 62 1.2.6. Other hydrated phases: hydrogarnet phases and amorphous aluminium hydroxide Al(OH)3·nH2O ...................................................................................... 65 1.3. Methodologies ...................................................................................................... 69 1.3.1. Laboratory X-Ray Powder Diffraction (LXRPD) and Rietveld Method ........ 69 1.3.2. Synchrotron X-Ray Powder Diffraction (SXRPD) (BL04 – MSPD, ALBA) 73 1.3.3. Isothermal conduction calorimetry ............................................................... 74 1.3.4. Thermal measurements ................................................................................ 76 1.3.5. Scanning electron microscopy (SEM) .......................................................... 76 2. Objectives..................................................................................................................... 81 3. Articles section ....................................................................................... (not numerated) a#1: Mechanism of stabilization of dicalcium silicate solid solution with aluminium a#2: Reactive belite stabilization mechanisms by boron-bearing dopants a#3: Structure, atomistic simulations, and phase transition of stoichiometric yeelimite a#4: Pseudocubic crystal structure and phase transition in doped ye'elimite a#5: Hydration mechanisms of two polymorphs of synthetic ye'elimite a#6: Ye'elimite: structures and hydration mechanisms a#7: Hydration of C4AF in the presence of other phases: a synchrotron X-ray powder diffraction study 4. General results and discussion ...................................................................................... 85 4.1. Dicalcium silicate ................................................................................................. 85 4.1.1. Synthesis and sample characterization of dicalcium silicate.......................... 85 4.1.2. Dicalcium silicate structural study ................................................................ 91 4.2. Ye'elimite ............................................................................................................. 93 4.2.1. Ye'elimite synthesis...................................................................................... 93 4.2.2. Ye'elimite structural study ............................................................................ 95 4.2.3. Thermodiffractometric study and phase transition ........................................ 99 4.2.4. HT-polymorphs of ye'elimite: structural study ............................................ 102 4.3. Synthesis of C4AF .............................................................................................. 103 4.4. Application of the structural studies to the analysis of CSA and BCSA cements 105 4.5. Hydration studies................................................................................................ 107 4.5.1. Hydration of ye'elimite ............................................................................... 109 4.5.2. Reactivity of ye'elimite with gypsum .......................................................... 115 4.5.3. Reactivity of ye'elimite with C4AF ............................................................. 118 4.5.4. Reactivity of ye'elimite with dicalcium silicate ........................................... 124 4.5.5. Hydration of C4AF ..................................................................................... 129 4.5.6. Reactivity of C4AF with gypsum ................................................................ 133 4.5.7. Influence of ye'elimite in C4AF hydration .................................................. 135 4.6. On-going research .............................................................................................. 136 5. Conclusions ................................................................................................................ 145 5. Conclusiones .............................................................................................................. 151 6. References .................................................................................................................. 157 Annex I: Collaborations .................................................................................................. 183 Annex II: Copyright permissions .................................................................................... 187 1 Nomenclature In order to simplify the writing and presentation of chemical formulations, cement nomenclature will be used hereafter: C = CaO S = SO3 H = H2O S = SiO2 N = N2O C = CO2 A = Al2O3 K= K2O P = P2O5 F = Fe2O3 M = MgO T = TiO2 B = B2O3 The compounds are written as follows: Formula Oxides Cement nomenclature Name 2Al(OH)3·nH2O Al2O3·(3+n)H2O AH3·nH Amorphous aluminium hydroxide CaSO4 CaO·SO3 CS Anhydrite CaSO4·0.5H2O CaO·SO3·0.5H2O CSH0.5 Bassanite 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 Ca2SiO4 2CaO·SiO2 C2S Dicalcium silicate/ Belite Ca6Al2(OH)12(SO4)3·26H2O 6CaO·Al2O3·3SO3·32H2O C6AS3H32 Ettringite (AFt) Ca3Al2(SiO4)3 3CaO·Al2O3·3SiO2 C3AS3 Garnet Ca2Al2SiO7 2CaO·Al2O3·SiO2 C2AS Gehlenite 2Al(OH)3 Al2O3·3H2O AH3 Gibbsite CaSO4·2H2O CaO·SO3·2H2O CSH2 Gypsum 2 Formula Oxides Cement nomenclature Name Ca3Al2(OH)12 3CaO·Al2O3·6H2O C3AH6 Hydrogarnet or katoite 2Fe(OH)3 Fe2O3·3H2O FH3 Iron hydroxide Ca3(Al0.5Fe0.5)2(SiO4)(OH)8 3CaO·0.5Al2O3·0.5Fe2O3 ·SiO2·4H2O C3A0.5F0.5SH4 Iron (siliceous) hydrogarnet or iron-katoite Ca12Al14O33 12CaO·7Al2O3 C12A7 Mayenite Ca4Al2(OH)12(SO4)·6H2O 4CaO·Al2O3·SO3·12H2O C4ASH12 Monosulfate (AFm) Ca(OH)2 CaO·H2O CH Portlandite Ca3Al2(SiO4)(OH)8 3CaO·Al2O3· SiO2·4H2O C3ASH4 Silicious hygrogarnet Ca2Al(OH)6[AlSiO2(OH)4·3H2O] 2CaO·Al2O3·SiO2·8H2O C2ASH8 Stratlingite Ca5(SiO4)2(SO4) 5CaO ·2SiO2·SO3 C5S2S Ternesite Ca4Al2Fe2O10 4CaO·Al2O3·Fe2O3 C4AF Tetracalcium aluminoferrite Ca3Al2O6 3CaO·Al2O3 C3A Tricalcium aluminate Ca3SiO5 3CaO·SiO2 C3S Tricalcium silicate/ Alite Ca4Al6O12(SO4) 4CaO·3Al2O3·SO3 C4A3S Ye'elimite Abbreviated names or initials to refer to some terms are listed below in alphabetical order:  a#x: article number x  ACn: Amorphous and Crystalline non-quantified  ADP: Atomic Displacement Parameter  AFm: Al2O3-Fe2O3-mono 3  AFt: Al2O3-Fe2O3-tri  ATR-FTIR: Attenuated Total Reflectance-Fourier Transform Infrarred Spectroscopy  BCSA: Belite Calcium Sulfo-Aluminate  BCSAF: Iron-rich Belite Calcium Sulfo-Aluminate  BSE: BackScattered Electrons  B.M. EoS: Birch-Murnaghan Equation of State  CAC: Calcium Aluminate Cement  CSA: Calcium Sulfo-Aluminate  DAC: Diamond Anvil Cell  DSC: Differential Scanning Calorimetry  DTA: Differential Thermal Analysis  FW: Free Water  FWHM: Full Width at Half Maximum  HT-LXRPD: High-Temperature Laboratory X-Ray Powder Diffraction  LXRPD: Laboratory X-Ray Powder Diffraction  NPD: Neutron Powder Diffraction  OPC: Ordinary Portland Cement  PXCT: Ptychographic X-ray Computed Tomography  RQPA: Rietveld Quantitative Phase Analysis  RT: Room Temperature  SEM: Scanning Electron Microscopy  SXRPD: Synchrotron X-Ray Powder Diffraction  TGA: ThermoGravimetric Analysis  w/s: water to solid ratio  XRPD: X-Ray Powder Diffraction Ana María Cuesta García 10 polymorphs of dicalcium silicate, plays an important role in the preparation of active BCSA cements in the laboratory and industrial trials. Ca2SiO4 or dicalcium silicate (belite in cement nomenclature) presents five stoichiometric forms , , α'L, α'H and α. The low temperature polymorph is γ-C2S, however, this phase is avoided in cement manufacture because it is hydraulically non-active. The -form is usually found in OPC and in almost all CSA cements. In addition, α-phases are supposed to be more reactive and they are mainly found in some active BCSA cements. The stabilization of α'-polymorph by introduction of foreign oxides has been studied over the years. Recently, it has been reported that the addition of a combination of B2O3 and Na2O successes to stabilize α'H-C2S (Wesselsky and Jensen 2009). Dicalcium silicate hydration may yield C-S-H and portlandite or stratlingite in case an aluminum source is present and portlandite is absent. However, the hydration kinetic of dicalcium silicate is slow when compared to other phases. Ye'elimite, Ca4[Al6O12]SO4, present a sodalite type-structure with general composition, M4[T6O12]X. Nevertheless, stoichiometric ye'elimite crystal structure at room temperature was poorly understood since it has been reported to be orthorhombic, tetragonal or cubic. Furthermore, it was know that the presence of different amounts of Na+ and Fe3+ can stabilize the cubic form. On the other hand, ye'elimite is very reactive and most of its hydration heat is released during the first eight hours. During early age hydration, ettringite (AFt) phase is the main crystalline hydration product formed from the dissolution of ye'elimite in the presence of calcium sulfate(s). However, monosulfate (AFm) phase is mainly formed in the absence of a source of soluble sulfate. A similar hydration mechanism can be found in the early age hydration of CSA and BCSA cements because their hydration procedure is mainly dominated by the ye'elimite phase. For this reason, commercial cements with large amounts of ye'elimite have special applications such as high strength developments at early-ages. Abstract 11 Tetracalcium aluminoferrite, is the major iron-containing phase in OPC and is also present in BCSA. However, the reactivity of tetracalcium aluminoferrite during hydration in these cements is slower than that for ye'elimite and it is not well understood. In the hydration of tetracalcium aluminoferrite usually C3(A,F)H6 is formed and in the presence of sulfate AFt and/or AFm precipitate. Moreover, in the presence of silica, iron containing siliceous hydrogarnet will precipitate instead of C3(A,F)H6. Taken into account which has been discussed previously, the general aim of this PhD Thesis has been to synthesize and characterize these single phases of ecocements to better understand their behavior during hydration. To do so, this PhD work is divided in two main blocks: i) structural studies of anhydrous phases and ii) hydration studies of selected phases. SXRPD, high-temperature LXRPD, NPD, DSC and permittivity measurements have been employed for the structural studies. On the other hand, XRPD is a powerful tool for material characterization in general, and for in-situ studies of hydration processes in particular. This technique has been extensively used for the characterization of pastes. Moreover, the use of an intense X-ray source, i.e. synchrotron X-rays, coupled with fast Xray detectors permits time-resolved diffraction experiments allowing in-situ quantitative phase analysis during the early ages of cement hydration. Firstly, for the study of β-dicalcium silicate samples, Ca2Si1-2xAl2xO4-xx (from x=0 to 0.03) solid solution was prepared and studied by LXRPD and the Rietveld method. The sample Ca2Si0.972Al0.028O3.9860.014 which contained 99(1) wt% of β-polymorph was selected for further studies. Thermal analysis measurements of this sample in a wet atmosphere indirectly have confirmed that aluminum was incorporated into the framework of dicalcium silicate and stabilized the β-form by replacing the silicate units and generating oxygen vacancies. However, aluminum substitution at both calcium and silicon sites, Ana María Cuesta García 12 Ca2-xAlx(SiO4)1-x(AlO4)x, was not achieved, conversely to the aluminium doping in tricalcium oxysilicate. Then, the coupled Na/B-doping of dicalcium silicate has also been investigated and Ca2-xNax(SiO4)1-x(BO3)x series have confirmed to exist for a large range of x values being α'H-C2S the main phase (for x0.10). In this case, the borate group was planar-triangular, BO33-. Moreover, Ca2-xBx(SiO4)1-x(BO4)x was also studied and in this case tetrahedral SiO44groups were replaced by tetrahedral BO45units. This solid solution also allowed to stabilize high amounts of α'Hform. However, this series was not further studied as we were more interesting in the stabilization of α'H-C2S by borax for mimic the industrial conditions. The boron anion nature was determined by infrared studies. Other compositions Ca2-x/2x/2(SiO4)1-x(BO3)x, Ca2(SiO4)1-x(BO3)xOx/2 and Ca2-3xB2xNax(SiO4)1-x(BO4)x were also tested but they were not achieved. Then, α'HCa1.85Na0.15(SiO4)0.85(BO3)0.15 was selected to perform a deep structural study, this was due to the fact that the sample presented sharp diffraction peaks and a high content of α'H-phase (90 wt%). Finally, a new revised structural description for α'H-Ca1.85Na0.15(SiO4)0.85(BO3)0.15 sample was obtained. The final refined unit cell parameters for α'H-Ca1.85Na0.15(SiO4)0.85(BO3)0.15 were a=6.8432(2) Å, b=5.4555(1) Å, c=9.2346(2) Å and V=344.76(2) Å3. The new revised crystal structure yielded to low disagreement values, RWP=6.5% and RF(α'H-C2S)=2.4% and fitted better this phase in active BCSA cements allowing more accurate Rietveld mineralogical analysis. For ye'elimite, stoichiometric (Ca4Al6O12SO4) and doped (Ca3.8Na0.2Al5.6Fe0.2Si0.2O12SO4) samples have been prepared in order to characterize the orthorhombic and pseudocubic polymorphs, respectively. Stoichiometric ye'elimite was successfully prepared but it contained small amount of impurities such as Ca3Al2O6, CaAl2O4 and Ca12Al14O33. Our structural study for stoichiometric ye'elimite by LXRPD and NPD joint Rietveld refinement Abstract 13 showed that this sample presents an orthorhombic symmetry with space group Pcc2. Final unit cell values were a=13.0356(7) Å, b=13.0350(7) Å, and c=9.1677(2) Å. Moreover, several structures were determined by using density functional theory calculations being the lowest energy structure Pcc2, in agreement with our experimental result. For doped ye'elimite, Ca3.8Na0.2Al5.6Fe0.2Si0.2O12SO4, dopants were added as Fe2O3, Na2CO3 and SiO2 obtaining a single crystalline phase. The structural model for this sample has been reported to be based on pseudocubic symmetry with I 3m as space group. Final unit cell value was 9.1974(7) Å. The thermal behavior of the doped sample has been crucial for the determination of the pseudocubic symmetry. DSC measurements of this sample showed a phase transition at 525ºC. Moreover, the high-temperature LXRPD studied showed a sharpening of the diffraction peaks over 500ºC which is likely related with the dynamical disordering of the sulfate anions. Both results indicate that the RT form of doped ye'elimite had a lower symmetry than cubic, consequently is can be assigned to a pseudocubic. Thermodiffractometric studies of stoichiometric ye'emilite showed the transition of the orthorhombic form to a higher symmetry polymorph. Moreover, for stoichiometric ye'elimite, DSC and permittivity measurements were used to measure the reversible phase transition from orthorhombic to cubic which was characterized to occur at 470ºC. In addition, an structural study for both samples at 800ºC was carried out. The crystal structure of stoichiometric ye'elimite was found to be cubic. For doped ye'elimite, the structural study at 800ºC also suggested a truly cubic structure at this temperature. The motivation of revising the crystal structures of both polymorphs of dicalcium silicate and ye'elimite was to use them to perform Rietveld quantitative phase analysis of sulfoaluminate cements. Our reported crystal structures enable Ana María Cuesta García 14 more accurate mineralogical phase analysis of commercial calcium sulfoaluminate cements since lower disagreement values (RF and RWP) were obtained using these structures when compared to results achieved with the previous published ones. After the initial characterization of anhydrous ye'elimite samples, orthorhombic and pseudocubic, hydration studies were carried out. Another important aim of this thesis has been to understand the early age hydration