Loading...

Study of hydration processes of Portland cements blended with supplementary cementitious materials

Master's Thesis 2012 144 Pages

Chemistry - Materials Chemistry

Excerpt

Contents

Aknowledgement

Abstract

Kurzfassung

List of Figures

List of Tables

List of Abbreviations and Symbols

1 Introduction
1.1 Outline of the Development of Ordinary Portland Cement
1.2 Research Objectives

2 Fundamentals
2.1 Ordinary Portland Cement
2.1.1 Mechanisms of Portland Cement Hydration
2.1.2 Development of Microstructure
2.1.3 The System C3A-CaSO4-H2O
2.2 Fly Ash
2.2.1 The Pozzolanic Reaction and the Hydration of Fly Ash
2.2.2 Interactions between Ordinary Portland Cement and Fly Ash
2.2.3 Activation of Fly Ash

3 Materials
3.1 Portland Cement
3.2 Fly Ash
3.3 Anhydrite
3.4 Laboratory synthesised C3A

4 Sample Preparation and Methods
4.1 Mix Design
4.2 Procedures of Sample Preparation
4.2.1 Paste Samples
4.2.2 Mortar Samples
4.3 Investigation of Flexural and Compressive Strength
4.4 Isothermal Calorimetry
4.5 Thermogravimetric Analysis
4.6 Chemical Shrinkage
4.7 X-Ray Diffraction
4.8 Scanning Electron Microscopy
4.9 Mercury Intrusion Porosimetry
4.10 Thermodynamic Modelling

5 Experimental Results and Discussion
5.1 Effect of Fly Ash, Anhydrite and C3A on the development of compressive strength
5.1.1 Compressive strength
5.1.2 Isothermal Calorimetry
5.1.3 Thermogravimetric Analysis
5.1.4 Chemical Shrinkage
5.1.5 X-ray Diffraction
5.1.6 SEM Image Analysis
5.1.7 Mercury intrusion Porosimetry
5.2 Effect of elevated Anhydrite and C3A contents upon presence of Fly Ash
5.2.1 Compressive strength
5.2.2 Isothermal Calorimetry
5.2.3 Thermogravimetric Analysis
5.2.4 Chemical Shrinkage
5.2.5 X-ray Diffraction
5.2.6 SEM Image Analysis
5.2.7 Mercury Intrusion Porosimetry
5.3 Effect of System Activation via Na2SO4
5.3.1 Mechanical Properties
5.3.2 Isothermal Calorimetry
5.3.3 Thermogravimetric Analysis
5.3.4 Chemical Shrinkage
5.3.5 X-ray Diffraction
5.3.6 SEM Image Analysis
5.3.7 Mercury intrusion Porosimetry
5.4 Thermodynamic Modelling
5.4.1 Calculated volume of the phases as a function of time
5.4.2 Calculated volume of the phases at complete reaction

6 Conclusions

References

Appendix
Appendix I: Complete display of the Compressive strength Properties
Appendix II: Inspection Records Mechanical Strength
Appendix III: TGA curves of the matrix under investigation
Appendix IV: Loss of bound water (exclusive Portlandite) detected via TGA; tem- perature range 50 - 420°C
Appendix V: Weight loss due to dehydroxylation of portlandite detected via TGA;
Appendix VI: Calculated Bound Water and Portlandite normalised to 100 g OPC .
Appendix VII: X-ray Diffractograms
Appendix VIII: SEM images
Appendix IX: Histograms of SEM images
Appendix X: Mercury Intrusion Porosimetry
Appendix XI: Comparison of Mix 4 and Mix 5
Appendix XII: PSD Analysis of the Raw Materials
Appendix XIII: Protocol for the Fly Ash Analysis

List of Figures

2.1 Total heat flow during early hydration of OPC at 23 °C. The contribution of the silicate reaction (Si) and aluminate reactions (Al1, Al2) to the total heat flow is indicated by dotted and dashed lines, respectively. I — initial period, II — induction period, III — acceleration phase, and IV — slow down phase (nomenclature according to Taylor[106] ). Adapted from Hesse et al.[45]

2.2 Development of microstructure during the hydration of OPC. Fss = ferrite solid solution. From[90]

2.3 The zonal structure of the glass matrix of FA[28]

2.4 Modell of the pozzolanic reaction of FA as described by Blaschke[14]

2.5 Scheme of the hydration in a system with pozzolanic grains (e.g. FA) and C3S grains[78] (left) and C3A grains[108] in the presence of Ca(OH)2 and CaSO4 · H2O

3.1 Regime of the heat treatment during the synthesis of C3A

4.1 Cut-away drawing of a TAM Air calorimeter[113]

4.2 Example of different curves gained via isothermal calorimetry displaying the heat flow (solid lines) and the released heat (dashed lines) of hydration of a OPC and a OPC blended with 30% FA

4.3 Weight loss and corresponding reactions taking place during TGA in an OPC containing 30% FA

4.4 Schematic progress of chemical shrinkage as a function of hydration time adapted from[56]

4.5 XRD pattern for a OPC/FA blend with a FA share of 30 % cured for 7 d. E - ettringite, H - hemicarbonate, M - monocarbonate, B - brownmillerite, V - mullite, P - portlandite, Q - quartz, C - calcite, D - dolomite, X - C2S, Y - C3S, Z - C3A

4.6 Interactions of the specimen with the electron beam in a SEM

4.7 Backscattered image of a FA blended cement at 28 days and its corresponding grey level histogram. P - porosity, HP - hydration products except portlandite, CH - portlandite, An - unreacted clinker particles, FA - fly ash[9]

4.8 Schematic description of MIP. a - evacuation, b - filling with mercury, c - application of pressure

4.9 Mercury intrusion and extrusion curves. Volume of incorporated mercury as a function of applied pressure (left) and as a function of the pore radius (right)

4.10 The volume of the different phases in a hydrating cement paste containing FA and LS as a function of time[24]

5.1 Effect of FA, AH and C3A: Compressive strength properties

5.2 Effect of FA, AH and C3A: I - specific heat flow with respect to the amount of OPC, II - released heat of the samples with respect to the amount of OPC

5.3 Effect of FA, AH and C3A: Compressive strength as a function of released heat of hydration

5.4 Effect of FA, AH and C3A: Bound water (I) and portlandite content (II) normalised to the amount of OPC up to 90 d of hydration

5.5 Effect of FA, AH and C3A: Compressive strength as a function of bound water

5.6 Effect of FA, AH and C3A: I - Chemical shrinkage normalised to the amount of OPC, II - Compressive strength as a function of chemical shrinkage

5.7 Effect of FA, AH and C3A: Detailed X-ray diffractograms up to 90 d. E - ettringite, H - hemicarbonate, M - monocarbonate, B - brownmillerite, V - mullite, P - portlandite, Q - quartz, C - calcite, D - dolomite, X - C2S, Y - C3S, Z - C3A

5.8 Effect of FA, AH and C3A: Compressive strength as a function of porosity and amount of anhydrous clinker phases for samples cured for 1 d and 28 d

5.9 Effect of FA, AH and C3A: Exemplary SEM images of samples cured for 1 d (left) and 28 d (right)

5.10 Effect of FA, AH and C3A: Compressive strength as a function of porosity as measured by MIP

5.11 Effect of FA, AH and C3A: 1st and 2nd mercury intrusion cycle. Corrected pore size calculation and neck pore entrance calculation for samples cured for 1 d (left) and 28 d(right)

5.12 Effect of elevated AH and C3A contents upon presence of FA: Compressive strength properties

5.13 Effect of elevated AH and C3A contents upon presence of FA: I - specific heat flow with respect to the amount of OPC, II - released heat of the samples with respect to the amount of OPC

5.14 Effect of elevated AH and C3A contents upon presence of FA: Compressive strength as a function of released heat

5.15 Effect of elevated AH and C3A contents upon presence of FA: Bound water (I) and portlandite content (II) normalised to the amount of OPC up to 90 d of hydration

5.16 Effect of elevated AH and C3A contents upon presence of FA: Compressive strength as a function of bound water

5.17 Effect of elevated AH and C3A contents upon presence of FA: I - Chemical shrinkage normalised to the amount of OPC, II - Compressive strength as a function of chemical shrinkage

5.18 Effect of elevated AH and C3A contents upon presence of FA: Detailed X-ray diffractograms up to 90 d. E - ettringite, H - hemicarbonate, M - monocar- bonate, B - brownmillerite, V - mullite, P - portlandite, Q - quartz, C - calcite, D - dolomite, X - C2S, Y - C3S, Z - C3A

5.19 Effect of elevated AH and C3A contents upon presence of FA: Compressive strength as a function of porosity and amount of anhydrous clinker phases for samples cured for 1 d and 28 d

5.20 Effect of elevated AH and C3A contents upon presence of FA: Exemplary SEM images of samples cured for 1 d (left) and 28 d (right)

5.21 Effect of elevated AH and C3A contents upon presence of FA: Compressive strength as a function of porosity as measured by MIP

5.22 Effect of elevated AH and C3A contents upon presence of FA: 1st and 2nd mercury intrusion cycle. Corrected pore size calculation and neck pore entrance calculation for samples cured for 1 d (left) and 28 d(right)

5.23 Effect of system activation via Na2SO4: Compressive strength properties

5.24 Effect of system activation via Na2SO4: I - specific heat flow with respect to the amount of OPC, II - released heat of the samples with respect to the amount of OPC

