USF Libraries
USF Digital Collections

Influence of C₃S content of cement on concrete sulfate durability

MISSING IMAGE

Material Information

Title:
Influence of C₃S content of cement on concrete sulfate durability
Physical Description:
Book
Language:
English
Creator:
Shanahan, Natalya G
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla.
Publication Date:

Subjects

Subjects / Keywords:
compressive strength
cement composition
alite
sulfate attack
expansion
Dissertations, Academic -- Civil Engineering -- Masters -- USF   ( lcsh )
Genre:
government publication (state, provincial, terriorial, dependent)   ( marcgt )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Summary:
ABSTRACT: The influence of tricalcium silicate content of cement on concrete durability has long been a topic of discussion in the literature. The objective of this investigation was to determine whether increasing tricalcium silicate content of cement has a negative effect on concrete sulfate durability. Several mill certificates were reviewed to select cements with similar tricalcium aluminate content and variable tricalcium silicate contents. Cements selected for this study were randomly labeled as cements C, D, D2, E, and P. The following properties were assessed for the as-received cements: Blaine fineness, particle size distribution, chemical oxide content, and mineralogical content. Three different methods were employed to determine the mineralogical composition of the as-received cements: Bogue calculation, internal standard method, and Rietveld refinement analysis. Despite the attempt to select cements with similar composition, it was determined that the as-received cements had compositional differences other than their C3S content. These cements had a variable tricalcium aluminate and alkali content, as well as differences in the amount and form of calcium sulfates. In order to eliminate these variances, doped cements were prepared by increasing the C3S content of the as received cements to 69 % by Bogue calculation. Durability of as-received cements and doped cements was assessed through several measurements including length change, compressive strength, and phase transformation in sodium sulfate solution. For as-received cements, compressive strength of mortar cubes stored in saturated lime solution was evaluated as well. Semiquantitative x-ray diffraction analysis and scanning electron microscopy observations were performed on mortar bars to evaluate the relative amounts and morphology of the hydrated phases. It was concluded at the end of this study that cements with high tricalcium silicate content generally have poor durability in sodium sulfate environment. All the cements experienced higher expansion with increased C3S content. High C3S content combinedwith high C3A content was particularly detrimental to mortar resistance to sodium sulfate attack.
Thesis:
Thesis (M.S.C.E.)--University of South Florida, 2003.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Natalya G. Shanahan.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 100 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
oclc - 54958998
notis - AJQ2293
usfldc doi - E14-SFE0000238
usfldc handle - e14.238
System ID:
SFS0024934:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 2200421Ka 4500
controlfield tag 006 m d s
007 cr mn
008 040406s2003 flua sbm s000|0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0000238
035
(OCoLC)54958998
9
AJQ2293
b SE
040
FHM
c FHM
090
TA145
1 100
Shanahan, Natalya G.
0 245
Influence of CS content of cement on concrete sulfate durability
h [electronic resource] /
by Natalya G. Shanahan.
260
[Tampa, Fla.] :
University of South Florida,
2003.
502
Thesis (M.S.C.E.)--University of South Florida, 2003.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
500
Title from PDF of title page.
Document formatted into pages; contains 100 pages.
520
ABSTRACT: The influence of tricalcium silicate content of cement on concrete durability has long been a topic of discussion in the literature. The objective of this investigation was to determine whether increasing tricalcium silicate content of cement has a negative effect on concrete sulfate durability. Several mill certificates were reviewed to select cements with similar tricalcium aluminate content and variable tricalcium silicate contents. Cements selected for this study were randomly labeled as cements C, D, D2, E, and P. The following properties were assessed for the as-received cements: Blaine fineness, particle size distribution, chemical oxide content, and mineralogical content. Three different methods were employed to determine the mineralogical composition of the as-received cements: Bogue calculation, internal standard method, and Rietveld refinement analysis. Despite the attempt to select cements with similar composition, it was determined that the as-received cements had compositional differences other than their C3S content. These cements had a variable tricalcium aluminate and alkali content, as well as differences in the amount and form of calcium sulfates. In order to eliminate these variances, doped cements were prepared by increasing the C3S content of the as received cements to 69 % by Bogue calculation. Durability of as-received cements and doped cements was assessed through several measurements including length change, compressive strength, and phase transformation in sodium sulfate solution. For as-received cements, compressive strength of mortar cubes stored in saturated lime solution was evaluated as well. Semiquantitative x-ray diffraction analysis and scanning electron microscopy observations were performed on mortar bars to evaluate the relative amounts and morphology of the hydrated phases. It was concluded at the end of this study that cements with high tricalcium silicate content generally have poor durability in sodium sulfate environment. All the cements experienced higher expansion with increased C3S content. High C3S content combinedwith high C3A content was particularly detrimental to mortar resistance to sodium sulfate attack.
590
Adviser: Abla M. Zayed
653
compressive strength.
cement composition.
alite.
sulfate attack.
expansion.
690
Dissertations, Academic
z USF
x Civil Engineering
Masters
773
t USF Electronic Theses and Dissertations.
949
FTS
SFERS
ETD
TA145 (ONLINE)
nkt 4/12/04
4 856
u http://digital.lib.usf.edu/?e14.238