of stoichiometric (orthorhombic) and doped (pseudocubic) ye’elimite phases at early ages in order to understand “eco-cement” performances. Consequently, the influence of different water/cement ratios and the addition of different calcium sulfate sources on hydration kinetic and mechanism have been studied. Hydration studies have been performed by using LXRPD with external standard methodology (G-factor) and/or by using SXRPD with internal standard methodology to determine the full phase assemblage including ACn contents. The internal standard method gives a total ACn value which includes FW. On the other hand, the external standard method was applied to stopped-hydration samples; consequently it enables to obtain only ACn contents. In this case FW was determined by the comparison of DTA-TG weight losses of stopped-hydration samples and the total added water. It is important to highlight that the results obtained by the internal standard method are in agreement with those obtained at later ages by external standard method, showing the consistence of both methodologies to follow hydration reactions with time. Calorimetry studies have also been carried out in order to complement the data obtained by XRPD. The sample preparation was slightly different for each methodology. Samples for SXRPD were prepared in capillaries sealed with grease and for the exsitu LXRPD study, pastes were prepared in a sealed cylinder shape mold and the hydration of the pastes were stopped after specific ages. Firstly, stoichiometric ye'elimite was studied with different amount of water in order to check the influence on the hydration mechanism of ye'elimite of Abstract 15 this parameter. The main conclusion reached is that higher amounts of water enhanced ye'elimite reactivity, as expected. For stoichiometric ye'elimite, AFm was the main hydration product at all ages jointly with AFt. Higher w/s ratio favored the formation of larger amounts of AFm and less AFt. From the mass balance point of view the same AFm/AFt ratio should be present if the reaction is complete. Consequently, we speculate that the higher w/s ratio has yielded to almost complete reaction. The role of ye'elimite polymorphism on the hydration mechanisms has also been established. Their reactivity was compared and it was found that doped ye'elimite (pseudocubic) presented a faster kinetics and yielded higher amounts of AFt than stoichiometric (orthorhombic) ye'elimite under the same experimental conditions. Calorimetric data also confirmed this behavior. The hydration mechanisms of both ye'elimite compounds with a source of soluble sulfate, such as gypsum or anhydrite, were studied at very early ages using SXRPD. It is known that the presence of sulfate phases leads to the formation of AFt. Here, we showed that for a fixed gypsum and water content, AFt was the only crystalline hydrated product and AFm was not found at any hydration time. However, it is important to have in mind that there should be an amorphous phase with general formula, AH3·nH2O which is formed jointly with AFt. The main difference between both ye'elimites was that the hydration kinetics of stoichiometric ye'elimite was faster than that of doped-ye'elimite at very early ages. Concretely, at 14 h of hydration 100% of stoichiometric ye'elimite had reacted while only 85% of doped ye'elimite was consumed after 14h. Moreover, the role of the addition of anhydrite was studied at early ages for the sake of comparison with the hydration behavior when gypsum is added to ye'elimite. It was found that both hydration mechanisms were very similar between both polymorphs but the reaction kinetics using anhydrite were slower due to the lower anhydrite solubility. Ana María Cuesta García 16 Hydration of cements can be summarized as a process of dissolution and crystallization/precipitation of different phases. Although ye'elimite is the main responsible of the hydration CSA cements, it is important to study the hydration of ye'elimite in combination with other phases. For this reason, another important target of this Thesis has been to study the hydration of both ye'elimites, stoichiometric and doped, in combination with other phases such as tetracalcium aluminoferrite, β-C2S and α’H-C2S. Firstly, the hydration of both ye'elimites in combination with C4AF and gypsum has been studied by in-situ SXRPD. In the case of stoichiometric ye'elimite, AFt was firstly formed and then when gypsum was exhausted AFm started to precipitate. Moreover, this ye'elimite was also studied with different amount of w/s in order to determine the water influence on its hydration, obtaining that higher amounts of water enhanced the AFm precipitation and accelerated the reaction kinetic of stoichiometric ye'elimite. The ACn values have also been calculated for these samples. It can be observed that these values slightly diminished with time during the crystallization of AFt and they increased at the time that AFm started to appear. In the case of the hydration of doped ye'elimite with C4AF, the main difference after 46 hours was that this sample yielded much larger relative amounts of AFt (28.6(1) wt%) than with stoichiometric ye'elimite (15.5(3) wt%). So, here, we have proved that the differences in the hydration mechanisms between both ye'elimite polymorphs are exacerbated when tetracalcium aluminoferrite is in the reaction medium. In addition, SEM-EDS has been used in order to check if the iron which belongs to C4AF phase is incorporated in the crystalline hydrated phases of these samples. It can be stated that AFm and AFt formed in the hydration of ye'elimite with gypsum and C4AF does not contain appreciable iron content in their structures. Therefore, an iron-bearing amorphous phase should be present. Abstract 17 Hydration of ye'elimite has also been studied in combination with dicalcium silicate by LXRPD. Here we have reported the role of ye'elimite and dicalcium silicate polymorphisms on their hydration mechanisms. These studies will help to unravel the hydration mechanisms of BCSA. It is known that BCSA cements which contain the α'H-C2S polymorphs develop higher mechanical strengths values than cements that contain β-C2S polymorph. This behavior was previously proposed but not directly demonstrated. However, it is essential to establish the role of both-polymorphs of ye'elimite in these mixtures. Here, we have studied all the combinations between dicalcium silicate and ye'elimite until 6 months. Powder diffraction data at 28 days showed some small differences between samples. The sample which contained both α'HC2S and doped-ye'elimite presented a higher degree of reaction and higher amounts of stratlingite at that age. However, important differences are expected at long ages. In the tested experimental conditions, both ye'elimite polymorphs yielded mainly crystalline ettringite. However, we have demonstrated that the hydration degree of α'H-dicalcium silicate was much larger than that of β-C2S in the investigated time range (up to six months). At that time, α'H-dicalcium silicate was totally dissolved and, conversely, high amounts of β-C2S were found in the samples. It is important to bear in mind that there were no differences between stoichiometric ye'elimite and doped ye'elimite in these samples due to the fact that studies were done from 7 days and ye'elimite hydration occurs earlier. After the hydration studies of ye'elimite, the characterization of tetracalcium aluminoferrite hydration has also been carried out. The anhydrous sample was successfully prepared obtaining a phase assemblage of 98.1(1) wt% of C4AF and 1.9(1) wt% of C3A as an impurity. Hydration behavior of C4AF in selected experimental conditions has been thoroughly studied. C4AF has been hydrated in the absence and presence of gypsum. C4AF in the presence of water hydrated to form mainly a hydrogarnet- Ana María Cuesta García 18 type phase, C3A0.845F0.155H6. This hydrated phase was measured by high resolution SXRPD in order to perform a structural study. In addition, TEM combined with EDS studies were used to confirm that iron was incorporated in the hydrogarnet phase, C3A0.845F0.155H6. A crystal structure for this phase is reported here. The hydration of tetracalcium aluminoferrite in the presence of gypsum was different than the previous one and yielded a mixture of AFm and AFt. Firstly, AFt precipitated jointly with amorphous aluminum hydroxide and once gypsum was completely dissolved crystalline AFm started to precipitate. Calorimetric data showed a broad signal which correspond to the AFm formation. SEM-EDS studies have been used to corroborate that AFt and AFm phases incorporate some amount of iron. Finally, new on-going research are being carried out using complementary techniques for the characterization of pure hydrated phases and for the study of some hydration mechanisms. Firstly, ptychographic X-ray computed tomography (PXCT) has been used here to study the hydration of ye'elimite samples. After different hydration times, the samples (cement pastes) were investigated at cSAXS beamline is SLS (Swiss Light source). The main goal is to characterize the microstructures of the pastes and to quantify the electron and mass densities of the phases present in these samples. Then, the densities were compared with the theoretical values in order to identify the phases. The measured mass densities matched well with the expected values. In order to obtain better results, the histograms of a VOI (volume of interest) were carried out for each sample excluding the capillaries. Moreover, a high pressure SXRPD experiment has also been carried out. Hydrated samples were placed in DAC to determine their behavior under pressure. This technique enables to determine the bulk modulus and the stability of these cement phases. This pressure study has been carried out for an iron-hydrogarnet sample, C3A0.845F0.155H6, obtaining a bulk modulus of 54(2) GPa. Resumen Ana María Cuesta García 26 En el caso de la ye'elimita dopada, Ca3.8Na0.2Al5.6Fe0.2Si0.2O12SO4, se obtuvo una única fase cristalina, en este caso los dopantes se añadieron como Fe2O3, Na2CO3 y SiO2. Es importante destacar que esta composición se eligió basándose en un clínker tipo BCSA previamente preparado en el cual ya se había estabilizado la forma pseudocúbica de la ye'elemita. El modelo estructural se basa en una simetría pseudocúbica con grupo espacial I 3m. El valor final de la celda unidad es 9.1974(7) Å. El comportamiento térmico de la muestra dopada fue clave para determinar que su simetría es pseudocúbica. Las medidas de DSC muestran una transición de fase reversible a 525ºC, y los estudios de LXRPD a alta temperatura mostraron un estrechamiento de los picos de difracción a partir de 500ºC, que está relacionado con el desorden de los aniones sulfato. Ambos resultados indican que la forma de la ye'elimita dopada a temperatura ambiente tiene que presentar una simetría menor que la cúbica, consecuentemente se le puede asignar una simetría pseudocúbica. Los estudios termodifractométricos de la ye'elimita estequiométrica muestran una transición de fase desde la forma ortorrómbica a un polimorfo de mayor simetría (cúbico). Mediante DSC y medidas de permitividad se determinó que la transición reversible está centrada a 470ºC. Además, se llevó a cabo un estudio estructural para las dos muestras de ye'elimita a 800ºC. A esta temperatura las dos presentaron una simetría cúbica real. Después de la caracterización de la ye'elimita, se llevó a cabo la síntesis y caracterización de la fase ferrito aluminato tetracálcico. Esta muestra se preparó obteniéndose un ensamblaje de fases de 98.1(1) % de C4AF y 1.9(1) % de C3A como impureza (porcentajes en peso). Debido a la alta calidad del ajuste obtenido por SXRPD se han optimizado los factores de ocupación de Fe/Al de la estructura obteniendose 0.746(2) de ocupación de Fe en el sitio octaédrico y por lo tanto, 0.254(2) de ocupación de Fe en el sitio tetraédrico. Resumen 27 Las estructuras cristalinas de ambos polimorfos de silicato dicálcico y de la ye'elimita se han revisado para utilizarlas en el análisis cuantitativo de fases de cementos de sulfoaluminato por el método de Rietveld. Las estructuras publicadas presentadas en la tesis doctoral permiten un análisis mineralógico de fases más exacto para cementos comerciales de sulfoaluminato de calcio y BCSA ya que con ellas se obtienen valores de desacuerdo más bajos (RF y RWP) que cuando se utilizan las estructuras publicadas anteriormente. Tras la caracterización inicial de las muestras anhidras de ye'elimita, ortorrómbica y pseudocúbica, se llevaron a cabo los estudios de hidratación. Un objetivo muy importante de la tesis es correlacionar el mecanismo de hidratación a edades tempranas con las prestaciones de los “ecocementos”. Se ha estudiado la influencia en la cinética y en el mecanismo de hidratación de diferentes relaciones agua/sólido y de la adición de fuentes de sulfato. Los estudios de hidratación se han realizado usando LXRPD con la metodología del estándar externo (factor-G) y/o usando SXRPD con la metodología del estándar interno para determinar el ensamblaje de fases incluyendo el material amorfo (ACn, del inglés Amorphous and Crystalline nonquantified). La metodología del estándar interno nos da un valor total de ACn, el cual incluye la cantidad de agua libre (FW, del inglés Free Water). Por otro lado, la metodología del estándar externo se aplicó a muestras donde la hidratación se detuvo a determinadas edades (más adelante “muestras con la hidratación parada”), por tanto esto permite obtener los valores individuales de ACn. El valor de FW se determinó por comparación entre el valor de las pérdidas de peso de las muestras con la hidratación parada obtenido por análisis térmico diferencial y termogravimétrico (DTA-TG, del inglés Differential Thermal AnalysisThermoGravimetric) y el valor de la cantidad total de agua añadida. Es importante destacar que los resultados obtenidos por el método del estándar interno están de acuerdo con los obtenidos a altas edades usando el método del estándar externo, mostrando una consistencia entre ambas metodologías que permite entender las Ana María Cuesta García 28 reacciones de hidratación en función del tiempo. Para complementar los datos obtenidos por XRPD también se han realizado estudios calorimétricos. La preparación de la muestra es ligeramente distinta para cada metodología: i) las muestras que se usaron para los estudios de SXRPD se prepararon en capilares sellados con grasa y, ii) en el caso de los estudios de hidratación ex-situ mediante LXRPD, las pastas se prepararon en moldes sellados con forma de cilindros y se pararon después a determinadas edades. El grado de cristalinidad y de hidratación de fases del tipo AFm depende de la metodología de preparación. En el caso de la muestra en capilar sellado la cristalinidad es mayor que aquellas en las que se ha parado la hidratación. La ye'elimita estequiométrica se estudió con distintas cantidades de agua para comprobar su influencia en el mecanismo de hidratación. Como cabe esperar a mayor cantidad de agua se intensifica la reactividad de la ye'elimita. El producto principal de hidratación a todas las edades ha sido AFm y AFt. Mayores relaciones agua/sólido favorecen la formación de mayores cantidades de AFm y desfavorece la de AFt. Desde el punto de vista del balance de masas se debe mantener la misma proporción AFm/AFt si la reacción es completa. Por lo tanto, se puede especular que mayores relaciones agua/sólido han conducido a una reacción casi completa. Se ha determinado el papel del polimorfismo de la ye'elimita en los mecanismos de hidratación comparándose la reactividad de ambos polimorfos. La ye'elimita dopada (pseudocúbica) presenta una cinética de hidratación más rápida y se forma una cantidad mayor de AFt que en el caso de la ye'elimita estequiométrica (ortorrómbica) en las mismas condiciones experimentales. Los datos de calorimetría también confirman este comportamiento. Los mecanismos de hidratación a edades tempranas de ambas ye'elimitas se han estudiado en presencia de una fuente de sulfato (yeso o anhidrita) usando SXRPD ya que se sabe que cuando hay suficiente sulfato en el medio se forma exclusivamente AFt. En este estudio se demostró que para una cantidad Resumen 29 determinada de yeso y agua, la única fase cristalina que se obtiene es el AFt, no encontrandose AFm a ninguna edad. Por otro lado, es importante tener en cuenta que en la hidratación de las muestras se debería formar una fase amorfa con fórmula general, AH3·nH2O, la cual se obtiene en la reacción de hidratación junto con AFt. La principal diferencia que se encuentra entre ambos polimorfos de la ye'elimita es que la cinética de hidratación de la estequiométrica es más rápida que la de la dopada a edades tempranas. Exactamente, a 14 h de hidratación reacciona un 100% de la ye'elimita estequiométrica mientras que solo lo hace un 85% de la dopada. Además, se ha estudiado el papel de la adición de anhidrita a edades tempranas para hacer la misma comparación que con la adición de yeso. En este caso, se encontró que el mecanismo de hidratación es muy similar para ambos polimorfos pero la cinética de reacción usando anhidrita es mucho más lenta debido a su menor solubilidad. La hidratación de los cementos se puede resumir como un proceso de disolución y cristalización/precipitación de diferentes fases. Aunque la ye'elimita es la principal responsable del mecanismo de hidratación de los cementos tipo CSA, es importante estudiar su hidratación en combinación con otras fases. En primer lugar, se ha estudiado la hidratación en combinación con C4AF y yeso usando in-situ SXRPD. En el caso de la ye'elimita estequiométrica, a primeras horas se formó AFt y después, cuando el yeso se consumió totalmente empezó a precipitar AFm. Además, se estudió también esta ye'elimita con diferentes cantidades de agua/sólido para poder determinar la influencia del agua. Se obtuvo que mayores cantidades de agua favorecen la precipitación de AFm y aceleran la cinética de hidratación para la ye'elimita estequiométrica. Como se explicó anteriormente, los valores de ACn para estas muestras se obtuvieron por SXRPD. En todos los casos, estos valores disminuyeron ligeramente con el tiempo durante la cristalización del AFt y se incrementaron en el momento que el AFm empezó a aparecer. Ana María Cuesta García 30 Este mismo estudio se ha realizado en ausencia de yeso. Se ha comprobado que el único producto de hidratación es AFm. Se demuestra que el yeso es el principal responsable de la formación del AFt. En el caso de la hidratación de la ye'elimita dopada en presencia de C4AF y yeso, la principal diferencia que se encontró fue que a 46 horas esta muestra formaba mucha más cantidad de AFt (28.6(1) % en peso) que en el caso de la ye'elimita estequiométrica (15.5(3) % en peso). Debido a estos resultados se ha probado que las diferencias en la hidratación entre ambos polimorfos se intensifican cuando hay ferrito aluminato tetracálcico en el medio. Además, mediante microscopía electrónica de barrido acoplada con espectroscopia de energía dispersiva (SEM-EDS, del inglés Scanning Electron MicroscopyEnergy Dispersive Spectroscopy) se ha comprobado que el hierro que pertenece al C4AF no se incorpora de forma apreciable en las fases cristalinas hidratadas (AFm y AFt). Por lo tanto, se debe haber formado una fase amorfa con alto contenido en hierro. Por último, se ha estudiado el papel de los polimorfos de la ye'elimita y del silicato dicálcico en sus mecanismos de hidratación mediante LXRPD. Se cree que estos estudios ayudaran a descifrar los mecanismos de hidratación de cementos tipo BCSA. Los cementos BCSA que contienen el polimorfo α'H-C2S desarrollan mayores resistencias mecánicas que aquellos cementos que contienen el polimorfo β-C2S. Este comportamiento se propuso anteriormente pero no se demostró directamente, por tanto, es esencial establecer el papel de ambos polimorfos de la ye'elimita en estas muestras. En la tesis se han estudiado todas las combinaciones entre el silicato dicálcico y la ye'elimita hasta edades de 6 meses. Los datos de difracción de polvo a 28 días muestran pequeñas diferencias. La muestra que contenía α'H-C2S y ye'elimita dopada presentó un mayor grado de hidratación y cantidad de stratlingita a esa edad. Sin embargo, se esperan aun mayores diferencias a largas Resumen 31 edades. En las condiciones experimentales estudiadas, ambos polimorfos de la ye'elimita formaron principalmente etringita. No obstante, se ha demostrado que el grado de hidratación del polimorfo α'H del silicato dicálcico es mucho mayor que el del polimorfo β en el rango de tiempo estudiado. A los 6 meses, el α'H-C2S se disolvió totalmente, en cambio, se encontraron grandes cantidades de β-C2S sin reaccionar en todas las muestras. Es importante destacar que no hubo diferencias importantes entre la ye'elimita dopada y la estequiométrica debido a que estos estudios se hicieron a partir de 7 días, y la hidratación de la ye'elimita ocurre mucho antes (en torno a 8 horas). Se ha determinado el comportamiento de hidratación de la fase C4AF en ausencia y presencia de yeso. En el caso de la hidratación del C4AF en ausencia de yeso se formó una fase tipo hidrogranate con fórmula C3A0.845F0.155H6. Esta fase hidratada se estudió por SXRPD de alta resolución y microscopía electrónica de transmisión (TEM, del inglés Transmission Electron Microscopy) combinada con EDS para confirmar que el hierro se incorpora a la fase tipo hidrogranate, C3A0.845F0.155H6. Se obtuvo una descripción estructural muy precisa con una celda unidad de a=12.60315(4) Å y V=2001.88(2) Å3. La estructura cristalina revisada condujo a unos valores de desacuerdo de RWP=8.1% y RF=4.8%. La hidratación del C4AF en presencia de yeso tuvo un comportamiento distinto. En este caso se formó una mezcla de AFm y AFt. En primer lugar, precipitó el AFt junto con hidróxido de aluminio amorfo y una vez que el yeso se disolvió totalmente empezó a precipitar el AFm. Los datos de calorimetría muestran una señal ancha que corresponde a la formación del AFm. De nuevo, los estudios de SEM-EDS se han utilizado para corroborar que las fases formadas, AFt y AFm, en este caso, sí incorporan hierro en su estructura. Se ha comprobado que en presencia de ye'elemita, la hidratación del ferrito aluminato tetracalcico se inhibe. Concretamente, la ye'elemita dopada tiene un Ana María Cuesta García 32 efecto de inhibición mucho más fuerte sobre el C4AF haciendo que el principal producto de hidratación que se forma sea mayoritariamente AFt en lugar de AFm. Actualmente se están llevando a cabo nuevas investigaciones con técnicas complementarias a las usadas en esta tesis para la caracterización de las fases hidratadas puras y para el estudio de los mecanismos de hidratación. En primer lugar, se ha usado pticografía-tomografía de rayos-X computerizada (PXCT, del inglés ptychographic X-ray computed tomography) para estudiar la hidratación de muestras de ye'elimita. Se han investigado diferentes pastas con distintas edades de hidratación en la línea cSAXS del sincrotrón Swiss Light Source (Villigen, Suiza). El objetivo principal es caracterizar la microestructura de las pastas y cuantificar la densidad electrónica y másica de las fases presentes. Posteriormente, estas densidades experimentales se comparan con las teóricas lo que permite identificar las fases presentes. Se han obtenido acuerdos muy buenos comparándolos con los valores teóricos. Además, para obtener mejores resultados se estudiaron los histogramas seleccionando una región de interés para cada muestra eliminando la zona del capilar. Por último, se ha realizado un experimento de SXRPD de alta presión. La muestras hidratadas se colocaron en celdas de diamante (DAC, del inglés Diamond anvil cell) para determinar su comportamiento bajo presión. Esta técnica permite determinar el módulo de comprensibilidad y la estabilidad de estas fases del cemento. Este estudio a presión se realizó para diversas muestras, destacando la muestra de hidrogranate con hierro, C3A0.845F0.155H6, para la cual se obtuvo un módulo de comprensibilidad de 54(2) GPa. 1. Introduction 1. Introduction 35 1. Introduction to cements OPC is the most used building material, which has become the dominant binder used in the preparation of concrete for construction applications. OPC is now prepared by heating a mixture of limestone and clay, or other materials of similar composition and sufficient reactivity, at a temperature of 1450ºC (Taylor 1997). Partial fusion occurs, and nodules of clinker are produced. The cement is finally prepared by grinding the clinker and mixing it with finely ground calcium sulfate, which will control the setting time. It is commonly described as gypsum, but this may be partly or wholly replaced by other forms of calcium sulfate. Some specifications allow the addition of other materials at the grinding stage. The main supplementary cementitious materials are fly ash, ground granulated blast furnace slag, silica fume, calcined clays and natural/industrial pozzolans. These materials are commonly blended with clinker to produce the different Portland cement types or used as a replacement for a portion of clinker in concrete mixtures (Juenger and Siddique 2015). 1.1. Ye'elimite-containing cements Since 1970s commercial ye'elimite-containing cements known as CSA cements have been manufactured and used on a large scale in China (Zhang et al. 1999). CSA cements are/were known as “the third series cements of China” (the first series being OPC and the second series being CAC) with special applications such as high strength developments at early-ages used in precast concrete and at moderate curing temperatures (Glasser and Zhang 2001; Quillin 2001), selfstressing materials (Péra and Ambroise 2004; Georgin et al. 2008), with expansive properties for shrinkage compensating concrete (Klein 1963; Chen et al. 2012) or for radioactive element encapsulation in high-density cement pastes (Zhou et al. 2006; Cau Dit Coumes et al. 2009; Sun et al. 2011). Ana María Cuesta García 42 1.2.1.1. Dicalcium silicate chemical compositions and structures Stoichiometric C2S has five polymorphs (Mumme et al. 1995) , , α'L, α'H and α, on heating, with -C2S being the form that commonly prevails in OPC and in BCSA without activation as it is stabilized by elements substitution(s). Its temperature evolution is shown in Figure 1.2. The structures of the polymorphs belong to a large family typified by that of glaserite, K3Na(SO4)2 (Taylor, 1997) except the -phase, which is essentially nonreactive with water and stable at ambient temperature and it crystallizes in an orthorhombic olivine-type structure (Smith et al. 1965; Czaya 1971; Udagawa et al. 1980). The -form is a metastable monoclinic phase at room temperature (Midgley 1952; Jost et al. 1977). This metastability has also been confirmed by the results of a structural simulation of possible hypothetical Ca2SiO4 phases (Yamnova et al. 2011). α'L and α'H orthorhombic phases are stable at higher temperatures (Susuki and Yamaguchi 1968). The α'L-polymorph is generally considered a superstructure of the α'H and it has been reported two possibilities for indexing: doubling the a and c parameters (Barnes et al. 1980) or tripling the b parameter (Saalfeld 1975; Jelenic and Bezjak 1982; II’inets and Bikbau 1990). Finally, the highest-temperature polymorph is the -form whose structure is still under discussion (Bredig 1950). It is noteworthy, that  →  polymorphic transformation on cooling is disruptive with and important change in volume. The density of the -modification of 2.94 g/cm3 is about 10% lower than the density of the -modification of 3.20 g/cm3. As a result of this volume change an originally compact burning product cracks and rapidly disintegrates into dust (“dusting”) as soon as the temperature during cooling falls below about 500ºC (Locher 2006) if the -phase has not been stabilized. 1. Introduction 43 Figure 1.2. Dicalcium silicate polymorphic transformations with temperature. The structures of all polymorphs are built from Ca2+ and SiO44ions. The arrangements of these ions are closely similar in the , 'H, 'L and  polymorphs, but that in -C2S is somewhat different. The crystal structure of -C2S has two calciums in regular six-coordinated oxygen environments. Meanwhile, the crystal structure of -C2S has two calciums surrounded by eight oxygens in distorted environments. Furthermore, the crystal structures of 'Hand - polymorphs have calcium cations in both eight and nine irregular coordinations (Midgley 1952; Smith et al. 1965; Regourd et al. 1968; Czaya 1971; Ghosh et al. 1979; Udagawa et al. 1980). It is worth noting that there are experimental evidences that an increase of the calcium coordination number seems to enhance the water reactivity (Kim and Hong 2004). Table 1.2 shows the crystallographic information of the corresponding dicalcium silicate polymorphs.    ' L  ' H  1425ºC 1160ºC 780-860ºC 690ºC < 500ºC    ' L  ' H  1425ºC 1160ºC 780-860ºC 690ºC < 500ºC Ana María Cuesta García 44 Table 1.2. Crystallographic data for dicalcium silicate “polymorphs” including temperature of stabilization or stabilizer. Polymorph Space group Unit cell parameters V/Z T(ºC)/ ICSD a (Å) b (Å) c (Å) (º) (Å3) Stabilizer  P63/mmca 5.420 5.420 7.027 90.0 89.4 - 81099 P63/mmcb 5.532(9) 5.532(9) 7.327(11) 90.0 97.1 1545/- 82998 P-3m1b 5.532(9) 5.532(9) 7.327(11) 90.0 97.1 1545/- 82999 'H Pnmaa 6.7673(4) 5.5191(4) 9.3031(6) 90.0 86.9 -/5% (molar) Ca3(PO4)2 81097 Pnmab 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 -/Ca1.8Sr0.2SiO4 49662 'L Pna21b 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 -/ Ca0.84Sr1.16SiO4 39203 P21cne 5.566 9.355 20.569 90.0 89.3 - 39100  P21/na 5.512(0) 6.758(0) 9.314(0) 94.6 86.5 -/0.5 wt% Cr2O3 81096 P21/nf 5.48(2) 6.76(2) 9.28(2) 85.5 85.7 - 24640 P21/ng 5.502(1) 6.745(1) 9.297(1) 94.6 86.0 - 963  Pbnma 5.082(0) 11.224(0) 6.764(0) 90.0 96.5 - 81095 Pbnmh 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) Il´inets and Bikbau 1990; (e) Udagawa et al. 1979; (f) Midgley 1952; (g) Jost et al. 1977 (h) Udagawa et al. 1980. Physical and chemical properties of phases can be altered by introducing within the defects or strains crystalline structures. The different type of defects can be introduced during the material preparation by the formation of solid solutions or by specific thermal treatments (Fukuda and Ito 1999). Moreover, these defects can even stabilize high-temperature forms of C2S at room temperature (Ghosh et al. 1979; Nettleship et al. 1992). Depending on the nature and concentration of the foreign substance or mixture it is possible to selectively produce certain modifications. There are many studies concerning the chemical-stabilization of - C2S by foreign ions such as SO3, 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; Zhao et al. 2013). Also, the presence of Fe2O3 can stabilize the  polymorph by sol-gel method using microwave and conventional heating (Gajbhiye and Singh 2010). The stabilization 1. Introduction 45 of '-forms by introduction of foreign oxides, such as MgO, P2O5, K2O, BaO and SO3 has also been studied (Bensted 1979; Fukuda et al. 2001; Park 2001; Zhao et al. 2013). These works stated that hydraulic properties were increased when compared to the materials without foreign ions. B2O3 has also been studied in order to stabilize  and '-forms at room temperature (Jelenic et al. 1978; Wesselsky and Jensen 2009). Some authors (Park 2001; Kim and Hong 2004; Wesselsky and Jensen 2009) have demonstrated that the effectiveness of addition of B2O3, with no other co-dopants, for stabilizing the α'-forms, is poor. On the other hand, the addition of a combination