5.25 Effect of system activation via Na2SO4: Compressive strength as a function of released heat

5.26 Effect of system activation via Na2SO4: Bound water (I) and portlandite con- tent (II) normalised to the amount of OPC up to 90 d of hydration

5.27 Effect of system activation via Na2SO4: Compressive strength as a function of bound water

5.28 Effect of system activation via Na2SO4: I - Chemical shrinkage normalised to the amount of OPC, II - Compressive strength as a function of chemical shrinkage

5.29 Effect of system activation via Na2SO4: Detailed X-ray diffractograms up to 90 d. E - ettringite, H - hemicarbonate, M - monocarbonate, B - brownmillerite, V - mullite, P - portlandite, Q - quartz, C - calcite, D - dolomite, X - C2S, Y - C3S, Z - C3A

5.30 Effect of system activation via Na2SO4: Compressive strength as a function of porosity and amount of anhydrous clinker phases for samples cured for 1 d and 28 d

5.31 Effect of system activation via Na2SO4: Exemplary SEM images of samples cured for 1 d (left) and 28 d (right)

5.32 Effect of system activation via Na2SO4: Compressive strength as a function of porosity as measured by MIP

5.33 Effect of system activation via Na2SO4: 1st and 2nd mercury intrusion cycle. Corrected pore size calculation and neck pore entrance calculation for samples cured for 1 d (left) and 28 d(right)

5.34 The volume of the different phases as function of time in hydrating cement pastes modeled by GEMS. I - OPC 100, II - OPC/FA 70/

5.35 Compressive strength as a function of modeled ettringite content. I - Effect of FA, AH and C3A, II - Effect of elevated AH and C3A contents upon presence of FA, III - Effect of system activation via Na2SO

5.36 Modeled phases after complete reaction. I - Addition of C3A, II - Addition of AH and C3A in the ratio 1/1.5, III - Addition of AH and C3A in the ratio 1/1.5 to systems activated with 1 wt.% Na2SO

List of Tables

2.1 Main clinker phases, adapted from[5]

2.2 Average chemical composition of clinker; adapted from[5]

2.3 Compositions of different fly ashes; data in [%]

3.1 Chemical analysis and phase composition of CEM I 42.5 R

3.2 Bogue calculation of the main clinker phases in [%]

3.3 Analysis of fly ash

3.4 Calculation of the composition of the glass phase of the FA

3.5 Analysis of anhydrite

3.6 Rietveld analysis of synthesised C3A; data in [wt%]

4.1 Compositions of the mixtures submitted to examination

5.1 Effect of FA, AH and C3A: Total porosity as measured in mercury intrusion experiment. Data in [%]

5.2 Effect of elevated AH and C3A contents upon presence of FA: Total porosity as measured in mercury intrusion experiment. Data in [%]

5.3 Effect of system activation via Na2SO4: Total porosity as measured in mercury intrusion experiment. Data in [%]

List of Abbreviations and Symbols

illustration not visible in this excerpt

Declaration

I hereby declare that I completed this work without any improper help from a third party and without using any aids other than those cited. All ideas derived directly or indirectly from other sources are identified as such. This declaration also refers to the representation of figures and visual material.

March 30, 2012 Axel Schöler

Eidesstattliche Erklärung

Ich versichere, dass ich diese Arbeit selbständig verfasst und keine anderen Hilfsmittel als die angegebenen benutzt habe. Die Stellen der Arbeit, die anderen Werken dem Wortlaut oder dem Sinn nach entnommen sind, habe ich in jedem einzelnen Fall unter Angabe der Quelle als Entlehnung kenntlich gemacht. Diese Versicherung bezieht sich auch auf die bildlichen Darstellungen.

30. März 2012 Axel Schöler

Aknowledgement

In the course of my stay at the Laboratory for Construction Chemistry and Concrete at Empa, Materials Science and Technology, Dübendorf, Switzerland, a lot of people contributed to the completion of this thesis in one way or another. In general I have to feel obliged to every single person at the department for making my stay in Switzerland extraordinary pleasant and educational. As it would go beyond the scope to namecheck everyone I can think of only a few people whom I consider the main contributors will be listed by name in the following.

First of all I would like to thank my academic supervisors Dr.sc.nat. Barbara Lothen- bach and Dr.rer.nat. Frank Winnefeld for their support and assistance throughout, not only on the implementation on the necessary analytical tests and experiments but also on their help in terms of the professional evaluation of the gained data and the comple- mentation on this thesis. You always made time for discussion and at solving occurring problems and gave helpful advices and hints based on your wealth of experience and expertise.

I have to feel obliged to all the other scientists at the laboratory as well, especially Dipl.-Min. Florian Deschner for his thoughts on the role of fly ash within the cement hydration and for his committed support with the acquisition of SEM pictures. Thanks to the magician of the X-rays, Dr. Gwenn Le Saout, for help with the XRD experiments and the Rietveld refinement in terms of the fly ash. I am indebted to Dr. Josef Kaufmann for implementation of the MIP and PSD experiments and his dedicated help and guidance on the evaluation of the results thereof.

Of course the technical staff at the laboratory lend me a helping hand as well. Here a special thank you goes out to Luigi Brunetti and to Boris Ingold who never hesitated to answer my questions or support me on whatever issue even as they were always on the go.

My two industrial partners from HTC Leimen who in the end made this project possible due to their interest in the interrogation. Thank you Dr. Maciej Zajac and Dr. Dirk Schmitt. In terms of HTC another helpful contributor was found with Dr. Mohsen Ben Haha who carried out image analysis on my SEM pictures.

Concluding I have so send extra special thanks to my family for your never ending support and for the facilitation of my studies and last but not least cordial thanks to my dearest love, Veronika Märkl, for your encouragement, your support and criticism and for always coming up with new recipes to keep my mind in a good mood.

Abstract

The production of Portland cement clinker has a share of about 6% to 8% on the global CO2-emissions . Approximately 60% of those emissions are attributable to the decarbon- ation of limestone. A widespread approach for the reduction of the CO2-emissions is to replace the clinker in the cement by pozzolanic waste materials, e.g. fly ash. The reaction of fly ash is generally slow. Due to this slow reaction cement pastes mixed of Portland ce- ment and fly ash have a slower strength development than pure Portland cement, at least at early ages up to 7 days. The goal of this study was to increase early strength properties of Portland cement/fly ash blends by increasing the early ettringite formation in order to decrease the porosity. Therefore various amounts of anhydrite and laboratory synthesised C3A were added to Portland cement/fly ash systems that contained 30% fly ash. The behavior of these systems in terms of kinetics, phase development and microstructure was studied by means of strength tests, isothermal calorimetry, thermogravimetric analysis, chemical shrinkage, X-ray diffraction, backscattered electron image analysis, mercury in- trusion porosimetry and thermodynamic modeling. In addition the activation by Na2SO4 was investigated on certain systems.

It was found that the amount of ettringite formed increases gradually with increasing amounts of anhydrite and C3A. The formation of ettringite was completed at approx- imately 1 day. Further the activation with Na2SO4 increased the amount of ettringite and shifted the completion of its formation to approximately 2 days. Due to the addition of anhydrite and C3A the porosity decreased corresponding to the amount of ettringite formed. The effects on the development of early strength were most distinct for very high additions and for systems activated by Na2SO4. All systems showed equal strengths at

7 d and 28 d. For later ages up to 90 days, the benefits at early ages were found to be inversed, leading to lower strength values compared to systems without or with low addi- tions of anhydrite, C3A or Na2SO4. Eventually the increase in early strength was either too low to be significant or only relevant at very high additions of anhydrite and C3A with simultaneous system activation by Na2SO4. In terms of kinetics a remarkable impact of the added C3A was observed as this seemed to significantly accelerate the silicate reaction of the Portland cement.

Kurzfassung

Etwa 6% bis 8% der weltweiten CO2-Emissionen sind auf die Produktion von Portlandze- ment zurückzuführen. Mehr als die Hälfte (60%) dieser Emissionen entsteht bei der Cal- cinierung des Kalksteins während der Klinkerherstellung. Die teilweise Substitution des Klinkers durch puzzolanische Materialien wie beispielsweise Flugaschen ist eine gängige Methode um die Menge des freigesetzten CO2 zu reduzieren. Die Reaktion von Flugasche ist jedoch für gewöhnlich vergleichsweise langsam. Aufgrund dessen geht die Entwick- lung von Druckfestigkeiten bei Portlandzement/Flugasche-Gemischen innerhalb der er- sten Tage der Hydratation langsamer vonstatten als bei reinem Portlandzement. Das Ziel dieser Arbeit war es, die Frühfestigkeiten von Portlandzement/Flugasche-Gemischen durch die Erhöhung des Ettringitgehalts, und damit einhergehender sinkender Porosität, zu steigern. Hierfür wurden verschiedene Mengen an Anhydrit und synthetisiertem C3A zu Portlandzement/Flugasche-Gemischen gemengt. Der Flugascheanteil betrug stets 30%. Mittels Festigkeitsprüfungen, isothermer Kalorimetrie, thermogravimetrischer Anal- yse, Chemischem Schwinden, Röntgenbeugung, Bildanalyse von Rückstreuelektronen- bildern aus der Rasterelektronenmikroskopie, Quecksilberdruckporosimetrie und thermo- dynamischer Modellierung wurden entsprechende Systeme hinsichtlich ihrer Hydratation- skinetik, der Phasenentwicklung und der Entwicklung ihrer Mikrostruktur untersucht. Weiterhin wurde die Systemaktivierung mittels Na2SO4 an ausgewählten Systemen un- tersucht.