PAGE 1

Influence of C3S Content of Cement on Concrete Sulfate Durability by Natalya G. Shanahan A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Civil Engineering Department of Civil and Environmental Engineering College of Engineering University of South Florida Major Professor: Abla M. Zayed, Ph.D. Rajan Sen, Ph.D. Jeffrey G. Ryan, Ph.D. Date of Approval: December 15, 2003 Keywords: alite, cement composition, sulfate attack, expansion, compressive strength Copyright 2004, Natalya G. Shanahan

PAGE 2

ACKNOWLEDGMENTS The author would like to express her appreciation to Dr. Abla M. Zayed, her major professor and chairman of the superv isory committee, for her advice and guidance throughout this study. The author would also like to thank Dr. Rajan Sen and Dr. Jeffrey Ryan for serving on her supervisory committee. In addition, the author would like to thank Anthony Greco for his assistance during the course of this investigation. This work was supported by the funds from the Florida Department of Transportation and the University of S outh Florida Department of Civil and Environmental Engineering. The conclusions and opinions expressed in this report are those of the author and not necessarily of the funding agencies.

PAGE 3

i TABLE OF CONTENTS LIST OF TABLES iii LIST OF FIGURES iv LIST OF SYMBOLS AND ABBREVIATIONS vii ABSTRACT ix CHAPTER 1. INTRODUCTION 1 1.1 Research Objective 1 1.2 Clinker and Cement 2 1.2.1 Mineral Phases 3 1.2.2 Cement Hydration 5 1.3 Sulfate Attack 8 1.3.1 Expansion due to Ettringite Formation 10 1.3.1.1 Increase of Solid Volume 10 1.3.1.2 Swelling Hypothesis 11 1.3.1.3 Crystallization Pressure 12 1.3.1.4 Oriented Crystal Growth 13 1.3.2 Expansion due to Gypsum Formation 14 1.3.3 Expansion due to Monosulfoaluminate Conversion 16 1.4 Review of Previous Research 16 CHAPTER 2. EXPERIMENTAL METHODS 24 2.1 Chemicals 24 2.2 C3S (Alite), C3A, C4AF, and MgO 23 2.3 Cement 24 2.3.1 Blaine Fineness and Particle Size Distribution 24 2.3.2 Oxide Chemical Composition 25 2.3.3 Mineralogical Composition 25 2.3.3.1 Bogue Calculation 25 2.3.3.2 Internal Standard Method 25 2.3.3.3 X-Ray Powder Diffraction Analysis 27 2.4 Sand 29 2.5 Mortar Durability Tests 29 2.5.1 Mortar Cubes 29

PAGE 4

ii 2.5.2 Mortar Bar Preparation 31 2.5.3 X-Ray Powder Diffraction Analysis of Mortar Bars 31 2.5.4 Scanning Electron Microscopy 32 CHAPTER 3. EXPERIMENTAL RESULTS AND DISCUSSION 33 3.1 Physical Characteristics of As-Received Cements 33 3.1.1 Blaine Fineness 33 3.1.2 Particle Size Distribution 34 3.2 Oxide Chemical Content 36 3.3 Mineralogical Composition 37 3.3.1 Bogue Mineralogical Content 37 3.3.2 Internal Standard Method 38 3.3.3 Rietveld Refinement 46 3.4 Optical Microscopy 51 3.5 Durability 55 3.5.1 Length Change of Mortar Bars in Na2SO4 Solution 56 3.5.2 Compressive Strength of Mortar Cubes 63 3.5.3 X-Ray Diffraction Analysis of Mortar Bars 68 3.5.4 SEM 70 CHAPTER 4. CONCLUSIONS AND RECOMMENDATIONS 74 REFERENCES 76 APPENDICES 81 Appendix A. Calibration Curves 82 Appendix B. Rietveld Refinement Profiles 84 Appendix C. SEM Images 87

PAGE 5

iii LIST OF TABLES Table 1. Mix Proportions for As-Received Cements 30 Table 2. Mix Proportions for C3S-Doped Cements 31 Table 3. Blaine Fineness 34 Table 4. Oxide Chemical Composition of As-received Cements 37 Table 5. Bogue Mineralogical Content 38 Table 6. Phase Composition of As-Received Cements Based on the Internal Standard Method 45 Table 7. Rietveld Refinement Resu lts for the Ground As-Received Cements 50 Table 8. Length Change of Mortar Bars Prepared with As-Received Cements at 300 Days 58 Table 9. Compound Composition of As-Received and Doped Cements Based on Bogue, Internal Standard, and Rietveld Refinement Methods 60 Table 10. Expansion and Ettringite Intensity Ratios (Bulk) for Mortar Bars Prepared with As-Received Cements 69