of dopants, for instance B2O3 and Na2O, successes to stabilize α'-C2S (Wesselsky and Jensen 2009) although these authors did not state which α'-form, α'L or α'H, was stabilized. Moreover, previous studies reported that the stabilization of '-belite forms in a cement matrix can be attained by introducing minor elements, such as alkaline oxide, boron or phosphor, in raw materials (Morsli et al. 2007a,b; Li et al. 2007; Wesselsky and Jensen 2009; Cuberos et al. 2010; Morin et al. 2011; Álvarez-Pinazo et al. 2012). Figure 1.3 shows the simulated laboratory X-ray powder diffractograms of different C2S polymorphs. The stabilization of the high temperature polymorph is due to both ionic substitutions and quenching. Moreover, Remy and Andrault (1997) established new formulas which represent the variation of molar volume as a function of temperature for the dicalcium silicate polymorphs (see Table 1.3). Ana María Cuesta García 46 Figure 1.3. Range from 29 to 35º (2 CuKα) of the theoretical diffractograms at RT of: (a) -C2S (Mumme et al. 1995), (b) 'H-C2S (Mumme et al. 1996), (c) -C2S (Mumme et al. 1995) y (d) -C2S (Udagawa et al. 1980). Simulacion Alfa-H_C2S Hist 1 Lambda 1.5406 A, L-S cycle 6 Obsd. and Calc. Profiles 2-Theta, deg C o u n ts 29.0 30.0 31.0 32.0 33.0 34.0 35.0 X 1 0 E 5 0 .0 0 .5 1 .0 1 .5 2 .0 Simulacion Beta_C2S Hist 1 Lambda 1.5406 A, L-S cycle 4 Obsd. and Calc. Profiles 2-Theta, deg C o u n ts 29.0 30.0 31.0 32.0 33.0 34.0 35.0 X 1 0 E 5 0 .0 0 .5 1 .0 1 .5 Simulacion Beta_C2S Hist 1 Lambda 1.5406 A, L-S cycle 3 Obsd. and Calc. Profiles 2-Theta, deg C o u n ts 29.0 30.0 31.0 32.0 33.0 34.0 35.0 X 1 0 E 5 0 .0 0 .5 1 .0 Simulacion Gamm a_C2S Hist 1 Lambda 1.5406 A, L-S cycle 4 Obsd. and Calc. Profiles 2-Theta, deg C o u n ts 29.0 30.0 31.0 32.0 33.0 34.0 35.0 X 1 0 E 5 0 .0 0 .5 1 .0 1.5 I(u.a.)I(u.a.) I(u.a.) I(u.a.) (a) (c) (b) (d) Simulacion Alfa-H_C2S Hist 1 Lambda 1.5406 A, L-S cycle 6 Obsd. and Calc. Profiles 2-Theta, deg C o u nts 29.0 30.0 31.0 32.0 33.0 34.0 35.0 X 1 0 E 5 0 .0 0 .5 1 .0 1 .5 2 .0 Simulacion Beta_C2S Hist 1 Lambda 1.5406 A, L-S cycle 4 Obsd. and Calc. Profiles 2-Theta, deg C o u nts 29.0 30.0 31.0 32.0 33.0 34.0 35.0 X 1 0 E 5 0 .0 0 .5 1 .0 1 .5 Simulacion Beta_C2S Hist 1 Lambda 1.5406 A, L-S cycle 3 Obsd. and Calc. Profiles 2-Theta, deg C o u nts 29.0 30.0 31.0 32.0 33.0 34.0 35.0 X 1 0 E 5 0 .0 0 .5 1 .0 Simulacion Gamm a_C2S Hist 1 Lambda 1.5406 A, L-S cycle 4 Obsd. and Calc. Profiles 2-Theta, deg C o u nts 29.0 30.0 31.0 32.0 33.0 34.0 35.0 X 1 0 E 5 0 .0 0 .5 1 .0 1 .5 I(u.a.)I(u.a.) I(u.a.) I(u.a.) (a) (c) (b) (d) (a) (b) (c) (d) I(u.a.)I(u.a.)I(u.a.) I(u.a.) Simulacion Alfa-H_C2S Hist 1 Lambda 1.5406 A, L-S cycle 6 Obsd. and Calc. Profiles 2-Theta, deg C o u n ts 29.0 30.0 31.0 32.0 33.0 34.0 35.0 X 1 0 E 5 0 .0 0 .5 1 .0 1 .5 2 .0 Simulacion Beta_C2S Hist 1 Lambda 1.5406 A, L-S cycle 4 Obsd. and Calc. Profiles 2-Theta, deg C o u n ts 29.0 30.0 31.0 32.0 33.0 34.0 35.0 X 1 0 E 5 0 .0 0 .5 1 .0 1 .5 Simulacion Beta_C2S Hist 1 Lambda 1.5406 A, L-S cycle 3 Obsd. and Calc. Profiles 2-Theta, deg C o u n ts 29.0 30.0 31.0 32.0 33.0 34.0 35.0 X 1 0 E 5 0 .0 0 .5 1 .0 Simulacion Gamm a_C2S Hist 1 Lambda 1.5406 A, L-S cycle 4 Obsd. and Calc. Profiles 2-Theta, deg C o u n ts 29.0 30.0 31.0 32.0 33.0 34.0 35.0 X 1 0 E 5 0 .0 0 .5 1 .0 1 .5 I(u.a.)I(u.a.) I(u.a.) I(u.a.) (a) (c) (b) (d) Simulacion Alfa-H_C2S Hist 1 Lambda 1.5406 A, L-S cycle 6 Obsd. and Calc. Profiles 2-Theta, deg C o u nts 29.0 30.0 31.0 32.0 33.0 34.0 35.0 X 1 0 E 5 0 .0 0 .5 1 .0 1 .5 2 .0 Simulacion Beta_C2S Hist 1 Lambda 1.5406 A, L-S cycle 4 Obsd. and Calc. Profiles 2-Theta, deg C o u nts 29.0 30.0 31.0 32.0 33.0 34.0 35.0 X 1 0 E 5 0 .0 0 .5 1 .0 1 .5 Simulacion Beta_C2S Hist 1 Lambda 1.5406 A, L-S cycle 3 Obsd. and Calc. Profiles 2-Theta, deg C o u nts 29.0 30.0 31.0 32.0 33.0 34.0 35.0 X 1 0 E 5 0 .0 0 .5 1 .0 Simulacion Gamm a_C2S Hist 1 Lambda 1.5406 A, L-S cycle 4 Obsd. and Calc. Profiles 2-Theta, deg C o u nts 29.0 30.0 31.0 32.0 33.0 34.0 35.0 X 1 0 E 5 0 .0 0 .5 1 .0 1 .5 I(u.a.)I(u.a.) I(u.a.) I(u.a.) (a) (c) (b) (d) (a) (b) (c) (d) I(u.a.)I(u.a.)I(u.a.) I(u.a.) 1. Introduction 47 Table 1.3. Molar volume data as a function of temperature for dicalcium silicate. Polymorphs Vmol (cm3) = f(T(ºC)) ∆T (ºC)  58.085 + 0.00137(T) + 5.205 x 10-7(T)2 25-800  51.827 + 0.00191(T) + 4.527 x 10-7(T)2 25-700 'L 51.212 + 0.00326(T) + 1.377 x 10-7(T)2 800-1100 'H 44.842 + 0.01466(T) – 4.899 x 10-6(T)2 1200-1420  51.117 + 0.00466(T) 1500 -1700 The belite grains in Portland cement clinkers frequently show complex striated structures. These have been studied over a long period by workers using optical microscopy; Yamaguchi and Takagi(1969) and Ono et al. (1969) who also used XRPD and other methods, gave the first substantially complete interpretation. A very common type of belite grain in production clinkers, called Type I belite, is rounded, typically 20-40m in mean dimension, and shows at least two sets of parallel striations (Taylor 1997). Type I belite grains are those which have crystallized from the liquid at temperatures above 1420ºC. The primary striations arise on cooling through the α to α´H transition, in which the symmetry decreases from hexagonal to orthorhombic, each set of these striations is representing a different orientation of the α´H structure. The transition to α´L-C2S that occurs on further cooling does not increase the number of orientations, but that from α´L to , in which the symmetry falls to monoclinic, causes each orientation to split into two (Figure 1.4.a) Type II belites are irregular grains with one set of distinct parallel striations (lamellae). This is not a common variety in normal clinker. Parallel striations in Type II belite from the α´ to  transformation are polysynthetic twins, observed in clinkers produced at a very low temperature or cooled very slowly (Ono 1995), see Figure 1.4.b. Ana María Cuesta García 48 Type III belite, which does not present lamellae, is uncommon in normal plant clinkers and is a merely single crystal with an uniform internal microstructure (Campbell 1999). Figure 1.4. a) Type I belite (Insley 1936). Belite crystals exhibit relatively large areas which are either nonlamellar or show a single set of parallel lamellae b)Type II belite (Insley 1936). Polysynthetically twinned belite in lower left and slightly splintery gamma belite (upper right) in matrix of coarsely microcrystalline ferrite and aluminate (Campbell (1999)). 1.2.1.2. Hydration mechanisms of dicalcium silicate The C2S hydration is of special interest in BCSA. However, the hydration kinetic is slow when compared to other phases, i.e. ye'elimite. Some efforts are being performed to enhance the hydration kinetic, for instance, by adding minor elements during clinkering to obtain high temperature polymorphs of C2S (Gartner and Li 2006; Li et al. 2007; Cuberos et al. 2010; Álvarez-Pinazo et al. 2012). Thus, α-forms of belite are considered more reactive than -C2S. The way the polymorphs are stabilized can have a significant influence on its reactivity (Gartner et al. 2002). Jost and Ziemer (1984) argue that the sharing faces of CaOx polyhedral is an important feature controlling reactivity. Both α' and -forms have distorted CaO7 polyhedra, while -polymorph has a relatively symmetrical closed- 1. Introduction 49 packed CaO6 octahedra which do not share faces (Jost and Ziemer 1984). Nevertheless, the hydration reaction could be the same independently of the rate of reaction of αand - polymorphs, where calcium silicate hydrate gel and portlandite are formed. C2S + (x+2-y)H → CySHx + (2-y)CH [1.3] Calcium silicate hydrates (CySHx) are produced with a crystalline order which, according to X-ray diffraction investigations, is very low. They are extremely fine-grained, so they are also designated like a gel. The hydration of belite to form the amorphous gel CySHx and portlandite corresponds to reaction [1.3]. In the C-S-H, the Ca/Si molar ratio lies between 1.2 and 2.3 with an average value of about 1.75 (Taylor 1986; Richardson 1999; Scrivener 2015). The inner CS-H is produced as a dense mass at the place of the original cement grain. The outer C-S-H forms fibers in the water-filled interstitial spaces between the cement grains. The outer C-S-H is richer in CaO than the inner gel and also contains more amounts of Al2O3, K2O and SO3 (Taplin 1959; Rodgers and Aldrige 1977). However, only few authors have found the presence of portlandite and C-S-H in their hydration studies, as reaction 1.3 shows (El-Didamony et al. 2012). It is important to highlight that if belite coexists with aluminum rich amorphous hydrates (AH3·nH2O), the formation of stratlingite (AFm-type) will be favored (Palou et al. 2005; Cuberos et al. 2010; Gartner and Macphee 2011; Álvarez-Pinazo et al. 2013). C2S + AH3 + 5H→ C2ASH8 [1.4] Reaction [1.4] consumes the amorphous AH3 formed by the hydration reactions of aluminum-rich phases. The presence of stratlingite has been Ana María Cuesta García 50 confirmed, for example, in BCSAF and CSA pastes by XRPD and DTA techniques by numerous independent investigations (Cuberos et al. 2010; Aranda et al. 2011; Morin et al. 2011; Álvarez-Pinazo et al. 2013; Ioannou et al. 2014). Other reactions need to be taken into account. Some studies have confirmed the presence of katoite phases (see section 1.2.6), also known as siliceous hydrogarnet (Taylor 1997). It can be formed according to the following chemical equation [1.5]. C2S + 1/2C4AF + 5H→ C3(A,F)SH4 + CH [1.5] Reaction [1.5] requires the formation of portlandite, which is not detected by XRPD or DTA, through the consumption of ferrite. Consequently, it is speculated that this portlandite may be consumed by stratlingite, producing larger quantities of katoite, see reaction [1.6]: C2ASH8 + CH → C3ASH4 + 5H [1.6] 1.2.2. Ye'elimite (Ca4[Al6O12]SO4, C4A3S) Ye'elimite was introduced as a cementitious phase in the 1960s, when it was patented by Alexander Klein (Klein 1963) as an expansive or shrinkage compensating addition to cementitious binders ("Klein compound"). The interest in the ye'elimite structure (Klein’s salt or tetracalcium trialuminate sulfate (C4A3S)) has recently increased because it is the major component in calcium sulfoaluminate cements, CSA, (Odler 2000; García-Maté et al. 2012, 2013; Telesca et al. 2014), and the second most relevant phase in sulfobelite cements, BCSA, (~25 wt%) (Cuberos et al. 2010; Morin et al. 2011; Álvarez-Pinazo et al. 2013; Bullerjahn et al. 2014). 1. Introduction 51 1.2.2.1. Ye'elimite chemical compositions and structures Ye'elimite, Ca4[Al6O12]SO4, crystallizes in a sodalite type structure which has a tetrahedral framework and it is built from all-corner-connected tetrahedra. Sodalites of the general composition M4[T6O12]X have been known for many years as both naturally-occurring minerals and as synthetic compounds (Hurlbut and Klein 1985). This general formula refers to a structure that is (ideally) a bodycentered cubic unit cell with a lattice parameter of 9 Å and where M is a relatively low-charged caged cation such as Na+, Ca2+ or Sr2+ (Depmeier 1988); T occupies tetrahedral site(s) and is typically Si or Al; and X is the caged anion which is either spherical (in the case of Cl-) or tetrahedral (in the case of SO42-, WO42-, and CrO42-). Therefore, stoichiometric ye'elimite is a tectoaluminosilicate sodalite with M=Ca, T=Al and X= SO4. This sodalite forms from CaO, Al2O3 and CaSO4 at around 1350ºC, a temperature at which CaSO4 decomposes. Consequently, the sulfate group is trapped in the sodalite cages, without being able to escape (Depmeier 2005). Figure 1.5 shows a perspective view of a pseudocubic sodalite cage along the (100) direction, where the sulfate group is in the center of the cage. The cage is formed by Al-O bonds and the cations (Ca2+) tent to locate in the center of the six-member rings. In general, this cages are about 4-4.5Å wide. The guest species in adjacent cages are isolated from each other and interact only weakly and indirectly (Depmeier 2005). Depending on the Si/Al ratio and the type of cage ions, different sodalite compounds can be produced with a range of interesting properties and features such as negative thermal expansion (Leardini et al. 2012), ferroelectric behaviour (Setter et al. 1984), F centers (Barrer and Coler 1968), s-electron antiferromagnetism (Monnier et al. 1994), catalytic activity (Ogura et al. 2008), tunable electronic properties (Moran et al. 1996) and optical emission (Borgmann et al. 1999), to mention just a few. Ana María Cuesta García 58 rapid development of a very fluid melt over a short range of temperatures, (iii) formation of other iron bearing phases such as monocalcium ferrite in iron-rich compositions, (iv) progressive replacement with increasing iron content of C4A3S by CaSO4 as sulfate-containing solid phase. These factors, taken together, suggest that the control of the Al/Fe ratio may well be a key parameter in optimizing sulfoaluminate clinker production. 1.2.3.2. C4AF hydration mechanisms In the absence of any other phases, the hydration of brownmillerite appears to be similar to the hydration of the iron-free C3A, in which a C-A-H gel first coats the C3A grains from which metastable hexagonal C-A-H plates (Jupe et al. 1996; Meredith et al. 2004). Finally, these plates convert to the stable cubic hydrate, C3AH6. It has been reported that iron does not form separate phases in the hydration of ferrite, but the hydration products of this phase are (more often than not) assumed to incorporate some iron (Rogers and Aldridge 1977; Fukuhara et al. 1981; Meller et al. 2004a). The hydration of pure brownmillerite with water initially forms metastable C-(A,F)-H hydrates (hydroxy-AFm), possibly in a C-(A,F)-H gel (Meredith et al. 2004). The C-(A,F)-H hydrates convert to a hydrogarnet phase C3(A,F)H6 (also called katoite) over time (Ectors et al. 2013). The exact Al/Fe ratios of the hydrogarnets are as yet debatable but these crystalline phases have a higher A/F ratio than does the unhydrated material (Meyer et al. 2004b). Thus, to maintain stoichiometry an iron-rich hydroxide must be formed (or FH3). This phase is amorphous to X-rays (there is a lack of order in the large scale), but has been detected by SEM (Rogers and Aldridge 1977) and Mossbauer spectroscopy (Teoreanu et al. 1979). The hydration of brownmillerite (Meller et al. 2004a) in the absence of sulfates may be written formally as: 1. Introduction 59 C4AF + 10H → C3AH6 + FH3 + CH [1.9] If Fe is incorporated into the hydrogarnet product by consumption of the CH produced: C4AF + 10H → (4/3)C3(A0.75F0.25)H6 + (2/3)FH3 [1.10] The addition of calcium sulfates to C4AF inhibits the direct hydration of C4AF to hydroxy-AFm or C3(A,F)H6. The hydration product that is formed rapidly on first contact with water is most commonly ettringite. There are some theories about the exact mechanism governing the retardation process (Ectors et al. 2013). In the presence of sulfate the simplest (i.e. no Fe-solid solution) scheme is (Meller et al. 2004a): C4AF + 3CSH2+ 30H → C6AS3H32 + CH + FH3 [1.11] where again C6A 3H32 is the AFt trisulfoaluminate mineral ettringite. Successively, ettringite can decompose to form an AFm monosulfoaluminate hydrate in the presence of C4AF as stated next: C6AS3H32 + C4AF → 2C4ASH12 + CSH2 + 2H + FH3 + CH [1.12] The gypsum released in [1.12] can react with any remaining C4AF to form further AFm (C4ASH12). Meller et al. (2004a) reported that small amounts of gypsum are seem to remain in the paste even after AFm forms and for this reason a transient increase in gypsum accompanies the AFt to AFm transformation (Meyer et al. 2004a). C4AF + CSH2 + 14H → C4ASH12 + FH3 + CH [1.13] Ana María Cuesta García 60 It is known that a partial series of solid solutions can be formed between C3AH6 and C3FH6; hence all the formulae in the above equations can be written as Cx(A,F)yHz or Cx(A,F)ySwHz. However, no simple crystalline iron oxides, hydroxides or hydrous reaction products are observed in any system studied (Meller et al. 2004a). Moreover, the hydration process involving iron-containing phases may be more complex, as the formation of solid solutions between Fe and Al-containing hydrates may stabilize mixed solids, such as Fe-AFt ( schner et al. 2009) and Fe-AFm (Dilnesa et al. 2012). 