Die Untersuchungen zeigten, dass die Menge des gebildeten Ettringits mit Erhöhung der Anteile des zugesetzten Anhydrit und C3A sukzessive stieg. Die Ettringitbildung war nach etwa einem Tag abgeschlossen. Die zusätzliche Aktivierung mittels Na2SO4 führte zu einem höheren Ettringitgehalt und verschob den Abschluss der Ettringitbildung auf etwa zwei Tage. Die Porosität sank aufgrund der Zugaben von Anhydrit und C3A bed- ingt durch die Beeinflussung des Ettringitgehalts. Bei sehr hohen Anteilen an zugeset- ztem Anhydrit und C3A und bei Systemaktivierung mittels Na2SO4 wurden die höchsten Steigerungen hinsichtlich der Frühfestigkeit beobachtet. Nach einer Hydratationdsauer von 7 resp. 28 Tagen zeigten alle Systeme vergleichbare Festigkeiten. Weiterhin kon- nte ein inverser Effekt beobachtet werden, der sich dadurch äusserte, dass Steigerungen der Frühfestigkeit bei Systemen mit hohen Anteilen an zugesetztem Anhydrit und C3A zu Festigkeitseinbussen bei späteren Zeiten bis zu 90 Tagen führten. Dieser Effekt trat ebenfall bei Systemen, welche mittels Na2SO4 aktiviert wurden, auf. Letztendlich waren die Steigerungen der Frühfestigkeit entweder nicht signifikant oder nur bemerkenswert bei sehr hohen Anteilen an Anhydrit und C3A bei gleichzeitiger Systemaktivierung durch Na2SO4. Hinsichtlich der Hydratationskinetik führte das zugesetzte C3A scheinbar zu einer ausgeprägten Beschleunigung der Silikatreaktion des Portlandzements.

1 Introduction

1.1 Outline of the Development of Ordinary Portland Cement

Cement as a hydraulic binder undeniably cuts the most outstanding figure in the modern building and construction industry. Nevertheless it was a long way from the first known applications of hydraulic binders to the contemporary monopolistic position of Ordinary Portland Cement (OPC).

The earliest traces that verify the use of hydraulic binders reach back far into the past. In Nevali Çore, located in Turkey, Felder-Casagrande et al.[31] found mortars that were examined via carbon 14 dating techniques. Results imply an age of roughly 22600 years and proof that the mortar was burnt. Other investigations report on the discovery of a 7000 years old floor during excavation work in Jericho[64] or on screed pavements in Lepenski Vir that can be dated to a time frame from 5600 to 5000 BC [92, 93]. The mortars there were manufactured under the use of anhydrous lime. According to Kingery the use of lime even preceded the use of pottery[55].

To name a precursor of the most important representative of its class, OPC, it is repeatedly referred to comparing with binders as they were used in ancient Rome. It is likely that it was back then when pozzolanic materials were used in full range for the first time in order to improve civil engineering [63, 44]. However the Romans did not invent concrete, they just improved a technique they adapted from the Greeks. At about 300 BC the Greeks invented a new technique of building masonry. There a mixture of coarse and fine quarrystones was heaped up between two layers of masonry built out of square stone blocks. Then this bulk material was spilled with lime mortar and compacted via poking. This masonry was called Emplecton[99]. The Romans adapted this Greek technique and used the term Opus Caementitium therefore. First this term meant only the cementitious masonry made from quarry stone with anhydrous lime as a binder. Later on it was extended to cementum, cimentum, cäment or cement and included every additive, volcanic ash and brick-dust that could be processed into a hydraulic binder when blended with anhydrous lime. The use and application of this binder can still be comprehended today using the example of historic buildings like the Pantheon, the Pont du Gard or the Colosseum[59]. After the Roman Empire collapsed, however, no appreciable innovations can be named in the field of cementitious materials and moreover the lion’s share of the gained knowledge disappeared for whatever reason. To see the emergence of a new building material with the discovery of OPC it was necessary to wait almost 2000 years.

It seems to be impossible to exactly determine the natal hour of modern cement. In France it is believed that modern cement was invented by Louis Vicat when he found the proportions of limestone and clay to produce ‘chaux factice’ following a scientific process (proto-Portland cement)[15]. Vicat’s work constitutes a real milestone and the results of his work were published in 1818 and 1828 [109, 110]. Moreover the first French hydraulic lime plant was built as early as 1818 in Nemours and the first French Portland cement plant was built in 1840 in Boulogne-sur-Mer[5]. Yet it is most common to consider 1824 as the year of birth of OPC. It was the English master bricklayer Joseph Aspdin who applied for a patent of his invention: ” An Improvement in the Modes of Producing an Artificial Stone ” [41]. As his product, at least in its visual appearance was very similar to the well- known limestone of Portland island he choose the islands name as a eponym[100]. Aspdin used crushed calcinated limestone to produce anhydrous lime which he then grinded to slurry with water after adding argillaceous earth or clay. This mixture was broken and calcinated in a kiln to expell all carbonates. Finally the obtained material was ground to a fine powder. The result: Clinker. Indeed this clinker really had nothing to do with what we call OPC today. It is more likely that Aspdin’s son, William Aspdin, succeeded in producing an OPC-prototype around 1850 when he fired a mixture of limestone and clay at a sufficient high temperature to produce the main clinker phase, tricalcium silicate [15].

The term clinker arose from the kilns used for firing. At first there were no particular kilns for clinker production and as kilns for firing bricks were widespread available they were alienated. From around 1900 those in turn disappeared for the benefit of shaft or rotary kilns which could be operated continuously and therefore meant a major step forward to mechanisation. Moreover this innovation allowed vastly better transfer and production rates. Nowadays it is possible to produce about 10000 tonnes per day with modern rotary kilns[5].

The last major chemical innovation to crucially improve OPC or rather its processing dates back to around 1900. Historical cements were of such coarse milling that their reactivity was comparatively low and no set regulator in form of calcium sulfates was needed. But as mechanisation developed the fineness of the cement could be drastically improved and that meant a elevation of the reactivity due to the larger surface achieved. It was the French chemist Giron who initially added gypsum to the clinker during grinding [70]. This production stage allowed to control OPC setting and hardening precisely.

Yet the production of OPC still remains a very simple pyrotechnical process, at least in principle. It consists of firing an appropriate mixture consisting of lime, silica, alumina and iron oxide at about 1450°C. During this thermal treatment the blended raw materials are transformed into the four active phases that constitute the essence of OPC.

Easy to obtain raw materials and the comparatively basic production defined the triumphal procession that at the latest became unstoppable with the discovery of reinforced concrete in the middle of the 19th century which can be looked upon as one of the cornerstones of our modern society. According to CEMBUREAU the world cement production grew from 10 million tonnes in 1900 to an exorbitant number of 3000 million tonnes in 2009[5]. With a global population of about 6800 million people this means a per capita consumption of 440 kilogram cement or about 1500 kilogram concrete. Therefore cement is by far one of the most commonly used materials.

In spite of all the advantages OPC offers it would be a bit presumptuous not to mention drawbacks. The most serious difficulty in terms of the clinker production undeniably is the compulsory emergence of CO2 attributable to the used combustibles and to the conversion of calcium carbonate to calcium oxide. Presently 6% to 8% of the global CO2 emission can be attributed to the cement industry[6]. One option to reduce CO2 emissions is to substitute OPC with supplementary cementitious materials (SCMs). The used materials are not inert and therefore participate in the already very complex hydration reaction. In recent years a lot of attention was drawn to the role of limestone (LS), fly ash (FA) and new interest arouse in the already long in use ground granulated blast furnace slag (GGBFS) which lead to numerous studies [107, 111, 114, 35, 24, 11]. As the use of SCMs affects the whole cementitious system, the modulation of beneficial properties is the main challenge. Through knowledge of the exact effect of every participant in the cement hydration it seems to be possible to strictly adjust multi component systems and to improve their mechanical properties while simultaneously reducing CO2 emissions. The present study will contribute to this purpose.

1.2 Research Objectives

During the early hydration of OPC there are four main products formed: calcium silicate hydrate (C-S-H), portlandite (CH), AFt and AFm. These phases are responsible for setting and strength development in hydrating cement pastes.

The formation of C-S-H and CH is a result of the hydration of tricalcium silicate (C3S) and dicalcium silicate (C2S). C-S-H is known to be the main contributor in terms of strength. The individual clinker phases exert influence on the hydration of the other clinker phases present. For example tricalcium aluminate (C3A) hydration accelerates the hydration of C2S[98] and the addition of gypsum is known to quicken the hydration of C3S and C2S[5]. Within the hydration of C3S in the presence of C3A the role of calcium sulfate is of special interest. Quennoz et al.[85] found that C3A has an accelerating effect on C3S hydration at properly sulfated cements and a decelerating one in undersulfated cements when they investigated model cements containing C3A, C3S and gypsum. The catalysed reaction of these phases is accompanied by a faster development of microstructure which in turn means a faster strength development.