PAGE 6

iv LIST OF FIGURES Figure 1. Particle Size Distribution for As-Received Cements 34 Figure 2. Particle Size Distribution for Cement C 35 Figure 3. Particle Size Distribution for Cement D2 35 Figure 4. Particle Size Distribution for Cement E 36 Figure 5. Particle Size Distribution for Cement P 36 Figure 6. Five Characteristic Angles of the Laboratory-Prepared Alite Used for the Monoclinic Polymorph Identification 39 Figure 7. Characteristic Angles Used for the Monoclinic Alite Polymorph Identification 40 Figure 8. Characteristic Angles Used for the Monoclinic Alite Polymorph Identification 41 Figure 9. Established C3S Calibration Curve 42 Figure 10. Angular Range (2 = 18-24 ) Confirming the Presence of Cubic C3A in Cements 43 Figure 11. Angular Range (2 = 29-36 ) Confirming the Absence of Orthorhombic C3A in Cements 44 Figure 12. Mumme Sublattice Structur e of the Monoclinic Alite 47 Figure 13. Nishi Superlattice Struct ure of the Monoclinic Alite 47 Figure 14. Rietveld Refinement for As-Received Cement C 48 Figure 15. Reflected Light Micros copy Images of Clinker C 51 Figure 16. Reflected Light Micros copy Images of Clinker D2 52

PAGE 7

v Figure 17. Reflected Light Micros copy Images of Clinker E 53 Figure 18. Reflected Light Micros copy Images of Clinker P 54 Figure 19. Length Change of Bars Prep ared with As-Received Cements 57 Figure 20. Expansion of Mortar Bars Prepared with As-Received Cements at 300 Days vs. C3S Content of Cements Determined by Rietveld Refinement 59 Figure 21. Expansion of Mortar Bars Prepared with As-Received Cements at 300 Days vs. C3S/ C2S Ratio of Cements Determined by Rietveld Refinement 59 Figure 22. Expansion of As-Received Ceme nt C and Doped Cement C-69D 61 Figure 23. Expansion of As-Received Cement D2 and Doped Cement D2-69D 61 Figure 24. Expansion of As-Received Ceme nt E and Doped Cement E-69D 62 Figure 25. Expansion of As-Received Cement P and Doped Cement P-69D 62 Figure 26. Expansion of Doped Cements 63 Figure 27. Compressive Strength of Mortar Cubes Prepared with As-Received Cements Stored in Saturated Lime Solution 64 Figure 28. Compressive Strength of Mortar Cubes Prepared with Cement C, Stored in Lime (C) and Sodium Sulfate (CS) Solutions 65 Figure 29. Compressive Strength of Mortar Cubes Prepared with Cement D2, Stored in Lime (D2) and Sodi um Sulfate (D2S) Solutions 65 Figure 30. Compressive Strength of Mortar Cubes Prepared with Cement E, Stored in Lime (E) and Sodium Sulfate (ES) Solutions 66 Figure 31. Compressive Strength of Mortar Cubes Prepared with Cement P, Stored in Lime (P) and Sodi um Sulfate (PS) Solutions 66 Figure 32. Compressive Strength of Mortar Cubes Prepared with As-Received Cements Stored in Sodium Sulfate Solution 67

PAGE 8

vi Figure 33. Compressive Strength of Mortar Cubes Prepared with As-Received Cement P and Doped Cement P-69D Stored in Sodium Sulfate Solution 68 Figure 34. XRD of Bulk Sections of Mortar Bars Prepared with As-Received Cements 69 Figure 35. XRD of Outside Surfaces of Mortar Bars Prepared with AsReceived P Cement and D oped P-69D Cement 70 Figure 36. Ettringite Spherulite in the C Bar Cross-Section 71 Figure 37. Ettringite Crystals in the Cross-Section of Bar P-69D 72 Figure 38. Needle-Like Ettringite Crystals in the Cross-Section of Bar E 72 Figure 39. Calibration Curve for Cubic C3A 82 Figure 40. Calibration Curve for C4AF 82 Figure 41. Calibration Curve for MgO 83 Figure 42. Rietveld Refinement for Cement D2 84 Figure 43. Rietveld Refinement for Cement E 85 Figure 44. Rietveld Refinement for Cement P 86 Figure 45. Ettringite Spherulite in the D2 Bar Cross-Section 87 Figure 46. Ettringite Spherulite in the P Bar Cross-Section 88 Figure 47. Ettringite Spherulite in the P-69D Bar Cross-Section 88

PAGE 9

vii LIST OF SYMBOLS AND ABBREVIATIONS AASHTO American Association of State Highway and Transportation Officials ASTM American Society for Testing and Materials EDS Energy Dispersive Spectroscopy SAM Salicylic Acid/Methanol Extraction SEM Scanning Electron Microscope XRD X-Ray Diffraction QXRD Quantitative X-Ray Diffraction Cement Chemistry Abbreviations A Alumina, Al2O3 C Calcium Oxide, CaO F Ferric Oxide, Fe2O3 H Water, H2O S Silica, SiO2 S Sulfur Trioxide, SO3 C3A Tricalcium Aluminate, 3CaOAl2O3 C4AF Tetracalcium Aluminoferrite, 4CaOAl2O3Fe2O3 C2S Dicalcium Silicate, 2CaOSiO2