1.2.4. Ettringite or AFt (Ca6Al2(OH)12(SO4)3.26H2O, C6AS3H32). Ettringite is a mineral which rarely occurs in nature but is widely present in the mineralogy of hydrated cements. This phase is known in cement nomenclature as C6AS3H32 (6CaO·Al2O3·3SO3·32H2O). Ettringite, commonly named AFt, is calcium sulfoaluminate hydrate and have the general formula [Ca3(Al,Fe)(OH)6·12H2O]2·X3·nH2O, where n is, normally at least,  2 and X represents one formula unit of a doubly charged, or with reservations, two formula units of a singly charged anion. The term AFt refers to the three units of CX in an alternative way of writing the formula, C3(A,F)·3CX·yH2O [or C6(A,F)X3·yH2O], where y = x + 30. In case of ettringite, the anion X = SO42-. The crystal structure can be described as a compact columns of [Ca3Al(OH)6·24H2O]3+ composition, running parallel to the c-axis, with the X anions and, usually, H2O molecules in the intermediate channels (Taylor, 1997). It was proved that ettringite could be thermally unstable for temperatures higher than 80ºC, depending on the water pressure (Zhou and Glasser 2001). During the dehydration process, the removal of hydroxyls and water molecules from the structure occurs (Hartmann et al. 2006). 1. Introduction 61 The structural model for ettringite was proposed by Moore and Taylor (1970) by single-crystal X-ray examination with a naturally specimen. However, no consideration was given to hydrogen atoms. For this reason, the crystalline structure of ettringite was revised by Goetz-Neunhoeffer and Neubauer (2006) to obtain a more complete structure model for this phase. The symmetry of ettringite is trigonal, with a=11.229(1) Å, c=21.478(3) Å and Z=2, the space group is P31c, apparent higher symmetry being due to twinning or disorder. At the same time, Hartman and Berliner (2006) proposed a similar structure using time-of-flight neutron powder diffraction techniques. In addition, ettringite crystallizes in the form of needles or strips. Figure 1.7 shows a SEM micrograph where needles of ettringite are observed. Figure 1.7. SEM micrograph of the fracture surface of a CSA with gypsum after 1 day of hydration (Reprinted from García-Maté et al. 2015, Copyright (2015), with permission from Elsevier). Ana María Cuesta García 62 1.2.5. AFm-type phases AFm phases have a layer structure with the general formula [Ca2Al(OH)6]X·nH2O (n=8-14) where X denotes an exchangeable singly charged (e.g. OHor chloride) or half a formula unit of a doubly charged anion (for instance sulfate, carbonate and aluminosilicate) placed in the interlayer space jointly with water molecules. Some Fe(III) may also substitute aluminum. The AFm compounds crystallize in the form of hexagonal plates. The most ordinary AFmtype phases are kuzelite (Allmann 1977), Stratlingite (Rinaldi et al. 1990) and C2AH8. There are other AFm-type phases such as monocarbonates (François et al. 1998) or monochlorides (Renaudin et al. 1999) that used to appear in chemically aggressive environments. (i) The typical AFm which crystallizes in cements ([Ca2Al(OH)6](SO4)1/2·3H2O), whose formula is C4ASH12 in cement nomenclature, is called AFm-12. This phase is formed when the sulfate source (gypsum, bassanite or anhydrite) is depleted and there is enough free water available (Winnefeld and Lothenbach 2010). However, in the presence of carbonates hemiand/or monocarbonate are formed instead of C4ASH12. The crystal structure of AFm-12 was determined by Allman (1977) as it occurs naturally as the mineral kuzelite. The presence of related AFm-phases with different layer spacing is justified twofold: i) by the partial anion replacement OH-/SO42within the layers; and ii) depending on the supply of available water from the surroundings the intermediate layer contains not only anions but also a variable number of water molecules. The AFm phase has proven a difficult subject for analysis because of its low crystallinity, polytypic and changes to the position and intensity of reflections in their diffraction patterns (Matschei et al. 2007). Figure 1.8 shows a SEM micrograph of AFm. 1. Introduction 63 Figure 1.8. SEM micrograph of a BCSAF cement paste which displays AFm particles (Reprinted from Cuberos et al. 2010, Copyright (2010), with permission from American Chemical Society). (ii) Stratlingite, Ca2Al(OH)6[AlSiO2(OH)4]·3H2O, C2ASH8 in cement nomenclature, appears as hydration product of aluminium-rich cements, such as calcium aluminate, calcium sulfoaluminate and also belite calcium sulfoaluminate cements. It appears as a consequence of the reaction of C2S with AH3·nH2O. 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]-, where T can be Si or Al and □ represents a vacant tetrahedral site. This vacancies are around 45 % . The symmetry of stratlingite is R3m. The structure analysis (Kwan et al. 1995; Rinaldi et al. 1990) indicates that the octahedral layer shows an ordered scheme where each Al-octahedron is linked to 6 edge-sharing CaO7 polyhedra (2 out 3 positions are then occupied by the seven coordinated Ca-type cation). The structure also contains hydration water. Most of the water is located in the aluminum octahedral layer projecting towards the centre of the 6-membered rings of the double tetrahedral layers (Santacruz et al. 2015). Figure 1.9 shows the crystal Ana María Cuesta García 64 structure of stratlingite where selected atoms are labeled (Santacruz et al. 2015). In this work not only the structure but also the microstructure of stratlingite is discussed. Figure 1.9. Layered crystal structure of stratlingite. (Reprinted from Santacruz et al. 2015, Copyright (2015), with permission from ICE Publishing). (iii) C2AH8 is the main hydration product in CAC. It belongs to the group of AFm-type phases consisting of a main layer with the following composition [Ca2Al(OH)6·2H2O]+ and the interlayer containing the [Al(OH)4·nH2O]- groups. The interlayer has variable water content depending on temperature, relative humidity and pressure. Therefore, the formula C2AH8±X is generally used. The degree of reaction of this hydrated phase can be unravel by analyzing the first (0 0 1) peak, which correlates with the main layer distance (Raab and Pöllmann 2011). 1. Introduction 65 1.2.6. Other hydrated phases: hydrogarnet phases and amorphous aluminum hydroxide, Al(OH)3·nH2O Hydrogarnet phases have a cubic structure related to that of grossular or garnet (Ca3Al2Si3O12) 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. 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− (Dilnesa et al. 2014). 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). This katoite is reported to be the only thermodynamically stable calcium aluminate hydrate formed in calcium aluminate cements (Rivas-Mercury et al. 2007) and is a quaternary compound that crystallizes in the space group Ia3d and displays the cubic unit cell a=12.55695(3) Å and z=8 (Lager et al. 1987). Moreover, Fe-katoite (Ca3Fe2(OH)12 or C3FH6) also belongs to this group. In the mineralogical nomenclature recommended by Passaglia and Rinaldi (1984), phases in the C3AH6-C3AS3 series are collectively called hydrogrossular, and the names katoite and hibschite are used more specifically to denote those with SiO2/Al2O3 molar ratios below and above 1.5, respectively, see Figure 1.10. It has been reported an empirical equation which gives better agreement for hydrated compositions which appears in cement chemistry with the formula C3A1xFxS3-y/2Hy: a= 11.71 + 0.16Fe2O3 + 0.144H2O (1.1) where a is the cell parameter in Å and Fe2O3 and H2O are the values of x and y, respectively (Taylor 1997). In order to determine the composition of a phase of this general composition from the XRPD pattern, the intensity ratio of the 022 peak to Ana María Cuesta García 66 the 116 peak has been previously used. Table 1.6 gives some values of a and of this ratio, calculated from equation 1.1 and from the XRPD pattern respectively (Taylor 1997). Figure 1.10. Nomenclature of minerals of the hydrogarnet group Ca3(AlxFe1−x)2(SiO4)y(OH)4(3−y) (Reprinted from Dilnesa et al. 2014, Copyright (2014), with permission from Elsevier). Table 1.6. Calculated values of the cell parameters a and ratio of intensities of the 022 to the 016 XRPD powder reflections for some hydrogarnet phases (Adapted from Taylor 1997). x (moles Fe2O3 in formula) C3A1-XFXH6 C3A1-XFXSH4 C3A1-XFXS2H2 a(Å) I022/I016 a(Å) I022/I016 a(Å) I022/I016 1.0 12.73 2.22 12.45 1.77 12.16 1.36 0.5 12.65 1.15 12.37 0.78 12.08 0.43 0.0 12.57 0.41 12.29 0.17 12.00 0.03 The amorphous aluminum hydroxide, Al(OH)3·nH2O is incorrectly known as AH3, which represents its crystalline form, gibbsite. However, this crystalline gibbsite never appears as ye'elimite hydration product. Al(OH)3·nH2O is the main 1. Introduction 67 amorphous hydration product of ye'elimite containing cement and has been proposed by many researchers (Pelletier et al. 2010; Winnefeld and Barlag 2010; Winnefeld and Lothenbach 2010; Pelletier-Chaignat et al. 2011; Chen et al. 2012; Song et al. 2015). This phase is formed during the first hours of hydration along with AFt and/or AFm (see reactions [1.7] and [1.8]). However, it is very difficult to quantify because it is amorphous or ill crystalline nature. This phase occurs as an X-ray amorphous form, showing ones some very broad signals in the XRPD pattern, centered where crystalline gibbsite main reflections are placed. On the other hand, TGA analysis is a suitable tool to identify amorphous gibbsite type-gel, which loses its bound water around 250-280ºC (Winnefeld and Barlag 2010), see Figure 1.6. However, the presence of some free water molecules in the structure of aluminum hydroxide gel cannot be excluded, which could lead to an additional water loss around 100ºC during the TGA analysis (Pelletier et al. 2010). It appears that slightly less Al(OH)3·nH2O is produced in the hydration of ye'elimite with gypsum when compared to the hydration of the former in the absence of other sulfates (Winnefeld and Barlag 2010). Some studies (Wu 1995; Song et al. 2015) of the microstructure of amorphous aluminum hydroxide prepared by a special method in CSA pastes reveals that Al(OH)3·nH2O has a spherical morphology. Finally, it is important to highlight that other phases can appear during the hydration of sulfoaluminate cements. Table 1.7 lists a summary of additional hydrated phases that may be present in cement pastes. Ana María Cuesta García 74 Beamline (BL04-MSPD), the Spanish Synchrotron Radiation Facility (Barcelona, Spain). Patterns were collected in Debye-Scherrer (transmission) mode synchrotron (Fauth et al. 2013). The wavelengths, 0.61975(1) Å and 0.62020(1), were selected with a double-crystal Si (111) monochromator and determined from Si640d NIST standard (a=5.43123 Å). The diffractometer is equipped with two different detectors: i) a MYTHEN detector especially suited for time-resolved experiments and ii) a detector system based on crystal analyzers in the diffracted beam especially suited for high-resolution experiments giving also a flat background. Raw SXRPD patterns were normalized taking into account the loss of Xray beam flux with time due to the electron beam current decline in the storage ring. Normalized SXRPD patterns were analyzed by using the Rietveld methodology in order to obtain Rietveld Quantitative Phase Analysis (RQPA). Internal standard methodology has been also used with SXRPD data using quartz as internal standard. 1.3.3. Isothermal conduction calorimetry The isothermal (heat conduction) calorimetry is an efficient tool to study the stages related by the hydration of cement pastes or mortars at constant temperature. The calorimeter continuously measures and displays the heat flow related by the hydration reactions (and chemical processes in general) taking place in the cement paste after mixing. At least for a single process, the thermal power is proportional to the rate of a reaction and the produced heat is proportional to the extent of the reaction (Lawrence 2003). This methodology allows understanding the chemical origin of different regions/features (hydration reactions), the changes in reaction kinetics (Winnefeld and Barlag, 2010; Jansen et al. 2012b; Hargis et al. 2014b) and also supports the accuracy of RQPA (Hesse et al. 2011; Jansen et al. 2012a). Xray diffraction data 1. Introduction 75 on the hydration pastes allow an assignment of chemical reactions to the different peaks in heat evolution over periods of hydration approaching 250 h (Lawrence 2003). For instance, Winnefeld and Barlag (2010) studied the hydration kinetics of ye'elimite pastes in different conditions using conduction calorimetry. They observed that without gypsum, the pure ye'elimite phase showed two maxima in the calorimetry curve. The first peak appears directly when the water is added due to the very early reactions (see Figure 1.11). After that, a dormant period occurs with a low heat flow during about 10 hours. Later, it can be appreciated a second maximum after about 15 hours of hydration, this second peak covers the main part of the hydration reactions. In the ye'elimite with gypsum sample, the first peak appears to be a little more intense indicating a higher intensity of the reactions (see Figure 1.11). Figure 1.11. Conduction calorimetry of ye'elimite hydrated samples without and with different amounts of gypsum a) at short times and b) up to 18 hours (reprinted from Winnefeld and Barlag (2010), Copyright (2010), with permission from Springer). Ana María Cuesta García 76 In this work, the isothermal calorimetric study was performed in an eight channel Thermal Activity Monitor (TAM) instrument. Pastes were prepared ex-situ by mixing the sample with the appropriated amount of water. A stabilization period of 45 minutes was needed to start the measurements. 1.3.4. Thermal measurements DTA-TGA have been used to identify hydration phases in cement pastes and to determine chemically bounded water. The TGA analyses are usually performed to confirm the mineralogical observations made by XRPD. It is also possible to determine the free water content by the difference between the added water (theoretical) and the chemically combined water. Although the use of this analysis is straightforward, it allows the determination of different phases (f.i. calcium silicate hydrate, ettringite, gypsum, monosulfate, amorphous aluminum hydroxide and calcium hydroxide in a CSA cement paste) through the corresponding dehydration endothermic peaks (Telesca et al. 2014). It is important to highlight that in contrary to the XRPD, TGA is suitable to identify amorphous phases, which have associated a loss of water, such as Al(OH)3·nH2O (Winnefeld and Barlag 2010), see Figure 1.6. In this Thesis, differential thermal analysis and thermogravimetric (TGA) measurements were performed in a SDT-Q600 analyzer from TA instruments for stopped-hydration pastes. The temperature was varied from RT to 1000ºC at a heating rate of 10 ºC/min. Measurements were carried out in open platinum crucibles under nitrogen flow. 1.3.5. Scanning electron microscopy (SEM) SEM is a type of electron microscope that produces images of a sample surface by scanning it with a focused beam of relatively high-energy electrons. The 1. Introduction 77 signals that derive from the interactions between electrons and atoms of the sample contain information about the sample surface topography, chemical composition, microstructure and other properties such as electrical conductivity. The resolution depends on the probe size and the information volume that contributes to the signal, and thus on the specimen and the mode of operation. SEM–BSE (backscattered electron imaging) has been extensively used to study the microstructure of cement and cementitious materials and in combination with EDS (energy dispersive X-ray spectroscopy) to determine the elemental compositions of the hydrate assemblage (Winnefeld and Lothenbach 2010, Bizzozero et al. 2014). This technique can also allow to quantify the different phases (unreacted cement pastes and hydration products) and to follow the evolution of the porosity (Pelletier et al. 2010). A recent investigation using SEM– BSE showed that using this technique is possible to establish connections between the modeled porosity and the compressive strength (Le Saoût et al. 2013). Moreover, SEM is a good complementary technique to XRPD in order to follow the hydration product development with the time (Winnefeld and Lothenbach 2010, Chen et al. 2012). In this Thesis, prior to SEM observation, the hydration reactions of selected samples were stopped by immersing them in isopropanol for 3 days and then heated at 40 ºC for 24 h (Le Saoût et al. 2013). Microscopic characterization of samples was performed in a JEOL JSM-6490LV electron microscope. Samples were impregnated with low viscosity resin and polished down. Energy dispersive spectroscopy measurements were carried out with the OXFORD INCA Energy 350 attachment. 