The term AFt characterises all manifestations of ettringite which initially meant a trisulfoaluminate but actually is a complex mineral that can incorporate various minor elements/ions and solid solutions [82, 81]. Ettringite is a product of the reaction taking place between the alumina phases C3A and calcium aluminate ferrite (C2AF) in the presence of calcium sulphate [21, 19, 20] and always occurs in OPC as long as there is a source of sulphate present. The main role of ettringite in the course of the hydration is the set regulation due to the decelerating effect on the highly reactive C3A. The volume ettringite occupies is comparatively high and its growth preferentially takes place in air voids. This in turn means a reduction of porosity in the hydrating system and for this reason a contribution to strength development as well. The formation of ettringite is typical for the early hydration of OPC and is normally accomplished after approximately one day.

The so-called monophases are summarised with the term AFm. Depending on the sup- ply and the kind of ions AFm can incorporate OH- , SO[2] - and CO[2] -. The presence of carbonate provides thermodynamic stabilisation for the AFm phase. Accordingly mono- carboaluminate is stable at 25°C while the stability of monosulfoaluminate is marginal at 25°C and hydroxy-AFm as well as a part of the extensive range of solid solutions formed by partial replacement of sulfate by hydroxide are metastable at 25°C[68]. Monocar- boaluminate is formed in the presence of calcite and is the most stable AFm phase. It indirectly stabilises ettringite which is why the allowed incorporation of 5% calcite in CEM I[4] is beneficial.

Blends of OPC and FA are used to reduce the CO2 emissions attributable to the clinker production. Yet the pozzolanic reaction of the FA does not start right from the first contact with water like it is the case with the clinker. As a result the strength development of OPC/FA-blends at early ages up to 2 d is lower compared to pure OPC. This drawback can be tackled by increasing the reactivity of the FA with mechanical activation (grinding) or chemical activation. Another approach is not to activate the FA but to modify the system to the effect that the microstructure provides better strength prior to the onset of the pozzolanic reaction of the FA. Accordingly the main targets of this thesis in respect to 1 d and 2 d compressive strength are summarised in the following:

- It is known that lower porosity generally leads to higher strength. Thus it is aimed at the reduction of porosity at early ages up to 2 d. This reduction requires the increase of solid phases which will be obtained by the increase of the ettringite content. To provide the reactants for the formation of supplementary ettringite additions of AH (anhydrite) and laboratory synthesised C3A to OPC/FA-blends are used.
- The reaction of fly ash is generally slow. Consequently few C-S-H is formed due to the pozzolanic reaction at early ages. As C-S-H is the main responsible hydrate phase in terms of strength the acceleration of the fly ash reaction is supported by two methods. The fly ash is a) ground to a fineness according to Blaine of 6700 cm[2] /g in order to increase the surface available for the reaction and b) chemical activation by Na2SO4 is applied.

2 Fundamentals

As this thesis deals with the influence of several additives on the hydration of cement pastes it is useful to give a short general overview on cement chemistry, hydration of OPC and pozzolana and interactions thereof. This overview does by no means claim to provide to be thorough. Further details can be found in the comprehensive professional literature wherefore it is referred to this whenever possible.

2.1 Ordinary Portland Cement

Modern OPC is made out of a ground clayey limestone which is burnt in a rotary kiln at a temperature of about 1450°C in order to get the main clinker phases formed. After leaving the kiln it is cooled down rapidly, crushed and milled with calcium sulfate, providing OPC. Depending on the distribution of the main clinker phases, the fineness that is adjusted via milling and the addition of calcium sulfate as set regulator specific characteristics may be manufactured. Average chemical compositions of the main clinker phases formed during this heat treatment are given in table 2.1. Note the cement chemists’ notation which is used in the field of cement and concrete research. A listing of this notation is given in the prefix of this thesis.

Table 2.1 - Main clinker phases, adapted from[5]

illustration not visible in this excerpt

Since OPC is a mass product which is manufactured in a major industrial process and not a laboratory synthesised material of chemical purity there can always be found some other chemical compounds like sodium sulfate (Na2SO4), potassium sulfate (K2SO4), periclase (MgO), calcium carbonate (CaCO3) and free lime CaO. Other incorporated impurities in a slight extend of less than 0.3 % are CaF2, P2O5, MnO2, ZnO, Cr2O3 and Mn2O3. The impurities may affect the development of the properties of the hardened cement paste and result from the actual chemical composition of the crude materials and the combustibles. Hence the amount on impurities can not be stated as absolute. Table 2.2 provides a paradigmatic composition of an average clinker. All of these compounds affect the hydration reaction, but alkalies are of special interest. The alkalies are incorporated in the clinker as long as the amount of SO3 in the clinker is about 1%. At higher SO3 amounts in the clinker approximately 90% of the K2O is present as K2SO4 and approximately 50% of the Na2O is present as Na2SO4. The rest of the alkalies is incorporated in the clinker. There Na is preferrably incorporated in C3A which is of particular importance. Different amounts of alkalies lead to different changes of the symmetry of the lattice[100]. Alkalies from the alkali sulfates that are brought into solution after the mixing of the paste contribute to the high pH of the pore solution and therefore are of eminent importance concerning the reactivity of FA as can be seen later in chapter 2.2.3. The hydration of crystalline MgO (periclase) or free lime (CaO), when present in significant amounts in OPC, can cause expansion and cracking[58]. To guarantee good durability the amount of free lime in the clinker is adjusted to values lower than 2% while EN 197-1[4] claims the total MgO amount in the clinker not to be higher than 5%.

Table 2.2 - Average chemical composition of clinker; adapted from[5]

illustration not visible in this excerpt

2.1.1 Mechanisms of Portland Cement Hydration

The terms cement hydration and hydration reaction denote all chemical and thermody- namic processes that lead to stiffening and hardening of cement when it comes into contact with water. While this behavior is very easy to handle the underlying mechanisms are still not entirely resolved to date. Aïtcin enunciates this state with much apropos as follows:

” Concrete is, therefore, the fruit of a very simple technology and a very complex science: this duality is at the same time the cause of concrete ’ s success but also of its weakness when this complex science is not mastered ” [5].

Since Le Chatelier and Michaëlis, two trailblazers of OPC hydration, numerous investi- gators devoted themselves to this phenomenon which caused the publication of countless books and papers. Hereinafter only a brief outline on the principles of OPC hydration is given. More in-depth considerations on this issue include but are by far not limited to Taylor[106], Odler[76], Gartner et al.[37], Stark et al. [97, 98], Jensen et al.[49] and Bullard et al.[17].

In the course of the hydration of the siliceous phases C-S-H and CH are being formed. These hydrate phases are the main components of hardened cement paste due to the large amount of their reactants in the clinker. As the main hydrate phase C-S-H occupies variable chemical compositions the chemical equations of C3S and C2S can only be stated in general form[100]:

illustration not visible in this excerpt

The ratio of calcium oxide to silica is an important parameter within cement chemistry and is known as C/S-ratio. In OPC this ratio is about 1.5[106].

C-S-H is not only the most common reaction product but it is also responsible for most of the mechanical properties of cement pastes. This is due to its formation as continuous layer that binds together the other hydrate phases into a cohesive whole. All the other hydrate phases form more or less as discrete crystals that are intrinsically strong but do not form strong connections to the surrounding solid phases which is why they contribute to the overall strength to a lesser content. The ability of C-S-H to act as a binding phase arises from its nanometer-level structure.

A crucial aspect of the quality of OPC is its workability time, which describes the timeframe between the mixing of the paste and the first stiffening. As already been stated with reference to[70] a source of calcium sulfate is added to the clinker during grinding. If there is no sulfate source present in the cement the paste does not posses any workability as it stiffens almost instantly due to the very rapid formation of calciumaluminates what leads to rapid set1:

illustration not visible in this excerpt

C4AH13 and C2AH8 are not stable and transform to stable C3AH6:

illustration not visible in this excerpt

Therefore the addition of sulfate is by all means necessary because it drastically slows down the very intense and exothermic reaction of C3A as calciumsulfate reacts with calciumaluminate to form ettringite.2

illustration not visible in this excerpt

As soon as the sulfate present is consumed, ettringite is no longer stable and is being decomposed for the benefit of the formation of monosulfate. The C3A reacts with the sulfates to form monosulfate.

illustration not visible in this excerpt

Reactions (2.1) and (2.2) produce CH and therefore are affected by any process that lowers the calcium ion concentration. Reactions (2.5) and (2.7) in contrast consume calcium hydroxide, consequently they will compete with any pozzolanic reaction3 in terms of the lime released by the hydration of the calcium silicates.

Reactions (2.1) and (2.2) produce CH and therefore are affected by any process that lowers the calcium ion concentration. Reactions (2.5) and (2.7) in contrast consume calcium hydroxide, consequently they will compete with any pozzolanic reaction3 in terms of the lime released by the hydration of the calcium silicates.

It has to be noted that the foregoing explanations are very simplified and are only valid in systems without calcite content. As long as calcite is present monocarboaluminate is thermodynamically stabilised. Accordingly monocarboaluminate is stable at 25°C while monosulfoaluminate is marginally stable and hydroxy-AFm as well as a part of the ex- tensive range of solid solutions formed by partial replacement of sulfate by hydroxide are metastable at 25°C [68, 69]. As monocarboaluminate is the most stable AFm phase the use of calcite leads to stabilisation of the Aft phase or ettringite at normal conditions. This is important because it is permitted to add up to 5% of additives (e.g. ground limestone) to CEM I[4].