PAGE 10

viii C3S Tricalcium Silicate, 3CaOSiO2 CH Calcium Hydroxide, Ca(OH)2 C S H2 Gypsum, Ca2SO42H2O C S H0.5 Bassanite, Ca2SO40.5H2O C S Anhydrite, Ca2SO4 C-S-H Calcium Silicate Hydrate, nCaO SiO2mH2O C6A S H32 Ettringite, 3CaOAl2O33CaSO432H2O C4A S H12 Monosulfoaluminate, 3CaOAl2O3CaSO412H2O

PAGE 11

ix INFLUENCE OF C3S CONTENT OF CEMENT ON CONCRETE SULFATE DURABILITY Natalya G. Shanahan ABSTRACT The influence of tricalcium silicate content of cement on concrete durability has long been a topic of discussion in the literature The objective of this investigation was to determine whether increasing tricalcium silicate content of cement has a negative effect on concrete sulfate durability. Several mill certificates were reviewed to select cements with similar tricalcium aluminate content and variable tricalcium silicate contents. Cements selected for this study were randomly labeled as cements C, D, D2, E, and P. The following properties were assessed for the as-received cements: Blaine fineness, particle size distribution, chemical oxide content, and mineralogical content. Three different methods were employed to determine the mineralogical composition of the as-received cements: Bogue calculation, internal standard method, and Rietveld refinement analysis. Despite the attempt to select cements with similar composition, it was determined that the as-received cements had compositional differences other than their C3S content. These cements had a variable tricalcium aluminate and alkali content, as well as differences in the amount and form of calcium sulfates. In order to eliminate these variances, doped cements were prepared by increasing the C3S content of the as received cements to 69 % by Bogue calculation.

PAGE 12

x Durability of as-received cements and doped cements was assessed through several measurements including length change, compressive strength, and phase transformation in sodium sulfate solution. For as-received cements, compressive strength of mortar cubes stored in saturated lime solution was evaluated as well. Semiquantitative x-ray diffraction analysis and scanning electron microscopy observations were performed on mortar bars to evaluate the relative amounts and morphology of the hydrated phases. It was concluded at the end of this study that cements with high tricalcium silicate content generally have poor durability in sodium sulfate environment. All the cements experienced higher expansion with increased C3S content. High C3S content combined with high C3A content was particularly detrimental to mortar resistance to sodium sulfate attack.

PAGE 13

1 CHAPTER 1. INTRODUCTION 1.1 Research Objective Sulfate attack occurs in concrete when concrete comes in contact with a source of sulfate ions, which can be groundwater, seawater, soil, or rainwater. Sulfate attack usually manifests itself by cracking and spalling of concrete accompanied by expansion and/or loss of strength. The resistance of concrete to sulfate attack is determined by several factors, such as water/cement ratio, permeability, and cement characteristics, which include fineness and cement composition. It has long been recognized that controlling cement composition, specifically tricalcium aluminate content, improves concrete resistance to sulfate attack. Both ASTM C150 and AASHTO M85 limit the C3A content to 8 % for Type II cement, which is designated as a moderate sulfate resistant cement. AASHTO M85 also limits the maximum C3S content of this cement to 55 %, while ASTM only limits the C3A. Currently, there is an effort to unify these two standards. The objective of this research was to verify that high C3S content of cements results in decreased durability of concrete in sulfate environment.

PAGE 14

2 1.2 Clinker and Cement British Standard BS 6100 Section 6.1 defines clinker as a “solid material formed in high temperature processes by total or partial fusion.” European Prestandard ENV 197-1 specifies that clinker used for the production of portland cement has to be a hydraulic material consisting of calcium sili cates, aluminum and iron oxides, and other oxides [1]. In clinker production, cement plants use limestone, seashells, and chalk for calcareous compounds; clay, shale, slate, and sand are sources of silica (SiO2) and alumina (Al2O3); and iron ore renders iron compounds. The particular combination of the raw feed determines the chemical composition and type of cement that is being produced. The heat treatment of the raw feed is called clinkering. Clinkering is different from sintering and fusion because only partial melting occurs in a cement kiln. A cement kiln is a long steel cylinder that is lined with refractory brick and is slightly inclined. This inclination together with rotation (usually 60 to 200 rev/h) helps move clinker along the length of the kiln, which is divided into several zones: dehydration, calcination, clinkering, and cooling. Water is driven off in the dehydration zone and the raw feed is heated up to about 600C when it enters the calcination zone. In this zone, the raw feed becomes reactive, and calcium aluminates and ferrites are formed through the solid-state reactions around 1200C. The clinkering zone begins at about 1350C, when calcium silicates are formed during partial melting. At the end of the clinkering zone, temperatures reach 1400 to 1600C. The formation of the clinker is completed in the cooling zone where it starts to solidify. The rate of cooling in this zone is very important because it determines at what rate the clinker compounds crystallize, and thus their final reactivity [2].