2. Objectives 2. Objectives 81 2. Objectives The main aim of this PhD thesis is to synthesize and characterize some of the most important phases of interest in sulfoaluminate cements and belite sulfoaluminate cements, such as dicalcium silicate, ye'elimite and tetracalcium aluminoferrite. After the initial characterization of these anhydrous phases, the study of their hydration mechanisms (including some mixtures) have been also carried out. The specific objectives can be described as follow:  To investigate the stabilization mechanisms of β and α'H-dicalcium silicate at room temperature by adding different dopants and to obtain a new revised crystal structure for α'H-dicalcium silicate.  To synthesize and characterize stoichiometric and doped ye'elimites in order to revise their crystal structures.  To understand the thermal behavior of both polymorphs of ye’elimite.  To understand the hydration mechanisms of ye'elimite as a function of its polymorphism, amount of water and type and content of sulfate source.  To study the hydration of both polymorphs of ye'elimite jointly with tetracalcium aluminoferrite or dicalcium silicate phases.  To study the hydration of tetracalcium aluminoferrite phase in different chemical environments: in the presence and absence of gypsum and under the influence of ye’elimite.  To understand and characterize the amorphous gibbsite (or aluminum oxide hydrated) that is formed in these hydration reactions. 3. Articles section Ana María Cuesta García a#6: “Ye’elimite: structures and hydration mechanisms” Cuesta A, Garcia-Mate M, Leon-Reina L, De la Torre A.G, Santacruz I, Aranda M. A. G (2015), Proceedings of the 14th International Congress on the Chemistry of Cement, Beijing, Spain. Abstract Belite calcium sulfoaluminate, aka BCSA or sulfobelite, cements are environmentally friendly building materials, as their production may have up to 25% lower CO2 footprint than OPC fabrication. They are prepared by mixing the clinker with different amounts of a calcium sulfate set regulator. Hereafter cement nomenclature will be used: C=CaO, S=SiO2, A=Al2O3, F=Fe2O3, S=SO3, T=TiO2 and H=H2O. BCSA cements are based on belite (C2S), ye’elimite (also called Klein’s salt or tetracalcium trialuminate sulfate ( C4A3S)) and other minor phases, such as ferrite (C4AF) or calcium aluminates (C12A7). Additionally, calcium sulfoaluminate cements, CSA, contain ye’elimite as the main phase with belite being the second phase in terms of content. The phase composition variability in CSA cements is much larger than in BCSA cements. During early age hydration, i.e. up to 3 days, ettringite phase is the main crystalline hydration product formed in CSA and BCSA cement pastes arising from the dissolution and reaction of ye’elimite and calcium sulfate(s). Subsequently, and mainly in BCSA pastes, belite may yield stratlingite, amorphous C-S-H gel and portlandite. Minor phases present in the anhydrous cement may provoke the precipitation of other hydration products such as monosulfoaluminate, hydrogarnet or katoite. The reactivity of ferrite during hydration in these cements is slower than that of C4A3S and it is still not well understood. Stoichiometric and doped ye’elimite crystal structures, including their polymorphisms, with temperature, have been studied and reported very recently. Furthermore, the reactivity of stoichiometric and doped ye’elimite has also very recently reported by us. The aim and originality of this work is to better understand the hydration of stoichiometric (orthorhombic) and doped (pseudo-cubic) ye’elimite phases in combination with belite crystallized as both polymorphs (’H and  at different Articles section ages to improve “ecocement” performances. This work has allowed establishing kinetics and mechanisms for hydration of ye’elimite with belite samples using exsitu laboratory X-ray powder diffraction (LXRPD) in combination with the external standard method, G-factor, up to 6 months. This strategy has allowed quantifying the amorphous content, including free water (the latter determined through thermogravimetric measurements (TGA), for stopped-hydration pastes). It is known that active BCSA cements which contain the ’H-C2S polymorphs have a faster hydration reaction than related cements containing the’H-C2S polymorph. However, it is essential to establish t he role of both ye’elimite polymorphs in the hydration behavior of these mixtures. Here, we study all the combinations between belite and ye’elimite. To simplify the problem, we prepare these artificial mixtures without aluminates. Powder diffraction data at 56 days show some small differences between samples. Chiefly, the sample which contains both ’H-C2S and doped-ye’elimite presents a higher degree of reaction and higher amounts of stratlingite at that age. Moreover, the hydration of tetracalcium aluminoferrite (ferrite) in combination with ye’elimite (stoichiometric and doped) has also been studied by in-situ SXRPD (synchrotron X-ray powder diffraction) to understand the dissolution/crystallization processes that take place during hydration processes at early-ages. The main difference after 46 hours is that the sample with doped-ye’elimite yields much larger relative amounts of AFt. Ana María Cuesta García a#7: “Hydration of C4AF in the presence of other phases: a synchrotron X-ray powder diffraction study” Cuesta A, Santacruz I, Sanfélix S G, Fauth F, Aranda M G A and De la Torre A G (2015), Constr Build Mater, in press. doi: 10.1016/j.conbuildmat.2015.10.114 Abstract Hydration behaviour of C4AF in selected experimental conditions has been determined. C4AF has been hydrated in the absence and presence of gypsum, two polymorphs of ye’elimite and different water/solid ratios. C4AF in the presence of water hydrates to form mainly a hydrogarnet-type phase. The crystal structure of C3A0.845F0.155H6 is reported from the Rietveld analysis of its synchrotron X-ray powder diffraction pattern. The hydration of C4AF in the presence of gypsum gives AFt. However, the mixture tetracalcium aluminoferrite/gypsum/ye’elimite gives both AFt and AFm phases. C4AF hydrated with ye’elimite i n the absence of gypsum gives onl y AFm. Ye’elimite has inhibited tetracalcium aluminoferrite hydration. 4. Results and discussion 4. Results and discussion 4. General results and discussion This PhD Thesis is focused on the preparation and characterization of single phases of interest in calcium sulfoaluminate cements. After the characterization of the anhydrous samples, the hydration mechanisms of some pure samples and mixtures pastes have been profoundly studied. This work has been organized in two main blocks: i) structural studies of anhydrous phases such as β and α'H dicalcium silicate; stoichiometric and doped ye'elimite and finally a characterization of tetracalcium aluminoferrite and ii) hydration studies of selected phases and mixtures. The writing up of this Thesis has ended by July 20th, 2015. 4.1. Dicalcium silicate. 4.1.1. Synthesis and sample characterization of dicalcium silicate. Stoichiometric C2S has five polymorphs, see Figure 1.2, (Mumme et al. 1995), γ, β, α'L, α'H and α. An important aim of this thesis has been to prepare and to study high temperature polymorphs of dicalcium silicate metastabilized at room temperature. Some solid solutions were tested for the stabilization of the high temperature polymorphs in the dicalcium silicate samples by using different dopants (articles #1 and #2). However, some compositions were not achieved. Firstly, Ca2-xAlxSi1-xAl2xO4 and Ca2Si1-2xAl2xO4-xx compositions were tested in order to stabilize the β polymorph (article #1) using aluminum as dopant. It was reported that for tricalcium silicate, C3S, (Porras-Vázquez et al. 2007) the aluminum doping was achieved by substitution of calcium and silicon sites by 85 Ana María Cuesta García aluminum, Ca3-x/2Alx/2(Si1-x/2Alx/2O4)O. However, for the C2S samples, the similar mechanism was not achieved as all attempts to synthesize Ca2-xAlxSi1-xAl2xO4 series were not successful. Nevertheless, Ca2Si1-2xAl2xO4-xx solid solution was successfully prepared. Table 1 in article #1 gives weight fractions of all phases as well as the refined unit cell volumes of the series. The volume increases along the series as expected, as Al3+ has a larger ionic radius than Si4+. So, this result confirms the partial substitutions of silicon by aluminum. In addition, the limit of the series has to be close to x=0.014 because the x=0.030 sample presented some amount of C3A phase. Then, for stabilizing the α'H-polymorph (article #2), boron oxide and sodium oxide were used as dopants due to the fact that previous studies reported that borax stabilized α'H-form in a cement matrix (Cuberos et al. 2010). The first issue to be clarified in the following boron-containing solid solutions was to unravel if the borate group was tetrahedral BO45or planar-triangular BO33-. Firstly, Ca2-xNax(SiO4)1-x(BO3)x solid solution (x=0.00-0.25) was successfully prepared. This series was designed for stabilizing dicalcium silicate with planar BO33anions. Figure 3b (in article #2), shows the intensities of their vibration bands, close to 1250 cm-1 that confirmed the presence of BO33units in this sample. Moreover, Table 2 (in article #2) gives RQPA of this series. The first result shown in this table is the absence of free calcium oxide along this series. A second very relevant result is that high temperature polymorphs of C2S, i.e. βand α'H-C2S, are stabilized as x increases, as expected. For instance, β-C2S was present between x=0.00 and x=0.20. Otherwise, α'H-C2S appeared from x=0.025 up to x=0.25, being the last composition a single phase. Finally, it can be seen that samples with larger x values presented higher FWHM values (Figure 7 in article #2). This behavior may be due to small (variable) particle sizes and/or microstrain evolution. Moreover, Ca2-xBx(SiO4)1-x(BO4)x (0≤x≤0.20) series was also prepared. In this case only boron oxide was added as dopant. This mechanism was based on 86 4. Results and discussion substitution of tetrahedral SiO44groups by tetrahedral BO45units with the boron substitution of calcium for charge compensation. ATR-FTIR infrared spectra were also performed and Figure 3a (article #2) shows the ATR-FTIR infrared spectra of all the members of this series (up to x=0.10). These measurements show bands close to 1000 cm-1 which correspond to BO45stretching vibrations. Furthermore, tiny signals from 1150 to 1350 cm-1, corresponding to BO33- (or similar) units, are also observed (Heyns et al. 1990, Tukia et al. 2005). However, this could be due to the fact that BO33units can be locally formed where boron cations substitute calcium atoms as boron is much smaller than calcium. In addition, a large displacement within the site was expected which might have lead to the formation of a BO33--type groups. Taking all these results into account, it can be affirmed that boron stabilizes high temperature polymorphs by substituting silicon units as BO45and calcium cations by B3+ at the same time. Moreover, Table 2 (in article #2) gives RQPA for all members. Finally, Ca2-x/2x/2(SiO4)1-x(BO3)x, Ca2(SiO4)1-x(BO3)xOx/2 and Ca2-3xB2xNax(SiO4)1-x(BO4)x compositions were also tested in order to check other possible substitution mechanisms for the case of boron and, in some cases, sodium. However, all members of these series contained large amounts of free calcium oxide and/or other secondary phases which indicated that these solid solutions were not formed (see Table 2 in article #2). Ca2Si0.972Al0.028O3.9860.014 and Ca1.85Na0.15(SiO4)0.85(BO3)0.15 were selected to further study of β and α'H polymorphs, respectively. On the one hand, β-C2S with formula (article #1), Ca2Si0.972Al0.028O3.9860.014, was synthesized by sintering reaction using high purity oxides and carbonates as starting materials: CaCO3 (99.95%-100.05% from Alfa Aesar), SiO2 (99.7% from ABCR) and γ-Al2O3 (99.999% from Alfa Aesar). Raw mixtures were ground during 30 min in a planetary ball mill and preheated at 1050ºC for 4 hours. The resulting powders were reground, pelletized and a second 87 Ana María Cuesta García thermal treatment was carried out at 1500ºC for 6 hours before being slowly cooled. On the other hand, α'H-C2S, Ca1.85Na0.15(SiO4)0.85(BO3)0.15, was prepared using a similar experimental procedure (article #2). Dopants were added as Na2B4O7·10H2O (100% from Aldrich) and Na2CO3 (99.999% from Aldrich). Raw mixtures were ground for 10 min in an agate mortar, and preheated at 1000ºC for 4 hours. After cooling, the mixtures were milled during 1 h in a planetary ball mill. The resulting powders were pelletized and a second thermal treatment was carried out at 1300ºC for 30 min and quenched from high temperature with an air flow. In a second stage, a similar synthetic method was employed to obtain 100 g of each sample for the hydration studies. In this case, the raw material was ground in a Micro-deval machine with a cylinder container and steel balls and was pelletized (600 mm diameter and 1000 MPa). The heating and cooling conditions were those described above. The Rietveld quantitative phase analyses for both samples are shown in Table 4.1. It is important to underline that these thermo-mechanical treatments at high temperature ensured the absence of free lime, CaO, showing that both reactions were complete. After this initial characterization it is important to confirm that these solid solutions have been properly formed. Table 4.1. Rietveld quantitative phase analyses for Ca1.85Na0.15(SiO4)0.85(BO3)0.15 and Ca2Si0.972Al0.028O3.9860.014 γ-C2S(wt%) β-C2S(wt%) α'H-C2S(wt%) Ca1.85Na0.15(SiO4)0.85(BO3)0.15 - 9.9(2) 90.1(1) Ca 2 Si 0.972 Al 0.028 O 3.986  0.014 1.0(1) 99.0(1) - The β-form was synthesized with nominal formula Ca2Si0.972Al0.028O3.9860.014 where  stands for vacancies in the structure. The most 88 4. Results and discussion important point was to check the presence of those vacancies. In order to do that, both samples Ca2Si0.972Al0.028O3.9860.014 and Ca1.85Na0.15(SiO4)0.85(BO3)0.15, the later without vacancies, were studied by thermogravimetric analysis to evaluate the influence of Al-substitution. In this study, the increase in vacancy concentration due to the incorporation of water in the structure was checked. Water uptake depends on the available oxygen vacancies in the structure and the differences are expected depending on the vacancy contents. The thermogravimetric measurements recorded under humidified air showed reproducible curves on both cooling and heating cycles, see Figure 4.1. The curve of Ca2Si0.972Al0.028O3.9860.014 shows the typical behavior of a proton conducting material with weight increase upon cooling due to water uptake and the formation of protonic defects. For these samples the water uptake increases with the oxygen vacancies content from 0.09 wt% for Ca1.85Na0.15(SiO4)0.85(BO3)0.15 (no vacancies) to 0.22 wt% for Ca2Si0.972Al0.028O3.9860.014 due to the large concentration of oxygen vacancies. Consequently, these results confirmed that Si was being substituted by Al in the dicalcium silicate framework and that this substitution induced the presence of oxygen vacancies. Finally, an impedance spectroscopy study was carried out for the Ca2Si0.972Al0.028O3.9860.014 sample. This study finally confirmed the existence of proton conductivity in this material due to the presence oxygen vacancies in comparison with Ca1.85Na0.15(SiO4)0.85(BO3)0.15, as standard (no vacancies). See Figures 6 and 7 in article #1. 