The most unclear behavior in terms of the cement hydration can be attributed to ferrite. While the hydration mechanisms of C3S, C2S and C3A are the subject of many investigations the behavior of ferrite is somewhat neglected and thus not very well understood. The following interpretation of the ferrite hydration is a simplified description with reference to Stark and Wicht[100].

illustration not visible in this excerpt

The products resulting from this reaction are not stable. After their formation a subsequent transformation takes place as follows:

illustration not visible in this excerpt

A recent study by Dilnesa[27] is concerned with the fate of iron during cement hydra- tion. Within this study it was found that Fe-containing siliceous hydrogarnet is likely to occur particularly at high temperature. The found compositions contain both Al and Fe which is an indication that Al and Fe can be substituted by each other in the structure of hydrogarnet. The investigation via EXAFS (Extended X-ray Absorption Fine Struc- ture) showed that Fe was associated with C2(A,F) and Fe(OH)3 at early hydration ages whereby the latter dissolved and precipitated as C3(A,F)SH4. Furthermore calculated solubility products showed that Fe-ettringite is less stable than Al-ettringite which is why it is hardly present in hydrated cement pastes. The Fe-AFm phases were found to be more stable than the Al-AFm phases but their formation would require a rather high Fe concentration in the pore solution which is not the case in real systems. Consequently Fe is preferably bound in siliceous hydrogarnet.

As already repeatedly mentioned the above listed reactions describe a simplified representation of the hydration mechanisms. Thus some important details are briefly discussed in the following.

The composition of C-S-H is variable in certain borders. The average molar lime-silica ratios can vary in the range between 0.8 and 1.7. Thus it is better to simply speak about C-S-H not least because of its X-ray amorphous or cryptocrystalline character at ordinary temperatures with a structure often stated to be similar to tobermorite. C-S-H can incorporate alkalis, sulfur, iron and alumina and ions of Si[4] + and Ca[2] + can be exchanged with ions of Al[3] +, Fe[3] + and Mg[2] +[87]. Similarly to C-S-H the calcium sulfoaluminate and calcium sulfoaluminate ferrite hydrates describe hydrates of variable compositions and are referred to AFm and AFt, respectively. Depending on the composition of the paste different hydrates are present or stable.

The process of hydration can be expressed as the heat flow of the hydrating cement paste as a function of hydration time. Such heat flow curves are usually divided in five periods or three stages4 as can be seen later in this chapter with figure 2.1. The hydration of the very reactive C3A is controlled by sulfate. However the largest share on the process of the heat flow curves is attributable to C3S. The following explanations on the hydration of C3S are adapted from a summary by Juilland[50].

I. Initial period5

On the first contact with water, hydrolysis of the anhydrous surface layer releases rapidly ions into solution presumably as simple hydrated ionic species: Ca[2] +, OH- and H2SiO[2] -[4] [37]. This initial reaction is highly exothermic and lasts only a few minutes. The surface is thought to be altered in such a way as to be less reactive what leads to a first deceleration of the reaction. There are different theories that try to explain this behavior. The most common of these are discussed subsequently.

II. Dormant period

During this period the rate of reaction is low. Ca[2] + ions are brought into solution whereby the theoretical level of saturation for CH is exceeded until a maximum is reached. The reason for the end of this period is still unclear. The typical degree of hydration by the beginning of the acceleration period is normally in the range of 0.2 to 2.0 %[37]. Usually setting ocurs at the end of the dormat period.

III. Acceleration period

Right after the hydration and the rate of heat evolution thereof have come to a minimum a massive precipitation of C-S-H and CH occurs and the rate of heat evolution rises again. Nucleation and growth of these hydrates are commonly believed to be the rate controlling factor during this stage. Nevertheless there is still a lack of knowledge on the growth mechanism of C-S-H. Recently Bishnoi and Scrivener[13] proposed that during the acceleration period a loosely packed C-S-H fills a large frac- tion of the microstructure and the subsequently packing density increases with hydration.

IV. Deceleration period

As the naming elucidates the rate of heat evolution and hence the rate of reaction decreases continuously during this period. This trend is often referred to the start of a diffusion controlled reaction across the dense layer of C-S-H around the anhydrous grain. However, neither the nucleation and growth mechanism nor the diffusion controlled mechanism give sufficient clarification to this deceleration[22]. The most probable explanation is given by[13] assuming the densification process of the initially loosely packed C-S-H.

V. Final slow reaction

At this stage the curve flattens and the rate of heat evolution becomes really low which is due to gradual densification of the microstructure. Even if it is not entirely clarified yet, the limiting factor of this reaction is generally assumed to be transport processes or diffusion. This period affects the late strength properties of hardened cement pastes and concrete, respectively.

The reasons of the first deceleration of this reaction are still not clear and there are various theories that try to explain this behavior. The most important ones are the (i) protective membrane layer theory, (ii) nucleation and growth theory, (iii) double layer theory and (iv) dissolution theory.

(i) Protective Membrane Layer Theory

This is one of the earliest theories and it was first introduced by Stein and Stevels [102, 23] who stated that a hydrate layer that covers the C3S surface is built within early hydration. Kantro et al.[52] as well as Stein and Stevels stated that the first hydrate layer is converted into a hydrate layer that is fitting the anhydrous surface less closely and is more permeable to water. It was even stated that the hydrate layer is destroyed due to osmotic pressure[12] yet there was and still is little evidence to this behavior. Gartner and Jennings[36] give the most probable explanation for the presence of a protective layer. They stated that there exist two forms of C-S-H, one with higher solubility (SI) being metastable with respect to the other form (SII). They conclude that the first form of C-S-H has a protective character that is responsible for the period of slow reaction and that a solid-state transformation into a less protective form of C- S-H (SII) takes place as soon as the two phases have the same molar ratio of CaO to SiO2.

(ii) Nucleation and Growth Theory

Publications on this theory state that the length of the induction period is controlled by the delayed nucleation and growth of hydrates - either CH or C-S-H [33, 77]. As long as the nucleation of CH is considered to be rate determining it is suggested that CH is not being precipitated, even when the saturation of the solution with respect to this phase is reached. This is attributed to a poisoning of the nuclei by silicate ions[104]. The poisoning effect of adsorbed silicate ions is overcome when a corresponding high concentration of CH is reached and CH starts to precipitate. It can be assumed that the rate of reaction during both the induction and acceleration period is controlled by nucleation and growth of C-S-H nuclei [33, 77]. There the induction period ends when the growth of C-S-H nuclei starts. A point that stands in contrast to this theory is that there is substantially evidence that C-S-H is precipitated very early after the first contact with water[115].

(iii) Double Layer Theory

The double layer theory alleges the formation of a double layer and assumes that Ca[2] + ions are build up close to the surface which inhibits further dissolution[94]. The subsequent re-adsorption of the Ca[2] + ions on the negatively charged surface creates an electrical double layer and a positive ζ -potential. According to this theory the acceleration period is initiated by the precipitation of phases that consume Ca[2] + ions from the solution and leads to a incremental breakdown of the double layer[76]. A detailed discourse on the theory of the electrical double layer would go beyond the scope. More in-depth considerations on this topic are given by [96, 103].

(iv) Dissolution Theory

This is the latest theory that tries to explain the dormant period. Julliand[50] and Julliand et al.[51] present a very interesting and significant discussion of the importance of dissolution in the initial stages of alite hydration. Their model can be summarised in the following way. First the initial exposure to water starts a process of dissolution of the surface of the alite at dislocations which leads to the formation of etch pits. Some of the C-S-H forms on the surface as indicated by the heat generated in the first stage of the hydration process. The dissolution continues slowly whereby further growth of C-S-H is limited. When the induction period comes to an end the dissolution speeds up again and C-S-H resumes. Within this model the formation of a barrier layer on top of the alite as described in (i) is specifically excluded. This work has already been critically reviewed as the proposed model seems not to correspond with recently published and forthcoming experimental evidence[62].

As can be seen from the existence of the numerous theories there is no vividness on how the dormant period really proceeds and still a lot of work has to be carried out to finally understand the underlying mechanisms entirely. The complexity of the C3S hydration is even enhanced as the above discussed theories only deal with pure C3S but in real systems there arises interaction of C3S with other ions. Clinker grains do not exclusively consist of the four clinker phases and one of the most likely ion insertions in the C3S grain is the insertion of aluminum ions. According to investigations of Begarin et al.[8] impurities of aluminum in C3S lead to a longer delay before the acceleration compared to pure C3S. This delay is linked to the aluminum that passes into solution at the very early hydration. Due to these aluminates in the solution C-A-S-H precipitates instead of C-S-H. The first C-A-S-H nuclei do not grow and do therefore not support for growth as the pure C-S-H does. Another influence on the C3S hydration is exerted by calcium sulfates that always appear in modern OPC. A study of the C3S hydration in the presence of different kinds of calcium sulfate by Pourchet et al.[83] found that the sulfate type strongly modifies the early C3S CaSO4 hydration products and the rate of this hydration. They found that the replacement of gypsum by hemihydrate leads to an increase of the ettringite formation rate during at least the first five hours under their conditions.

Despite their different basic approaches and all the by now not understood mechanisms the discussed periods and models have one thing in common and that is their graphic representation as heat flow curves. A good and up to date interpretation of the observable heat flow is given in figure 2.1 by Hesse et al.[45]. This figure provides excellent insight to the contribution of the silicate and aluminate reaction to the total heat flow.