PAGE 15

3 Clinker comes out of the kiln in the form of dark gray nodules that are 6 to 50 mm in diameter. Before clinker can become a useful material, it needs to be ground to a fine powder. Clinker is transported to ball-mills where it is interground with small amounts of calcium sulfates in the form of gypsum (Ca2SO42H2O), bassanite (Ca2SO40.5H2O), anhydrite (Ca2SO4), or any combination thereof, a nd other grinding agents. Anhydrite can be present in two forms, as “soluble” anhydrite, or -Ca2SO4, and “insoluble” or “deadburnt” anhydrite. Both Ca2SO40.5H2O and -Ca2SO4 have higher solubility than gypsum [3]. The solubility of “dead-burnt” anhydrite is extremely low. If grinding temperatures in the mill exceed 115-130C, gypsum will dehydrate to hemihydrate, which can lead to false setting of concrete [1, 3]. The presence of calcium sulfates is necessary to prevent early hydration of tricalcium aluminate and to optimize concrete strength. The final mixture of clinker and calcium sulfate, usually 95 % of clinker and 5 % of calcium sulfate, is called cement [2]. 1.2.1 Mineral Phases The major compounds formed in the clinker are tricalcium silicate (3CaOSiO2), dicalcium silicate (2CaOSiO2), tricalcium aluminate (3CaOAl2O3), and tetracalcium aluminoferrite (4CaOAl2O3Fe2O3). The shorthand notation for these compounds as used in cement industry is C3S, C2S, C3A, and C4AF respectively. C3S and C2S are responsible for strength development; C3A and C4AF control setting and heat evolution during hydration. C3S forms when calcium oxide combines with silica; it crystallizes from clinker liquid at about 1450C. C3S has several polymorphs with three different crystal structures: triclinic, monoclinic, and rhombohedral. In pure C3S, the triclinic form exists

PAGE 16

4 up to 980C, which at this temperature converts to monoclinic, which in turn converts to rhombohedral when temperature exceeds 1070C. Therefore, pure compound cooled to room temperature, should be in the triclinic form. However, in clinker C3S structure commonly incorporates such impurity atoms as Mg2+, Al3+ or Fe3+. The presence of these substitution ions can cause the higher temperature polymorphs to remain, even when clinker is cooled to room temperature [3]. Therefore, in portland cement, C3S is most frequently observed in the monoclinic form. The C3S found in cement is often referred to as alite. The term alite implies the presence of impurities in the C3S, but does not necessarily imply any information about a particular crystal structure. The reactivity differences of various C3S polymorphs have been a point of controversy for some time. Currently, there is no agreement in the scientific community on this topic [1], although the modifications including substitutional ions are thought to be more reactive than the pure compound [3]. C2S also has several polymorphs that have hexagonal, orthorhombic, and monoclinic structures. -C2S and -C2S, which are respectively monoclinic and orthorhombic, can exist up to 670C at normal pressure. Up to 1160C, orthorhombic ’ L-C2S is present. Above that temperature, dicalcium silicate can exist either in the form of ’ HC2S, which is orthorhombic, or -C2S, which is hexagonal. The monoclinic structure is the stable form of dicalcium silicate at room temperature. The form of C2S that is most commonly present in clinker is -C2S [4]. The research on the reactivity of different C2S polymorphs has also been inconclusive [3]. C3A can exist in the cubic, orthorhombic, and monoclinic forms. The pure C3A is cubic and does not have any polymorphs. In clinker, it is rather common for Na+ ions to

PAGE 17

5 substitute for Ca2+ ions in the C3A structure. When the Na2O concentration in the structure exceeds 3.7 %, but is below 4.6 %, C3A structure becomes orthorhombic. Above 4.6 %, C3A is monoclinic. Both the cubic and the orthorhombic C3A are common in clinker. Cubic C3A is more reactive than the orthorhombic, which in turn is more reactive than the monoclinic C3A. In short, C3A reactivity decreases with increasing Na2O content [3]. It should also be noted that C3A becomes less reactive in the presence of calcium hydroxide [1]. C4AF also does not have any polymorphs [4]. The chemical formula C4AF is an approximation of the actual chemical composition of the phase present in clinker, as it exists in the form of a solid solution series [1]. The reactivity of this phase generally increases with increasing Al/Fe ratio [3]. 1.2.2 Cement Hydration Reaction of cement with water is called hydration. Cement hydration consists of a series of complex chemical reactions that proceed at the same time. Alite reacts with water to form poorly crystalline calcium silicate hydrate gel and calcium hydroxide, also called portlandite. The abbreviated notation for these compounds is C-S-H and CH. 3CaOSiO2 + (3 + m n)H2O (1) nCaO SiO2mH2O + (3 n)Ca(OH)2 The effect of alkalis present in cement on C3S reactivity has been a topic of controversy among the researchers [1, 5, 6]. Reaction of belite with water is very similar to that of alite, only is proceeds at a much slower rate. The reaction products of belite hydration are also C-S-H and CH, although hydration of belite produces 2.2 times less CH than alite hydration [7].