89 Ana María Cuesta García data for the stoichiometric sample, Ca4[Al6O12]SO4. Firstly, as the recorded pattern clearly contains splitted diffraction peaks the cubic sodalite structure was not tested here. Then, the tetragonal structural description reported by Peixing et al. (1992) was used but the fits were so poor that this symmetry choice had to be dismissed. For this reason, orthorrombic crystal structure description reported by Calos et al. (1995) was finally used as a starting model for the Rietveld refinement. Table 4.3. gives the starting and final disagreement values for the joint Rietveld refinements. Moreover, Figure 1 in article #3 shows the final LXRPD and NPD Rietveld plots using the refined orthorhombic structure. Finally, the following unit cell parameters were obtained, a=13.0356(7) Å, b=13.0350(7) Å and c=9.1677(2) Å, yielding a cell volume of 1557.78(6) Å3. Table 4.4 shows the final refined atomic parameters and the isotropic ADPs which were refined independently. Table 4.3. Disagreement factors for the Rietveld refinements of LXRPD, NPD, SXRPD and HT-LXRPD patterns for stoichiometric and doped ye'elimite samples by using previously-reported structures and after atomic parameter refinement. Orthorhombic Structure (a) Cubic Structure (b) Revised Structure T/ºC R WP /% R F /% R WP /% R F /% R WP /% R F /% Stoichiometric(LXRPD)* 26 17.1 14.6 13.6 5.5 Stoichiometric (NPD)* 26 6.5 8.3 5.4 4.7 Doped (SXRPD) 26 15.4 16.9 11.0 7.9 Stoichiometric (LXRPD) 800 11.4 6.5 9.9 3.2 Doped (LXRPD) 800 6.6 4.4 5.8 3.0 (a) Calos et al. 1995; (b) Kurokawa et al. 2004 *joint refinement It is important to bear in mind that this study was corroborated by using atomistic calculations and several conclusions could be deduced. Firstly, the most stable found phase was the orthorhombic Pcc2 structure, in agreement with our experimental results (see article #3). Secondly, the ideal cubic structure is less stable than the orthorhombic phase at room temperature. Finally, it is possible that a pressure-induced transition of ye'elimite from orthorhombic to cubic polymorph 96 4. Results and discussion may occur. The phase transition with temperature for ye'elimite has been also studied in this work and it will be discussed in the next section. Table 4.4. Refined positional coordinates and isotropic ADPs for stoichiometric ye'elimite at room temperature in the Pcc2 space group. Atom x y z ADPs/ Å2 S1 0.2691(7) 0.2693(7) 0.9654(10) 0.028(3) O1 0.6790(12) 0.2645(16) 0.6111(12) 0.099(8) O2 0.8260(8) 0.2042(9) 0.4666(15) 0.033(4) O3 0.7590(15) 0.3781(7) 0.4296(22) 0.103(8) O4 0.6598(10) 0.2280(14) 0.3507(17) 0.073(7) Ca1 0.0627(11) 0.2495(14) 0.1554(17) 0.020(4) Ca2 0.2549(14) 0.9417(12) 0.2945(16) 0.011(4) Ca3 0.2591(16) 0.5301(11) 0.2196(13) 0.035(4) Ca4 0.4910(9) 0.2548(13) 0.2196(13) 0.013(3) Al1 0.5000 0.5000 0.2242(15) 0.005(4) Al2 0.5000 0.0000 0.2466(12) 0.016(8) Al3 0.0000 0.0000 0.2292(15) 0.021(5) Al4 0.0000 0.5000 0.2011(11) 0.026(8) Al5 0.6269(7) 0.1202(7) 0.0036(10) 0.005(5) Al6 0.1238(6) 0.6282(7) 0.9519(8) 0.007(6) Al7 0.3708(6) 0.6293(7) 0.9760(13) 0.012(3) Al8 0.1233(7) 0.8793(7) 0.9787(13) -0.001(3) O5 0.3983(10) 0.2439(6) 0.4338(11) 0.027(6) O6 0.5628(6) 0.4028(5) 0.1320(13) 0.013(5) O7 0.5457(8) 0.1035(5) 0.1446(11) 0.032(5) O8 0.7546(5) 0.5910(9) 0.5102(9) -0.004(3) O9 0.5957(5) 0.4375(6) 0.8201(13) 0.011(5) O10 0.8984(5) 0.4459(7) 0.7976(8) 0.08(3) O11 0.7539(6) 0.9027(10) 0.5520(10) 0.018(4) O12 0.8916(5) 0.0345(7) 0.8323(14) 0.013(4) O13 0.6069(5) 0.0378(8) 0.8490(11) 0.024(5) O14 0.1050(11) 0.2433(6) 0.4026(11) 0.027(6) O15 0.9646(7) 0.1066(5) 0.1224(14) 0.014(4) O16 0.9571(7) 0.3947(5) 0.0996(8) -0.001(3) For the doped sample, Ca3.8Na0.2Al5.6Fe0.2Si0.2O12SO4, the cubic crystal structure recently reported for stoichiometric ye'elimite at 800ºC (Kurokawa et al. 97 Ana María Cuesta García 2014), space group I4 �3m, was used as a starting model for the Rietveld refinement due to the fact that the recorded pattern clearly contains the diffraction reflections of the cubic sodalite structure. The cubic crystal structure reported by Saalfeld and Depmeier (1972) was previously tested but was rejected since the disagreement values were much higher than those obtained with the structure reported by Kurokawa et al. (2014), see Table 2 in article #4. SXRPD data, obtained in the Xray powder diffraction station of ALBA with a MAD detector system, which is especially suited for high-resolution experiments, were used to revise the structure of the doped sample. The final Rietveld disagreement factors of SXRPD data of doped ye'elimite are also shown in Table 4.3. Moreover, Figure 1 in article #4 shows the SXRPD Rietveld plot using the refined cubic structure. For the doped ye'elimite, the refined unit cell parameter was a=9.1970(1) Å and the cell volume was 777.93(2) Å3. The anisotropic ADPs were also optimized. Table 4.5 shows the refined atomic parameters (positional coordinates and anisotropic ADPs) for the RT cubic structure of doped ye'elimite. It is important to point out that foreign elements were excluded from the crystal structural study as they did not make any improvement in the refinement due to their high degree of disorder. Table 4.5. Final (refined) atomic parameters (positional coordinates and anisotropic ADPs) for doped ye'elimite at room temperature. Atom Ca1 Al1 S1* O1 O2* Wyckoff position 8c 12d 2a 24g 24g occupation factor 1 1 1 1 0.3333 x 0.2944(1) 0.2500 0.0000 0.3479(2) 0.1129(5) y 0.2944 (1) 0.5000 0.0000 0.3479(2) 0.1129(5) z 0.2944 (1) 0.0000 0.0000 0.0533(3) 0.0136(16) u11 0.096(1) 0.007(1) 0.087(2) 0.028(1) 0.139(7) u22 0.096(1) 0.015(1) 0.028(1) u33 0.096(1) 0.015(1) 0.034(2) u12 0.069(2) 0.000 0.009(1) u13 0.069(2) 0.000 0.007(1) u23 0.069(2) 0.000 0.007(1) *isotropic displacement parameters 98 4. Results and discussion It is important to highlight that there are some issues which point towards a pseudocubic structure for this sample instead of cubic. That fact will be better explained with the termodiffractometric study. 4.2.3. Thermodiffractometric study and phase transition A thermodiffractometric study, up to 800ºC, has been performed for both ye'elimites. Firstly, for stoichiometric ye'elimite, we have studied in detail the lowto high-symmetry phase transition on heating previously reported by Depmeier (1983) without specific details. Figure 4.5 gives a selected view of the patterns at different temperatures. This figure clearly shows the transition of the orthorhombic form to a higher symmetry polymorph the latter being cubic above 400ºC. The unit cell parameter behavior observed during the phase transition, given in Figure 7 in article #3, is that expected for the evolution of a lower-symmetry phase due to long-range sulfate ordering to a higher-symmetry phase where sulfate anions display disorder, likely due to rotation. Furthermore, a thermodiffractometric study for doped ye'elemite was also performed (article #4) up to 800ºC. It is important to have in mind that doped ye'elimite is cubic at room temperature and for this reason no phase transition would be expected. Moreover, all attempts to index the low temperature patterns with symmetry lower than cubic were unsuccessful as the observed splitted peaks at high angles were very small and only observable for some diffraction reflections. Figure 4.6 shows a selected range of the high temperature LXRPD patterns of doped ye'elimite. However, although no phase transition was observed, Figure 4.6 display the sharpening of one of the cubic reflection (622) with temperature as an example. Moreover, changes in the widths of the diffraction peaks with temperature were clearly measured and were compared with silicon as cubic standard (see Figure 3, article #4). Those results showed that variation of FWHM of silicon peaks was negligible while the FWHM values of ye'elimite peaks were 99 Ana María Cuesta García decreasing with increasing temperature. This behavior indicates that the symmetry of the RT form has to be lower than cubic, i.e. pseudocubic. Figure 4.5. Temperature-dependence LXRPD (λ=1.5406 Å) patterns for stoichiometric ye'elimite shown over a selected range with Miller indexes displayed for the orthorhombic (bottom) and cubic (top) phases. Both ye'elimites have also been studied by DSC and permittivity measurements. The phase transition for stoichiometric ye'elimite takes place close to 470 ºC on cooling determined by DSC and permittivity studies. These results agree quite well with the reported value of 464ºC (Depmeier 1988). Figure 4.7 shows a comparative of the permittivity study for both ye'elimites. The permittivity 38.5 39.0 39.5 40.0 40.5 41.0 41.5 42.0 200º C 300º C 400º C 100 ºC 450ºC 500ºC Orthorhombic: √2 a c × √2 a c ×a c Cubic: a c ×a c ×a c 004 411 330 440 004 600 060 333 204 351 024 104 014 502432 I (a.u.) 2θ(º) 30ºC 100 4. Results and discussion curve for doped ye'elimite shows a phase transition close to 525ºC, Figure 4.7. This result was not expected if the RT form is truly cubic. However, this signal in the DSC curve shows that RT form has a symmetry lower than cubic, i.e. pseudocubic, in full agreement with HT-LXRPD study. Figure 4.6. Selected region of the HT-LXRPD (λ=1.5406 Å) data for doped ye'elimite. Stars indicate silicon reflections. Inset details the pseudocubic 622 reflection. As a summary, this study has proved that stoichiometric ye'elimite undergoes a cubic-to-orthorhombic phase transition at 470ºC (on cooling) likely due to the freezing and spatial ordering of the sulfate anions which can be rotating 20 30 40 50 60 70 º/2θ I (a.u.) 30ºC 100ºC 200ºC 300ºC 400ºC 500ºC 600ºC 700ºC 800ºC 30ºC 100ºC 200ºC 300ºC 400ºC 500ºC 600ºC 700ºC 800ºC ( ) FWHM: 0.199º FWHM: 0.130º FWHM: 0.116º FWHM: 0.104º 101 Ana María Cuesta García above the transition temperature. Moreover, doped ye'elimite undergoes a related cubic-to-pseudocubic phase transition at 525ºC also likely due to the freezing of the sulfate anions. However, the long-range ordering of the sulfate anions does take place, due to the presence of the dopants, avoiding the orthorhombic structure to develop. The difference in phase transition temperature between both polymorphs is likely due to the effect of the dopants which stabilize the static disorder of the sulfate anions within the cages. Figure 4.7. Permittivity signal for stoichiometric ye'elemite and doped ye'elimite. 4.2.4. HT-polymorphs of ye'elimite: structural study. Both ye'elimites were thoroughly studied at 800ºC (article #4). Firstly, the pattern at 800ºC of stoichiometric ye'elimite was auto-indexed using DICVOL06 (Boultif and Louer 2004) in a cubic unit cell. The cubic structure was also verified by fitting the powder pattern with the cubic structure recently reported by Kurokawa et al. (2014). Table 4.3 shows the agreement factors obtained using both 300 400 500 600 700 19.6 20.0 20.4 20.8 21.2 21.6 22.0 525ºC T (oC) ε’ Stoichiometric Doped 470ºC 102 4. Results and discussion the published and refined structures. The final unit cell parameter obtained was a=9.2497(1) Å. The 800ºC powder pattern for doped ye'elimite was also autoindexed using DICVOL06 (Boultif and Louer 2004) in a cubic unit cell. Figures of merit (De Wolff 1972) for this indexation were much higher than those obtained at RT (Table 1 in article #4). This is another indication that the HT-form is truly cubic. A Rietveld refinement of the 800ºC pattern of doped ye'elimite was performed using the previously reported split-atom model for the sulfate anion (Kurokawa et al. 2014). The refinement in space group I4 �3m was very good and a proof of this fact was the low R-factors (see Table 4.3) and the flatness of the difference curve of the Rietveld plot (Figure 5 in article #4). The final refined unit cell parameter at 800ºC was a=9.2544(1) Å. 4.3. Synthesis of C4AF. Tetracalcium aluminoferrite (C4AF) was prepared by mixing suitable amounts of CaCO3 (99.95%, Alfa Aesar), Al2O3 (99.997%, Alfa Aesar) and Fe2O3 (99.945%, Alfa Aesar), to obtain approximately 5 g of sample. The raw mixture was ground for 5 minutes in an agate mortar and heated at 1000ºC for 4 hours. After that, the powder was ground in an agate mortar and was pelletized. Finally, the pellets were heated at 1350ºC for 4 hours (heating rate of 10 ºC/min) followed by quenching from high temperature with an air flow (article #7). This first synthesis of 5 grams of C4AF was studied by LXRPD and the Rietveld method. The analysis of the sample was of 95.2(3) wt% of C4AF with minor amounts of two crystalline phases: 2.6(3) wt% of Fe2O3 (hematite) and 2.2(3) wt% of Ca12Al14O33 (C12A7). The derived unit cell parameters for C4AF were: a=5.5687(2) Å, b=14.5250(7) Å, c=5.3500(2) Å and V=432.73(4) Å3. 103 Ana María Cuesta García In a second stage, this synthetic method was employed to obtain 50 g of sample for the hydration studies. In this case, the powder was ground in a Microdeval machine with a cylinder container and steel balls for 45 min and was pelletized (600 mm diameter and 1000 MPa). The heating and cooling conditions were the same just described. This sample was studied by SXRPD and the Rietveld method. For this synthesis, the only additional phase was C3A, 1.94(6) wt%. The derived unit cell parameters for C4AF from SXRPD, were: a=5.56638(4)Å, b=14.5227(1) Å, c=5.34835(4) Å and V=432.354(8) Å3. Figure 4.8 shows the Rietveld plot for this sample. As the quality of the SXRPD was very high, we could optimize the Fe/Al ratio in the octahedral and tetrahedral sites (constrained to an overall Fe/Al ratio of 1.00). The resulting value was 0.746(2) of Fe occupancy at the octahedral site and therefore 0.254(2) of Fe occupancy at the tetrahedral site. Figure 4.8. SXRPD Rietveld plot for C4AF (λ=0.6202(2) Å). The tic marks are the allowed Bragg reflections. C4AF: lower row, Quartz (internal standard): middle row and C3A: upper row. We are aware that different cooling rates could change Fe/Al distributions at these sites but to study the possible consequences of this variation on the hydration properties of C4AF is not a target of this Thesis. C 3 A I (a.u.) 2θ(º) 104 4. Results and discussion It is important to point out that the SXRPD Rietveld plot of the synthetic C4AF (Figure 4.8) shows narrower reflections than C4AF present in a Portland clinker (De la Torre, 2003). This fact indicates that the tetracalcium aluminoferrite found in a clinker matrix has smaller particle sizes. Moreover, the Fe/Al ratio of this phase can also be different in clinkers. 