Even if it is not entirely correct6 the heat flow curve of OPC seems to be the sum of the silicate and aluminate reactions. The bulk of the heat of hydration results from the reaction of C3S as it is the most represented clinker mineral. The delayed reaction of the calcium aluminates is due to the formation of the ettringite layer and the peak visible at approximately 12 hours in the OPC curve results from the cracking of the layer when calciumsulfate is entirely consumed and the not yet reacted C3A reacts to form monosulfate. Despite this delay there reacts some C3A prior to the formation of the ettringite layer. This early reaction is responsible for the huge amounts of heat produced in the first minutes of the hydration as can be seen from the explanations to period I in terms of the aluminate reaction from figure 2.1.7 Due to the sulfate caused deceleration of the C3A reaction, C3S reacts somewhat faster with a maximum of produced heat at about 8 hours.

illustration not visible in this excerpt

Figure 2.1 - Total heat flow during early hydration of OPC at 23 °C. The contribution of the silicate reaction

(Si) and aluminate reactions (Al1, Al2) to the total heat flow is indicated by dotted and dashed lines, respectively. I — initial period, II — induction period, III — acceleration phase, and IV — slow down phase (nomenclature according to Taylor[106] ). Adapted from Hesse et al.[45].8

2.1.2 Development of Microstructure

The study on microstructures of cement pastes via scanning electron microscope (SEM) or transmission electron microscope (TEM) supplies visual insight on the above described reactions. Figure 2.2 describes schematically the sequence of changes undergone by a typical, polymineralic cement grain[90].

illustration not visible in this excerpt

Figure 2.2 - Development of microstructure during the hydration of OPC. Fss = ferrite solid solution. From [90].

Early Stage

Soon after mixing an amorphous and colloidal layer or membrane containing alumina, silica and also calcium and sulfate forms on the surface of the grains. The composition depends on that of the underlying surface. After about 10 minutes stubby AFt rods (2nd drawing) can be seen. They seem to be abundant near the surface of the aluminate phase and appear to be nucleated both, in the solution and on the outside surface of the formed gel layer.

Middle Stage

Within this stage which begins at about 3 hours and ends at about 24 hours approx- imately 30% of the cement react and a strong heat evolution occurs due to the rapid formation of C-S-H and CH (3rd drawing at approximately 10 hours). Undried C-S-H has a filmy and foil-like morphology while drying leads to give fibres9, 2 μ m in diameter, as soon as space is available or reticular networks when it is more restricted. CH forms massive crystals in the former water-filled space and nucleation sites are relatively rare which is why the growing crystals may engulf some of the smaller cement grains. C-S-H forms outwards in a thickening layer around cement grains and possibly also on AFt rods. After about 12 hours the C-S-H layer will make contact with layers growing on adjacent grains as it is by now 0.5-1.0 μ m thick. At this point a structure of interconnected shells is present which will play an important role in determining the mechanical properties that depends on the particle size distribution of the cement. A space of 0.5 μ m devel- ops between the shells and the anhydrous grains which is likely to be filled with highly concentrated colloidal solution. This space proves the reaction process to be based on dissolution and precipitation

The middle period ends with a renewed growth of AFt crystals typically 1-8 μ m of length. Their formation is associated with the right shoulder on cement heat evolution curves and implies an increase in the reaction of aluminate. C-S-H inner products will start to form on the inside of shells from continuing alite hydration (4th drawing; approximately 18 hours).

Late Stage

As the interconnected shells increase their permeability will decrease and C-S-H will also be deposited on the inside of the shells.[10] This happens very quickly and the formation of the inwarding C-S-H is often taking place more quickly than the retreat of the alite grains. More and more inner C-S-H is formed and fills the space between anhydrous grains and the hydrated shell (5th drawing; approximately 1-3 days). It seems that remaining spaces are filled up after about 7 days. The shells have by then become approximately 8 μ m thick. The aluminate phases react and the concentration of SO[2] -[4] drops rapidly inside the shell. AFm begins to form either from C3A or from AFt. As long as there is calcite present AFt is stable and monocarbonate is precipitated instead of monosulfate. AFt outside the shells may be stable way longer and perhaps indefinitely. In the final stage (6th drawing; 7-20 days) sufficient C-S-H has formed to fill the space between the anhydrous grain and the shell. After the complete filling of the spaces between the shells and the anhydrous cores any further reaction that leads to a densification of the paste will be slow.

2.1.3 The System C3A-CaSO4-H2O

As already discussed in paragraph 2.1.1 the addition of sulfate is absolutely necessary to gain a good workability of OPC. The sulfate source does furthermore influence the whole hardened cement paste and its properties. There are different manifestations of calcium sulfate and it is likely to find one or a mix of several of the following sulfates in OPC today: natural anhydrite, gypsum and hemihydrate[43]. All these manifestations stand out due to different solubilities and dissolution rates[63] which is why it is very important to gain exact analytical data on the sulfate source used. Calciumsulfate dihydrate has a solubility of 2.6 g CaSO4 · 2 H2O per litre H2O at 23°C. The solubility of calciumsul- fate hemihydrate is approximately four times that of calciumsulfate dihydrate and the solubility of anhydrite is 2.7 g CaSO4 per litre H2O. A in depth study on the modeling of calcium sulfate solubility in aqueous solutions can be found with[71]. The generally added calcium sulfates are dihydrate and anhydrite. Due to the heat evolution during the milling process there can possibly occur a further dehydration which means the manifes- tations and amounts of added calcium sulfate before milling do not necessarily represent those after the milling. Many authors who dedicated themselves to the investigation of the hydration mechanisms in the system C3A CaSO4 conclude that the initial rate of C3A hydration is significantly influenced by the morphology of the sulfate source used [105, 10, 89]. Haecker et al.[40] investigated the influence on different sources of CaSO4 on the mechanical properties. Their results can be summarised as follows:

- CaSO4 accelerates the cement hydration at an equal maximum value, whereby small amounts shorten the induction period while increasing amounts further accelerate the rate of hydration.
- High amounts of CaSO4 retard the hydration of C3A but despite retardation more ettringite is formed in the end.
- Different modifications have different solubilities and solubility rates11. Hemihy- drate has a higher solubility than anhydrite and hence the retardation is more distinct. Anhydrite reacts more slowly, is therefore longer available and there is a high rate of reaction as long as there is a high Ca[2] + concentration available.

2.2 Fly Ash

Fly Ash (FA), a waste product that is collected in the dedusting systems of power plants that burn hard coal or lignite coal, is a synthetic pozzolan often used to produce blended cements or to be added to concrete. Similar to the one of cement the composition of FA can be very different. Even in one power plant the quality of the FA varies considerable as the composition and the phenotype of the FA hinge on to the burned material and the process of combustion itself. The chemical composition of the FA depends mainly on the composition of the used combustible and the particle size distribution (PSD), the morphology and the mineralogy of the FA are subject mainly to the mode of firing[86]. Table 2.3 shows a small assortment of compositions that can be found with FA.

Table 2.3 - Compositions of different fly ashes; data in [%]

illustration not visible in this excerpt

Reference[7] [7] [34] [16]

It is widely known that the stake of FA on the hardening of pastes and the properties of those depends on the reactivity of the contained phases and consequently on their amount within the FA. ASTM C618 standard[2] describes two main families of FA based on the value of the sum of SiO2, Al2O3 and Fe2O3:

- If the value of SiO2 + Al2O3 + Fe2O3 is greater than 70 per cent, the FA is said to be a Class F FA.
- If the value of SiO2 + Al2O3 + Fe2O3 is lower than 70 per cent and the CaO content higher than 10 per cent, the FA is said to be a Class C FA.

The European standard EN 197-1[4] gives a different classification as it divides in between silicious FA (V) and Ca-rich FA (W):

- Silicious FA is a spherical pozzolanic material that mainly consists of reactive SiO2 and reactive Al2O3. The amount of reactive CaO is not allowed to be higher than 10% and the amount of reactive SiO2 has to be at least 25%.
- Ca-rich FA is a powdery pozzolanic and/or hydraulic material that mainly consists of reactive CaO, reactive SiO2 and Al2O3. The amount of reactive CaO is not allowed to be lower than 10% and the amount of reactive SiO2 has to be at least 25% when the amount of reactive CaO is in the range of 10 to 15%.12

The higher the amount of these three oxides the higher the ability of the FA to partic- ipate on the hydration, assumed a sufficient high degree of reactivity is reached. As discussed in chapter 1 pozzolans, even though natural ones, have already been used as hydraulic binders in ancient times. Nowadays their use arouses new interest in the course of the ongoing debates on CO2 emission. Since the use of FA changes the development of mechanical properties of hardened cement paste dramatically, the underlying process of the pozzolanic reaction and its interactions with the hydration reactions of OPC have to be understood. These mechanisms and, with the chemical activation, a method that is used in order to accelerate the pozzolanic reaction will be discussed.

2.2.1 The Pozzolanic Reaction and the Hydration of Fly Ash

The characterising reaction of any pozzolanic material is the cracking of the SiO2 or Al2O3 SiO2 framework of the glass through OH- , what is known as alkaline corrosion of silicate glasses[95]. This reaction is based on the alkalinity of the paste and the products of this decomposition react with lime to form C-S-H. The scheme of a pozzolanic reaction can be written as follows:

illustration not visible in this excerpt

Pozzolanic materials are composed not only out of SiO2 and therefore equation 2.9 gives only a very simple assumption. The following remarks will give an idea of the complexity of the pozzolanic reaction.