PAGE 18

6 Tricalcium aluminate reacts with calcium sulfate and water to form ettringite. 3CaOAl2O3 + 3CaSO42H2O + 26H2O (2) 3CaOAl2O33CaSO432H2O If C3A is more reactive than calcium sulfate, it will form calcium aluminate m onosulfate hydrate according to reaction (3) or calcium aluminate hydrate according to reaction (4), depending on the differences in reactivity. 2(3CaOAl2O3) + 3CaOAl2O33CaSO432H2O + 4H2O (3) 3(3CaOAl2O3CaSO412H2O) 3CaOAl2O3 + Ca(OH)2 + 12H2O 4CaOAl2O313H2O (4) Reaction (3) can also take place if the supply of sulfate ions is exhausted before all the C3A has hydrated [1]. Although the exact mechanism of ettringite formation is not known, many researchers speculate that ettringite is formed topochemically or by a through-solution mechanism. Topochemical reactions are defined as reactions that “form a true crystalline overgrowth on the reactant with matching d-spaces … in the space occupied by the reactant ” However, the strict definition of topochemical reactions is rarely used in cement chemistry. Rather, any reaction during which “the reaction product precipitates only within the space that became available by the dissolution process or in its immediate vicinity” is considered to be topochemical. In through-solution reactions products are transported away from the reactants and precipitate from solution at random [8]. The particular mechanism of ettringite formation seems to be connected to the lime or Ca(OH)2 concentration in the pore solution. Some scientists think that hydroxyl ion concentration affects the mechanism of ettringite formation [9], while others suggest that

PAGE 19

7 calcium ions, and not the hydroxyl ions, influe nce the mode of ettringite formation [10, 11]. Lime and alkalis can serve as a source of OH[9] and Ca2+ ions as well as CH produced during the hydration of calcium silicat es. Presumably, formation of ettringite proceeds as a through-solution reaction at lime concentrations up to 0.020 M and as a topochemical reaction at higher concentrations [9]. An increase in Ca(OH)2 concentration forces ettringite to deposit on top of the C3A grains as the solution becomes supersaturated with respect to ettringite [8]. Many researchers support the theory that it is the topochemical ettringite that causes expansion and not the through-solution ettringite [8, 9]. It should be noted, that the paste can accommodate any volumetric changes that might occur due to ettringite formation while it is still in the plastic state. This ettringite formation is not disruptive. However, any volumetric changes that occur after the paste has hardened are damaging to the concrete matrix. High Ca(OH)2 concentrations also have an affect on the size of ettringite [12,13]. Mehta [12] observed that ettringite crystals formed in the absence of lime were at least 6 times longer and twice as thick as those formed in the presence of lime after 24 hours. Mehta described ettringite crystals formed in the presence of lime (1 micron long and 0.25 microns thick) as “colloidal” or “microcrystalline.” In the absence of lime, ettringite precipitated throughout the sample, forming long needle-like crystals arranged in rosettes and spherulites in cavities and “compacted bundles” of prismatic crystals on top of the C3A grains. This random formation of ettringite is characteristic of a through-solution mechanism. Addition of lime decreased both the size and the rate of formation of ettringite [13].

PAGE 20

8 Odler [5] observed that sodium oxide decreases the rate of initial C3A hydration as well as the rate of gypsum consumption. Potassium hydroxide, on the other hand, accelerates both of these processes as well as increases the transformation rate of ettringite to monosulfoaluminate. However, Taylor [3] stat es that the presence of alkalis, particularly KOH, decreases ettringite formation. The presence of alkali sulfates does not seem to affect the rate of C3A hydration, but it does increase gypsum consumption and ettringite to monosulfoaluminate transformation. Also, a decreased compressive strength is usually associated with the presence of alkali sulfates in cement [6]. The hydration of tetracalcium aluminoferrite yields products very similar to those produced during C3A hydration, although ettringite produced during C4AF hydration incorporates a fairly large number of Fe3+ ions. This ettringite is considered to be impervious to sulfate attack, probably due to the Fe3+ substitutions. Generally, the C4AF reaction rate is slower, although its reactivity depends on its exact chemical composition and Fe/Al ratio [1]. 1.3 Sulfate Attack One of the major durability issues associat ed with concrete exposed to sulfate environment is sulfate attack. Sulfate attack usually occurs when concrete structures are exposed to seawater, groundwater, and rainwater containing sulfate ions. In 1936, the U.S. Bureau of Reclamation formed a classification of potential for sulfate attack on concrete based levels of soluble sulfates in soils and groundwater. Concentrations of soluble sulfates below 0.1 % in soils and below 0.15 g/L in groundwater are considered too low to cause sulfate attack. Higher concentrations, between 0.1 and 0.2 % in soils and