4.4. Application of the structural studies to the analysis of CSA and BCSA cements. After the study of these revised crystal structures, dicalcium silicate and ye'elimite, they were used to perform RQPA of cements in order to prove the importance of choosing the best crystal structure description for each component. Firstly, the refined orthorhombic structure of ye'elimite was used to describe a commercial CSA cement which contained large amounts of this phase. The RF values from the Rietveld refinement of ye'elimite phase in CSA were 6.9% and 4.8% for the reported orthorhombic structure (Calos et al. 1995) and for the new revised structure, respectively. Moreover, the disagreement value for the whole pattern was 8.7% using the Calos et al. (1995) structure and 7.6% for the structure reported in this Thesis. Lower RF values indicates that the structure description used fits better the experimental patterns yielding to more accurate results (article #3). Secondly, the combination of the structural descriptions of α'H-C2S and doped-ye'elimite revised in this PhD Thesis were used to quantify these phases in an active BCSAF clinker (Álvarez-Pinazo et al. 2012) doped with 2 wt% B2O3, added as borax. This study has not been previously published and it is original of this Thesis. Only the revised structure of α'H-C2S was used to perform the RQPA of an active BCSAF clinker in article #2. This clinker contains three main phases, α'H105 Ana María Cuesta García Table 4.7. Quantitative phase analysis results (wt%) for st-C4A3S-1.16 paste, as a function of hydration time obtained by SXRPD* and LXRPD#. to* 12h* 18h* 24h* 31h* 2d# 7d# st-C4A3S 44.9(1) 43.0(1) 29.4(1) 10.3(3) 9.2(3) 4.6(2) 2.9(2) AFt - - 2.3(1) 1.1(2) 1.0(2) 9.2(3) 6.8(3) AFm$ - - 7.1(1) 22.0(2) 22.2(2) 25.1(4) 25.8(5) ACn +FW 1.4(1) +53.7**=55.1 57.0(1) 61.2(1) 66.5(2) 67.6(2) 44.1(5) +17***=61.1 46.5(6) +17.9***=64.4 **Theoretical free water content *** Obtained from the TGA study $SXRPD patters were fitted with one AFm (Allmann 1977); In LXRPD pattern fit the value is the total of the three phases. It is important to bear in mind that stopping procedure has affected the mineralogical composition, especially AFm-type phases. For the in-situ synchrotron experiment, the hydration was not stopped and higher crystallinity of AFm was found. It was fitted by using the crystal structure of AFm reported by Allmann (1977) and refining unit cell parameters. On the other hand, as the ex-situ LXRPD data were collected for stopped-hydration samples, broader peaks due to AFm-type phases were observed (see stars in Figure 4.11). In order to fit those broader peaks, the structure reported by Allmann (1977) was introduced three times in the GSAS experimental file and c values were modified. This was due to the lack of different crystal structures published for AFm-type phases. Table 4.7 includes the ACn values obtained from internal and external standard methodologies of the mixture of st-C4A3S with w/s=1.16. The first column gives t0 values obtained from the SXRPD pattern of the anhydrous sample. FW in t0 column is the theoretical value. Remaining values obtained from internal standard method encompass not only ACn but also FW (not chemically bound water) and are expressed as a single value. This is due to the inability of the internal standard methodology to distinguish between different not-diffracting phases. On the other hand, data obtained with G-factor methodology from LXRPD, 2 days and 7 days, corresponds only to ACn values, since FW was removed by the stopping hydration procedure. The FW contents were determined by the difference between the ‘theoretical/mixed’ water and the combined water determined from 112 4. Results and discussion TGA study (from RT to 600ºC) of the stopped-hydration sample. It is important to highlight that the results obtained by the internal standard method are in agreement with those obtained at later ages with the external standard showing the consistence of both methodologies. Secondly, the hydration mechanism for doped or solid solution ye'elimite (ssye'elimite) has been also studied with a w/s of 1.16. Table 4.8 shows RQPA results at the measured ages. It is important to know that the results reported in the article #5 were obtained using the crystal structure for doped ye'elimite published by Saalfeld and Depmeier (1972) since this study was published before obtaining a revised crystal structure for the doped polymorph. However, RQPA results have been updated in this Thesis by using the revised pseudocubic ye'elimite structure discussed in the previous section. The main difference has been that the amorphous contents have decreased due to the better Rietveld fits and hence the RQPA results are more accurate with the revised crystal structure. Table 4.8. Quantitative phase analysis results (wt%) for ss-C4A3S-1.16 paste, as a function of hydration time obtained by SXRPD* and LXRPD#. to* 6h* 8h* 12h* 15h* 18h* 24h* 2d# 7d# ssC4A3S 45.6(1) 45.1(1) 44.3(1) 12.2(2) 6.1(2) 3.9(3) 1.9(2) 0.8(1) 0 AFt - 0.6(2) 0.9(2) 5.6(2) 6.4(3) 6.8(3) 7.0(3) 14.7(1) 12.7(1) AFm$ - - - 13.6(2) 16.5(2) 18.8(2) 18.5(2) 13.4(1) 14.3(1) ACn +FW 0.7(2) +53.7**=54.4 54.4(2) 54.8(2) 68.7(2) 71.0(2) 71.4(2) 72.6(2) 58.0(5) +13.1***=71.1 61.0(5) +12***=73.0 **Theoretical free water content *** Obtained from the TGA study $SXRPD patters were fitted with one AFm (Allmann 1977); In LXRPD pattern fit the value is the total of the three phases. Comparing Tables 4.7 and 4.8 and observing Figure 4.10, it can be stated that solid-solution ye'elimite reacts at a faster pace as more than 13 wt% of AFm and 5 wt% of AFt are quantify after 12 hours while no hydration phase can be found in st-ye'elimite at the same time. Moreover, the increase of w/s ratio affected the hydration of st-ye'elimite by promoting the formation of AFm. As an example, Figure 4.12 a and b shows Rietveld plots of st-C4A3S-1.16 and ss-C4A3S-1.16, 113 Ana María Cuesta García respectively, at 12 hours of hydration in order to show the big differences in the reaction degree. In addition to a faster kinetics, solid-solution ye'elimite yielded much larger relative amounts of AFt. Observing Figure 4 in article #5, AFt crystallization in ss-C4A3S is higher than in the stoichiometric sample. Moreover, ~17% and ~10% of hydrated sulfate groups for st-C4A3S and ss-C4A3S, respectively were mainly incorporated into ACn phase(s) and/or in pore solution. Figure 4.12. SXRPD (λ=0.61975(1) Å) Rietveld plots for a) st-C4A3S-1.16 and b) ssC4A3S-1.16, at 12 hours of hydration, with the main peaks due to a given phase labeled. 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 I (a.u.) 2θ(º) (a) (b) C 4 A 3 S Quartz AFm AFt 114 4. Results and discussion It is important to highlight that calorimetric data, showed that more heat was released by the solid-solution sample (577 J/g at 7 days) in comparison to stoichiometric ye'elimite (555 J/g), due to a larger reaction degree and also to a larger relative amount of crystallized ettringite at seven days. In summary, comparing both polymorphs, some findings should be highlighted: i) stoichiometric ye'elimite reacts slower than solid-solution ye'elimite and ii) solid-solution ye'elimite produces higher amounts of ettringite than stoichiometric ye'elimite in similar hydrating conditions. Moreover, the increase of w/s ratio affects the hydration of st-ye'elimite by promoting the formation of AFm. 4.5.2. Reactivity of ye'elimites with gypsum. The hydration of ye'elimite, stoichiometric and solid solution, with gypsum was studied by in-situ SXRPD at very early ages because the presence of gypsum accelerates the hydration reaction of ye'elimite. In order to understand the role of polymorphism in the hydration mechanism, st-C4A3S-g-1.42 and ss-C4A3S-g-1.42 mixtures were prepared in which each polymorph was mixed with gypsum in the molar ratio of reaction [1.7] and water was added in excess. For both pastes, the reaction almost finished after 14 hours of hydration. RQPA results obtained from SXRPD are given in Tables 4.9 and 4.10. It can be observed that the overall reactivity of these two samples has followed reaction [1.7]. AFt is the only crystalline hydrated product, consequently there should be an amorphous phase with stoichiometry close to AH3·nH2O. It can be also observed that ACn and FW contents for both pastes slightly diminished with time. It is worth noting that the opposite behavior is observed for samples without gypsum (Tables 4.7 and 4.8). This is mainly due to the larger amounts of ettringite and the absence of AFm-type phases. 115 Ana María Cuesta García Table 4.9. Quantitative phase analysis results (wt %) for st-C4A3S-g-1.42 paste, as a function of hydration time obtained by SXRPD. to 1h45m 2h30m 3h30m 14h 18h st-C4A3S 22.9(1) 23.4(1) 17.3(1) 9.7(1) 0.4(1) 0.3(1) CSH2 12.8(1) 12.0(1) 8.2(1) 3.1(1) 0.3(1) 0.3(1) AFt - 5.4(1) 15.0(1) 28.0(1) 47.8(1) 47.5(1) ACn+FW 6.1(1)+58.1*=64.2 59.2(1) 59.6(1) 59.2(1) 51.4(1) 51.9(1) *Theoretical free water content Table 4.10. Quantitative phase analysis results (wt %) for ss-C4A3S-g-1.42 paste, as a function of hydration time obtained by SXRPD. to 1h30m 3h 14h 18h ss-C4A3S 26.3(1) 25.8(1) 25.3(1) 3.2(1) 0.7(1) CSH2 13.4(2) 14.3(2) 14.3(2) 1.5(1) 0.3(1) AFt - 0.2(1) 0.8(2) 32.9(1) 37.8(1) ACn+FW 2.2(1)+58.1*=60.3 59.6(1) 59.6(1) 62.5(1) 61.2(1) *Theoretical free water content The main difference found between the hydration of st-C4A3S-g-1.42 and that of ss-C4A3S-g-1.42 is related to the reaction pace, Figure 4.10(b). The hydration kinetics in st-ye'elimite is faster than in ss-ye'elimite at early ages. Figure 4.13 shows the SXRPD raw patterns for both samples at similar early hydration ages for the sake of comparison. This effect was corroborated by calorimetric measurements since most of the heat evolved by stoichiometric sample is centered around 6 hours while that the signal of ss-C4A3S is placed close to 12 hours. Figure 4.14 shows the calorimetric data for both samples. It is important to highlight that the hydration of stoichiometric ye'elimite has been accelerated in the presence of gypsum when compared to the hydration behavior of this phase without gypsum, Figure 4.10. However, for the solid solution ye'elimite, the presence of gypsum has only a little effect in the kinetics of hydration. 116 4. Results and discussion Figure 4.13. Selected range of a SXRPD (λ=0.61975(1) Å) raw pattern for (a) ss-C4A3S-g1.42 at 3 h and (b) st-C4A3S-g-1.42 at 3.5 h, with the main peaks due to a given phase labeled. Figure 4.14. Conduction calorimetric curves for st-C4A3S-g-1.42 (blue line) and ss-C4A3Sg-1.42 (red line). 46810 12 2θ( 0 ) I (a.u.) (a) (b) C 4 A 3 S Quartz AFt CSH 2 0 5 10 15 20 25 30 0.000 0.005 0.010 0.015 0.020 0.025 Normalizedheat flow (W/g) t (h) 117 Ana María Cuesta García Finally, anhydrite was also used to study its influence on the hydration mechanism of ye'elimite, stoichiometric and doped (see Table 6 and 7 in article #5). The main difference with reference to the gypsum study was that the hydration mechanism was much slower because anhydrite was dissolved at a very low pace, as expected, Figure 4.10(b). In that case the main hydration product was again AFt. 4.5.3. Reactivity of ye'elimite with C4AF. The hydration of ye'elimite (stoichiometric and doped) in combination with C4AF has been studied by in-situ SXRPD to characterize the hydration processes which have taken place at early-ages (article #7). Firstly, st-ye'elimite, ferrite and gypsum were mixed in proportions close to those obtained for BCSAF reported by Álvarez-Pinazo et al. (2012) with a water/solid ratio of 1.0 (C4AF-st-C4A3S-g-1.0). This paste has been studied at early ages, up to 46 hours. Figure 4.15(a) shows the full quantitative phase analysis results as a function of time. Initially, only reaction to form AFt took place. Then, when gypsum was totally consumed AFm phase crystallized and AFt started to consume/decrease. The formation of AFm was favored not only from ye'elimite reaction but also likely from AFt dissolution, see reaction [1.12]. The time evolution of ACn and FW contents had a complementary behavior. It can be observed that these values slightly diminished with time during the crystallization of AFt and it is worth noting that the opposite behavior was observed when AFm started to appear. 118 4. Results and discussion Figure 4.15. Full quantitative phase analysis results for (a) C4AF-st-C4A3S-g-1.0 and (b) C4AF-ss-C4A3S-g-1.0. Moreover, st-ye'elimite with C4AF and gypsum was also studied with 0.7 and 1.3 w/s ratios up to 12 hours in order to study the water influence at very early ages. Tables 5 and 6 (in article #7) show the phase assemblages with time for both mixtures. It is well known that higher amounts of water enhance ye'elimite 110 0 10 20 30 40 50 60 110 0 10 20 30 40 50 60 t(h) wt(%) (b) (a) C 4 A 3 S AFm AFt CSH 2 ACn+FW C 4 AF 119 Ana María Cuesta García reactivity (Winnefeld and Barlag 2010) and this behavior has also been confirmed here. Figure 4.16 shows degree of reaction of ye'elimite in the mixture with gypsum and C4AF with different w/s ratios. This figure shows the enhanced ye'elimite hydration with higher amounts of water. Moreover, the results obtained for these samples indicate that higher amounts of water favor the formation of AFm because the reaction is complete and the system is closer to the equilibrium, in agreement with the previous study of the hydration of ye'elimite samples only with water (section 4.5.1). Figure 6 in article #7 shows the Rietveld plots of these mixtures with different w/s ratios where it is clearly shown that AFt phase is favored when w/s is low. Figure 4.16. Degree of reaction [α] of ye'elimite as a function of time for C4AF with ye'elimite pastes. Solid lines are just guide to the eyes. 0 5 10 15 20 45 50 0 20 40 60 80 100 C4AF-st-C4A3S-g-0.7 C4AF-st-C4A3S-g-1.0 C4AF-st-C4A3S-g-1.3 C4AF-st-C4A3S-1.0 C4AF-ss-C4A3S-g-1.0 t(h) α(%) 120 4. Results and discussion Secondly, doped ye'elimite was mixed with C4AF and gypsum (C4AF-ssC4A3S-g-1.0) to determine its hydration behavior in these conditions. Figure 4.15(b) shows the full quantitative phase analysis results as a function of time. Here, ye'elimite started to consume fastly and AFt precipitated at early stage. Only when ss-ye'elimite ended up dissolving, C4AF started to consume and AFm appeared. Furthermore, the ACn content was in agreement with this behavior, since it diminished with time during the crystallization of AFt and the opposite behavior was observed when AFm started to appear. Moreover, the total heat evolved at 2 days was 382 and 409 J/g for C4AF-st-C4A3S-g-1.0 and C4AF-ss-C4A3S-g-1.0, respectively. The total heat released for the C4AF-ss-C4A3S-g-1.0 sample was higher due to the total consumption of ye'elimite and the precipitation of bigger amounts of ettringite. The main difference between stoichiometric and doped ye'elimites, after 46 hours of hydration, is that the C4AF-ss-C4A3S-g-1.0 sample yields much larger relative amounts of AFt, see Figure 4.17. Keeping this direct observation in mind, we can conclude that the differences in the hydration mechanisms between both ye'elimite polymorphs are intensified when ferrite is in the reaction medium. A SEM-EDS study was also performed for C4AF-st-C4A3S-g-1.0 and C4AF-ss-C4A3S-g-1.0 pastes at 48 hours of hydration. Figure 4.18(a) shows Fe/Ca vs Al/Ca of these samples and Figure 4.18(b) shows a representative SEM photograph. It shows that most particles, with needle shape identified as AFt (Figure 4.18(b)) and with close compositions to theoretical ettringite (circle in Figure 4.18(a)) do not contain significant amounts of iron. Moreover, those particles without needle shape with composition similar to that of AFm neither contain iron in their structures, Figure 4.18. In addition, bright particles with high iron contents were also observed in SEM studies which may correspond to ironrich amorphous phases and also a dark gray background with a very large Al/Ca ratio which could be related to aluminum-rich ACn phases (Song et al. 2015). 121