The first equation assumes that all pozzolanic reactions are lime reactions with siliceous pozzolanic materials. If it is assumed that the reaction of a pozzolan and lime produces the same hydration products as cement the following reaction is proceeding[29]:

illustration not visible in this excerpt

As the reactive phase is not only composed of silica the other compounds like alumina and ferric oxide participate in the reaction as well. OPC blended with FA shows a higher amount of AFm phases compared to pure OPC [25, 24]. This is caused by the additional alumina provided by the FA.

It is quite argumentative that the reactivity of FA is caused by its amorphous glass phase. Figure 2.3 shows the zonal structure of the glass matrix of a FA particle. There are different explanations that describe the pozzolanic reaction of FA. According to Blaschke [14] a film of pore solution is formed at the beginning of the hydration of the cement due to chemical shrinkage of the matrix. FA particles react with alkalies of the pore solution and form potassium stabilised silicate hydrate. This compound forms a film of about 0.3 μ m around the FA particles and a further reaction of the FA is prevented. On the outer surface of the film calcium hydroxide crystallises out of the pore solution and is slowly transformed into C-S-H. According to this model the silica of these C-S-H does not originate from the FA but from the pore solution. A graphical visualisation of this process is given in figure 2.4.

illustration not visible in this excerpt

Figure 2.3 - The zonal structure of the glass matrix of FA[28]

illustration not visible in this excerpt

Figure 2.4 - Modell of the pozzolanic reaction of FA as described by Blaschke[14]

Another theory developed by Rudert[88] claims that the film on the surface of the FA particles consists of calcium silicate hydrates out of the hydration of the clinker. The film is called contact zone and the observable seam of C-S-H grows perpendicular out of the pore solution.

Both theories describe the zonar structure of the shell on the FA surface which is formed out of reaction products.

However Blaschke claims the enhanced conversion of Ca(OH)2 to be a result of the fine dispersion of Ca(OH)2 which is caused by the additional nucleation space provided by the FA particles. Rudert, on the other hand, claims that the higher the number of crystallisation seeds the higher the number of early formed calcium silicate hydrates, the higher the decrease of Ca(OH)2 content.

Another interesting study on the mechanism of action of FA is given by Hüttl[47]. Within this investigation an artificial FA that consists out of an elevated content of the silicone isotope[29] Si in comparison to FA produced the common way in the dedusting systems of power plants was synthesised. Through the knowledge of the incorporated sili- cone isotopes it became possible to know from where the silicone in the different hydration products originates. This work revealed that[29] Si could be found in the matrix which is an evidence for the pozzolanic reaction of FA. While Blaschke and Rudert draw upon pictures gained via electron microscopy where FA particles seemed not to have developed any bond with the matrix, Hüttels’ investigations on polished sections via SEM showed that there in fact is a bond however it is not very distinct. This is in good correlation with Ogawa et al.[78] who stated that the bond between pozzolans and C-S-H is normally not very strong and strongly depends on the alkali content of the pozzolans.

2.2.2 Interactions between Ordinary Portland Cement and Fly Ash

Accordingly to equation 2.9, hydration of a blended OPC containing a pozzolan can be written in the following manner:

illustration not visible in this excerpt

Within the OPC hydration, lime, produced by the hydration of C3S and C2S, is being consumed in the reaction of the decomposed FA to C-S-H similar to that produced by the direct hydration of C3S and C2S.13 Another crucial aspect is that the pH necessary for reaction is relatively high with values of more than 13.0[11]. Indeed the pH in cement pastes reaches values up to about 13.0 to 13.8 with respect to the saturated calcium hydroxide solution and the dissolution of alkalies (KOH and NaOH)[101]. Within the OPC hydration CH is formed. A pozzolanic reaction consumes lime to form C-S-H and can therefore improve the intrinsic quality of the paste and, consequently, the quality of the concrete[66].

Uchikawa et al.[108] and Ogawa et al.[78] investigated the hydration of cement in the presence of pozzolans and described the mechanism between FA and C3S and between FA and C3A as can be seen in figure 2.5.

First, calcium ions dissolved from C3S are captured and adsorbed on the surface of the negatively charged pozzolan grains. C-S-H resulting from the hydration of C3S pretici- pates as C-S-H I on the surface of C3S and as C-S-H II on the surface of the pozzolan grains. Due to the dissociation of water, the pozzolan surfaces are attacked by OH- and because of the resulting dissolution of K+ and Na+, a Si- and Al-rich amorphous layer is formed on the surfaces. As more and more alkalies are dissolved, the whole process is accelerated which results in the thickening of the layer. It is assumed that owing to osmotic pressure resulting from the difference of concentration of ions the layer swells and a hollow space is formed in between layer and pozzolan grain. This hollow space is filled with Si, AI, Na+ and K+ rich liquid. With a certain osmotic pressure the layer is broken and SiO[4] -[4] and AlO-[2] diffuse outside. Together with Ca[2] + precipitation starts on the outer surface of the C3S grains. The remaining vacant space inside the film on the FA grains remains due to the high concentration of alkalies and the vitreous phase of the FA which contains corresponding high alkali concentrations causes the typical stripping off of the hydrate from the fly ash which can be observed.

illustration not visible in this excerpt

Figure 2.5 - Scheme of the hydration in a system with pozzolanic grains (e.g. FA) and C3S grains[78] (left) and C3A grains[108] in the presence of Ca(OH)2 and CaSO4 · H2O

Protonical attack starts right after the first contact of C3A with water due to OH- produced by dissociation of water. Ca[2] + is released and a layer rich of alumina is formed on the surface of the C3A grains. Due to osmotic pressure a hollow space appears between the hydrated particle and the layer. This space becomes filled with AlO-[2] -richsolution and near the outside of the layer the solution is rich with Ca[2] + and SO[2] -[4] which results from the dissolution of calcium sulfate and calcium hydroxide. Chemisorption of Ca-ions on the surface of the layer produces positively charged particles and temporarily stops the rate of C3A dissolution. SO[2] -[4] -ions are adsorbed on this charged C3A grains and protect the grains from attack of hydronium ions. With a certain osmotic pressure the

illustration not visible in this excerpt

speaking, the FA has to be decomposed which can be related to the alkaline reaction of the clinker phases in contact with water. The pH in the liquid phase of the cement paste is created by the dissipation of calcium hydroxide and the presence of alkalies in the solution. Since alkalies are essentially, their effect on the activation of FA has already been well examined. Y. Fan et al.[30] investigated the influence of Na2SO4 as an activation agent and reported that there is an acceleration effect on the pozzolanic reaction in both, the short term and the long term hydration, which leads to a strength increase in mortar prisms compared to mortars without FA acceleration. Other investigations attribute a simillar beneficial impact to Na2SO4 [53, 91].

3 Materials

3.1 Portland Cement

For all the investigated mixtures CEM I 42.5 R supplied from HTC Leimen was used. All analytical data on this cement as given in table 3.1 was supplied from HTC as well.

Table 3.1 - Chemical analysis and phase composition of CEM I 42.5 R

illustration not visible in this excerpt

The four main clinker phases were calculated according to Bogue as given in table 3.2.

Table 3.2 - Bogue calculation of the main clinker phases in [%]a

illustration not visible in this excerpt

a Considering the refinement according to Taylor

3.2 Fly Ash

Also the FA and the analytical data thereof were delivered by HTC Leimen. The FA was ground in the laboratory ball mill at HTC. All analytical data that classify the FA to be a type V FA (EN 197-1) can be found in table 3.3. An additional analysis was carried out by Empa to confirm the rather low content of amorphous phases. All data on this investigation are listed in appendix XIII.

Table 3.3 - Analysis of fly ash

illustration not visible in this excerpt

The composition of the glass phase was calculated from the XRD Results are listed in table 3.4.

Table 3.4 - Calculation of the composition of the glass phase of the FA.

illustration not visible in this excerpt

and XRF data.

As additional source of sulfate Microanhydrit A was used. This material was delivered by HTC from where the analytical data as given in table 3.5 originate as well.

Table 3.5 - Analysis of anhydrite

illustration not visible in this excerpt

For the investigations cubic C3A was used. In the following the synthesis is described. First the two educts CaCO3 and Al2O3 were mixed in the mass ratio CaCO3/Al2O3 = 2.95 and filled in a 2000 ml PE bottle where alumina ceramic mixing balls with a diameter of 5 mm were added for homogenisation purposes. This mixture was homogenised via placing the bottle on rotating rolls over night. After granulation the mixture was fired in a laboratory kiln according to the temperature regime figure 3.1 shows.

Figure 3.1 - Regime of the heat treatment during the synthesis of C3A

illustration not visible in this excerpt

In the course of this treatment a chemical reaction is taking place between the two reactants and C3A is formed according to equation 3.1.

illustration not visible in this excerpt

Synthesising C3A in a laboratory oven principally is comparably easy to perform and leads to C3A of high purity. As the bulk that results from one production process is only of about 300 gram different charges had to be prepared. If the heat treatment is not successful and there is not a complete conversion of the reactants into C3A different contaminations, e.g. mayenite (C12A7) or free lime (CaOfree), arise. To make sure this is not the case, XRD was performed on every charge and the impurities were ascertained by means of Rietveld refinement. The results are listed in table 3.6.