PAGE 21

9 between 0.15 and 1.5 g/L in groundwater, are cla ssified as having moderate potential for sulfate attack. Levels above 0.2 % of soluble sulfates in soils and above 1.5 g/L in groundwater constitute high potential for sulfate attack [14]. Although the literature on the causes of concrete expansion due to sodium sulfate attack is inconclusive, formation of gypsum and ettringite are generally held responsible for expansion and subsequent deterioration of concrete during sulfate attack [12, 15, 16]. Secondary ettringite forms at the early stages of sulfate attack and is responsible for initial expansion. Gypsum begins to form at later stages of attack after both aluminate and sulfo-aluminate hydrates have been converted to ettringite [17]. It has been suggested that formation of gypsum is an expansive reaction as well [17, 18, 19, 20]. Although ettringite is considered to be a telltale sign of sulfate attack, its presence in concrete and the amount of ettringite formed are not always indicative of expansion. Odler and Colan-Subauste [16] monitored expansion of paste prepared by mixing different forms of calcium aluminates (C3A, AFm, C4AF, etc.) with gypsum and C3S. It was observed that there seemed to be a relationship between the dissolution rate of the aluminate compounds and ettringite formation rate. The amount of ettringite alone was not necessarily indicative of high expansion. Bars prepared with aluminum sulfate had the highest amount of ettringite (approximately 30 % by weight of total sample) and second lowest expansion. However, bars prep ared with tricalcium aluminate experienced the highest expansion (an order of magnitude higher than aluminum sulfate bars) while having the lowest ettringite quantities (13 %). Samples prepared with tetracalcium aluminate ferrite experienced significantly lower expansion, even though they contained the same amount of ettringite as the C3A bars. These findings emphasize the effect of

PAGE 22

10 solubility of aluminate compounds in cement on expansion due to ettringite formation. The researchers suggest that the differences in the expansion of different samples are connected to the mode of ettringite formation. They believe that the higher the fraction of topochemically formed ettringite and the more ettringite crystals are growing in the oriented fasion, the higher the expansion. Unfortunately, the microstructure of the bars was not observed on SEM. Odler and Colan-Subauste also speculate that swelling of ettringite might have contributed to expansion in some way because bars stored in saturated calcium hydroxide solution expanded more than bars stored in air. 1.3.1 Expansion due to Ettringite Formation The findings presented by Odler [16] raise several questions. It is yet not understood why ettringite is expansive in some cases, but not in others. Is expansive ettringite morphologically different from non-expansive ettringite? What is the mechanism of such expansion? Presently, there are several theories that attempt to explain the mechanisms of sulfate expansion. The most popular theories dealing with ettringite formation and/or presence in concrete are solid volume increase, swelling, expansion due to crystallization pressure, and oriented crystal growth. Although a number of studies have been done trying to confirm or dispute these theories, no decisive evidence exists as to which is the precise mechanism of concrete deterioration. 1.3.1.1 Increase of Solid Volume Skalny et al. [8] point out that ettringite formation reaction results in a decrease of the total volume. Hime and Mather [21] agree that formation of ettringite from gypsum, tricalcium aluminate and water is accompanied by a decrease in volume, given that all the

PAGE 23

11 reactants originated within the system However, during secondary ettringite formation, after the concrete has been placed in sulfate solution, not all the reactants originate within the system and the reaction becomes expansive. In an open system, only solid volume changes should be considered. 1.3.1.2 Swelling Hypothesis This theory proposes that ettringite can adsorb water into its crystalline matrix thus causing a volume increase. Mehta [12, 14, 22] believes that the primary mechanism of sulfate expansion is not the formation ettringite in hardened concrete due to ingress of sulfate ions, but the swelling of existing ettringite. He speculates that only “colloidal” ettringite is capable of attracting large amount s of water due to its high specific surface area and a net negative charge [12]. Mehta also states that before this expansion can occur, concrete stiffness needs to be low otherwise pressures caused by swelling of ettringite would not be enough to result in the expansion of the system [14]. He proposes that concrete stiffness could be lowered by formation of gypsum and adsorption of sulfate ions by CSH [22]. Although Mehta [12, 13] observed formation of “colloidal” ettringite in the presence of CH by scanning electron micr oscopy, he did not perform any length measurements on samples in these studies. In a different experiment, Mehta [22] monitored expansion of paste stored in a 4 % Na2SO4 solution maintained at a pH of 7. He also observed mineralogical changes in the paste via measuring powder x-ray diffraction peak intensities of ettringite, CH, and gypsum. Cement used in this study contained 58 % of C3S, 22 % C2S, 12 % C3A, and 4 % gypsum, according to the Bogue calculations. Before the samples were immersed in solution, no ettringite was detected in

PAGE 24

12 the paste, only monosulfoaluminate. During the first 7 days of immersion, small expansions (0.416 %) were recorded although most of the ettringite was formed during that time. A large drop in peak intensity of calcium hydroxide and a corresponding increase in the peak intensity of gypsum were observed between 3 and 7 days. After 7 days, expansion increased dramatically, reachi ng 2.07 % at 10 days. Mehta attributes this increased rate of expansion to the loss of stiffness of the paste caused by gypsum formation. He speculates that ettringite formed in the paste was “colloidal”, although no evidence (SEM) was presented to confirm this assumption, and once the elastic modulus of the paste was lowered, ettringite began to adsorb water and generated swelling pressures that led to expansion of the paste. Unfortunately, since the paste was not examined in the SEM, it is not obvious that ettringite formed in the paste was “colloidal”. Also, peak intensity measurements might not reflect the changes in the amounts of ettringite, especially if it is, as Mehta suggests, “colloidal” or poorly crystalline. An increase in the amount of a poorly crystalline substance might not necessarily result in large increase of peak intensity, but rather in peak broadening. Peak area measurements would have been more reliable in this case. 1.3.1.3 Crystallization Pressure Crystallization pressure hypothesis states that expansion occurs when the pore solution becomes supersaturated with respect to ettringite, ettringite filling the pores starts to exert pressure on the pore walls generating expansion [8]. Ping and Beaudoin [23] suggest that crystallization pressure starts to develop when ettringite fills the pores and all other available spaces in concrete. When there is no more room for it to precipitate and there is a constant ingress of sulfate ions, the higher the concentration of