Table 3.6 - Rietveld analysis of synthesised C3A; data in [wt%]

illustration not visible in this excerpt

The material was ground in the laboratory ball mill at Empa in order to receive the requested sepecific surface. The ground C3A was stored under sealed conditions in a climate cabinet at 24°C and 27 % relative humidity (RH) until it was used.

illustration not visible in this excerpt

4 Sample Preparation and Methods

4.1 Mix Design

In order to achieve the intended objective of improving early strength properties several mixtures of FA-blended CEM I 42.5 R were submitted to examination. Table 4.1 shows the compositions of each mixture and the experimental matrix respectively. This matrix can be divided into three groups of mixtures whereby group one reflects the effect of additions of FA, AH and AH plus C3A (mixtures 1, 3 and 4) while group two displays the influence of additions of AH plus C3A in different amounts upon the presence of FA (mixtures 6, 7 and 8) and group three deals with system activation via Na2SO4 (mixtures 2, 5 and 9).

Table 4.1 - Compositions of the mixtures submitted to examination

illustration not visible in this excerpt

All these mixtures were prepared in the following way: First the various components were weighed to give a total bulk of 1400 g. This amount resulted from the sum of the amount needed for every single sample to be prepared, e.g. for prisms or XRD samples plus about 200 additional grams. The bulk was filled in PE bottles of 2000 ml volume and subsequently mixed for two hours under the use of two alumina ceramic mixing balls of 30 mm diameter in a WAB Turbula mixer, Type T2A from Willy A. Bachofen Maschinenfabrik Basel, Switzerland. After mixing the bulk was decanted in two smaller PE vessels of 1000 ml volume. These vessels were sealed and stored in an air-conditioned room at 20°C and 35% RH until the samples were prepared. Every single sample needed for analysis was prepared out of these mixtures.

4.2 Procedures of Sample Preparation

4.2.1 Paste Samples

For TGA and SEM analysis paste samples were prepared for all mixes and all inspection dates with a w/c-ratio of 0.45. A Twister Evolution vacuum mixing device from Renfert was applied for preparation. The mixing container was filled with 200 gram of the dry mixtures and 90 gram deionised water tempered to approximately 20°C. Subsequently the tank was connected with the mixing device and vacuum was applied. In the first 15 seconds of mixing the binder was roughly mixed with the water. Thereupon proper mixing followed within the next 105 seconds. During the latter period the direction of the mixing tool was changed every 15 seconds. This whole procedure led to a total mixing time of 120 seconds. The pastes were filled into small PE vessels of 12 ml volume immediately after the mixing and sealed air tight with a PE cap and Parafilm®, a synthetic film consisting of equal shares of paraffin and PE. These vessels were stored in an air-conditioned room at 20°C and 90 % RH until the corresponding inspection date was reached. The samples for investigation via XRD were prepared according to the same procedure but in larger PE vessels with an inner diameter of 32 mm in order to guarantee the measurement not to be limited by divergence.

4.2.2 Mortar Samples

For the investigation of flexural and compressive strength mortar samples were prepared according to EN 196-1[3]. The mortar was produced out of one part of the corresponding mixture, three parts of standardised sand and half a part of deionised water leading to a w/c of 0.50[15]. Expressed in mass percentage rate this means the mix for each mould consisted of (450 ± 2g) cement, (1350 ± 5g) sand and (225 ± 1g) water. Mixing took place in a Type 1551 mixer from Toni Technik, Berlin, Germany. Immediately after the beginning of the filling of the mould which was fixed to a vibrating table the table was started and the moulds were filled completely in two steps. After covering these moulds with glass plates the storage took place in a climate cabinet at 20°C and 90 % RH. For each mixture 18 prisms under the use of two moulds that supplied 9 prisms, each with a size of (25 · 25 · 100) mm[3], were prepared for assessed inspection dates of 1, 2, 7, 28 and 90 days.[16] Prisms for inspection after 1, 2, and 28 days were prepared quadruple whereas the inspection for all other dates took place on three prisms respectively. 20 minutes prior to the first inspection after 1 day the prisms were demoulded. Prisms for later inspection dates were stored under water at 20°C until inspection.

4.3 Investigation of Flexural and Compressive Strength

For the detection of both, flexural and compressive strength, a testing machine from Walter + Baig AG, Löhningen, Switzerland, Typ LFM50, 50kN was used. Prior to measurement the prisms were measured in terms of height, width, length and weight.

Flexural strength was detected by applying a 3-point bending test where the test jig consists of two parallel supports for the sample and a single loading pin in the middle between the supports through which the force is introduced. In order to minimise the influence of friction on the measurement the support pins was mounted in such a way that they can rotate freely on their axis. The force was introduced perpendicularly to the specimen plane and/or to the support plane with a rate of increase of the load of (50 ± 10) N/s.

Compressive strength was tested on both parts of the broken prisms. A special device that allowed to apply pressure perpendicular to the specimen on a square base of (20 · 20) mm[2] was used therefore with a rate of increase of the load of (2400 ± 200) N/s. The investigation of mechanical properties was realised according to the requirements of EN 196-1[3].

4.4 Isothermal Calorimetry

The isothermal (heat conduction) calorimetry (IC) is a common method to measure the heat flow as a function of time. As the hydration reactions of cement consist of exother- mal and endothermal reactions the temperature of a corresponding sample will change. This gives rise to a heat flow which is why this technique has long found use in cement research[112]. The hydration process can be followed continuously at various tempera- tures whereas the overall rate of heat at certain points indicates the rate of reaction. A detailed review on the use of IC especially in the field of cement research and cement chemistry can be found in[113].

In this work a properly calibrated TAM Air isothermal calorimeter from Thermometric AB, Sweden, was used. It uses the principle of 8 parallel twin measurement channels that both consist of two cells. One of the cells is dedicated to the sample and the other one to a reference vessel. The heat from both, the sample and the reference vessel is conducted away to a heat sink to keep the sample temperature essentially constant. The cells are in contact with a heat flow sensor which in turn is in contact with the heat sink. What is actually been measured by the sensor is the heat flow from the sample, which for the major part of the cement hydration measurement is equal to the heat production rate in the sample, also called the thermal power P expressed in Watt. Figure 4.1 shows the principled setup of a corresponding facility. The reference vessel is used to reduce the signal to noise ratio and furthermore to correct measurement and temperature artifacts. Both, the sample and the reference container were assembled with 20 ml glass flasks.

[...]


1 Rapid set describes the already discussed phenomenon of rapid hardening of a cement paste when there is no or not enough sulfate present to delay the setting.

2 It is not clear that the formation of an ettringite layer is really responsible for the deceleration of the C3A hydration. Investigations from Minard et al.[73] and Jansen et al[48] show that the dissolution of C3A is rather stopped by the formation of an amorphous Al(OH)3-layer or adsorbed sulfate ions.

3 The pozzolanic reaction is discussed in chapter 2.2.1.

4 The fragmentation in stages happens according to the following nomenclature. Early stage (periods I and II), middle stage (periods III and IV) and late stage (period V). This subsumption is used to discuss the formation of microstructure in pastes in chapter 2.1.2.

5 Usually this is the first period to be displayed within a curve that shows the development of heat of hydration and the initial peak is not considered. Nevertheless the hydration reaction starts immedi- ately after the first contact of the cement with water and consequently an initial peak always exists. As the very early reaction is extraordinary exothermic and fast, it is somewhat difficult do detect it but as discussed in chapter 4.4 with isothermal calorimetry and in-situ mixing of the used paste samples this initial peak can be detected. It is possible to label this upstream period as period 0 as can be seen from[37].

6 There are some interactions that occur during hydration. For example C3A and C4AF both consume sulfate ions, but as C3A is more reactive it consumes SO[2] -[4] more rapidly.

7 Besides the early reaction of C3A the reaction of free lime that takes place at this early stage as well is another contributor to the heat produced in the first minutes of the hydration[75].

8 Figure 2.1 only displays the first four periods as it only contemplates the early hydration up to 22 hours.

9 The first morphology is known as C-S-H I with a C/S-ratio from 0.8 to 1.5 and the latter is known as C-S-H II with a C/S-ratio from 1.0 to 2.0 [100]

10 In contrast to the deposition of C-S-H on the inside of the shells hollow grains are likely to be found in hydrating cement pastes. This observation is not consistent with the hydration model of Powers[84] who stated that hydration products deposit both inside the original cement grain boundary (inner product) and outside in the original water filled spaces between the cement grains (outer product). These hollow grains are a characteristic feature in cement pastes and were first described by Hadley [39], hence the name ’Hadley’ grains.

11 The solubility describes the maximum amount of a certain matter that is to be solved in another matter. The solubility rate on the other hand describes the velocity at which the process of solution takes place.

12 Ca-rich FA is very rarely used as SCM due to its generally high compositional variability

13 The amount of FA or pozzolans is limited by the corresponding standards (e.g. EN 197-1[4] ) in order to avoid systems that are free from CH. The absence of CH would cause a low pH with the danger of steel corrosion in reinforced concrete.

Details

Pages
144
Year
2012
ISBN (eBook)
9783656258568
ISBN (Book)
9783656259480
File size
21.1 MB
Language
English
Catalog Number
v199418
Institution / College
TU Bergakademie Freiberg – Institut für Glas, Keramik und Baustofftechnik
Grade
1,3
Tags
study portland

Author

Share

Previous

Title: Study of hydration processes of Portland cements blended with supplementary cementitious materials