PAGE 25

13 the reactants in the pore solution, the higher the pressure that ettringite exerts on the concrete. Ping and Beaudoin [24] go on further to try to connect the crystallization pressure exerted by ettringite on a single pore to the overall expansion of the sample. They suggest that expansion is directly related to the concentration of the reactants, temperature, Young’s modulus and porosity of concrete, and number and shape of the solid products. In order to confirm their hypothesis, Ping and Beaudoin prepared mortar bars from an ASTM Type I portland cement with a w/c of 0.5 and stored them in a 5 % Na2SO4 solution. After the bars reached an expansion of 0.4-0.5 %, they were immersed in pure water and Ca(OH)2, Na(OH)2, and Ba(OH)2 solutions. All the bars continued to expand, except for the bars placed in the NaOH solution, which the scientists took as a confirmation of their theory. Water and Ca(OH)2 solution were supposed to increase the concentration of the ettringite reactants (H2O, Ca2+) in the pore solution, while NaOH solution was supposed to decrease the crystallization pressure of ettringite by increasing the OHion concentration. Expansion of bars in the Ba(OH)2 solution was attributed to the formation of barite. 1.3.1.4 Oriented Crystal Growth Oriented crystal growth theory is closely related to the crystallization pressure hypothesis. This type of expansion is usually associated with topochemical formation of ettringite. Expansion occurs when ettringite crystallizes perpendicular to the surface of C3A particles. Such crystal growth is thought to be able to generate crystallization pressures sufficient to disrupt the concrete matrix [8, 25]. Ogawa and Roy [10, 11] observed formation of ettringite crystals growing radially around C4A3 S particles. The researchers prepared C4A3 S from a mixture of

PAGE 26

14 calcium carbonate, gypsum, and Al2O3. C4A3 S was then mixed with CaSO4, Ca(OH)2, and water and placed in high humidity curing chambers at various temperatures. Expansion of this paste was monitored by measuring its linear length change [10]. The microstructure of the paste was also studied via the SEM at 0.5, 1, 3, and 7 days. Ogawa and Roy [11] observed that ettringite initially formed around C4A3 S grains had no particular orientation. However, as the hydration progressed ettringite crystals became oriented perpendicular to the surface of C4A3 S grains. The onset of expansion corresponded to ettringite formations around individual grains becoming interconnected. The researchers suggested that this expansion could be due to the fact that these large ettringite formations started to exert pressure on the paste. Although these studies vary greatly in their explanation of the mechanism for ettringite-induced expansion, all four recognize that high calcium hydroxide concentrations change ettringite behavior and adversely affect concrete durability. The relationship between CH concentration and expansion in the solid volume increase theory might not be very obvious at first because it is an indirect relationship. As mentioned previously, high CH concentration decreases C3A solubility, thus providing a larger source of aluminate ions for ettringite formation after the concrete has hardened. 1.3.2 Expansion due to Gypsum Formation When concrete comes in contact with a sulfate environment, sulfate ion that enter the concrete matrix may react with calcium hydroxide to form gypsum. Ca(OH)2 + SO4 2(aq) CaSO42H2O + 2OH(aq) Mindess [2] believes that gypsum formation becomes significant only when sulfate ion concentrations reach above 1000 ppm. When sulfate ion concentration is

PAGE 27

15 between 1000 ppm and 4000 ppm, gypsum formation is secondary to the ettringite formation in terms of the driving force of concrete expansion. However, above 4000 ppm, gypsum formation becomes the primary expansive reaction. The role of gypsum in concrete deterioration during sulfate attack has been unclear for a long time. Is it just a byproduct of an attack driven by ettringite formation or is gypsum formation expansive as well? Tian and Cohen [18] studied expansion of alite and alite plus silica fume pastes exposed to 5 % sodium sulfate solution. Alite was chosen to eliminate ettringite formation in the samples. They observed no expansion in any of the samples for approximately 360 days. After that period, alite paste specimens exhibited high rate of expansion, reaching 0.1% expansion by 480 days. At the same time, samples containing silica fume showed no expansion. After 480 days of curing, x-ray diffraction was performed to determine the solid phases present in the samples. The analysis indicated gypsum as the main product formed. The researchers concluded that gypsum formation causes expansion in alite pastes and that expansion can be significantly reduced by silica fume additions. Mehta [26], on the other hand, observed only minimal expansion in alite pastes, although expansion was monitored only for 75 days. However, after 6 years of storage in sulfate solution, the samples exhibited severe spalling and decrease in strength. When examined with x-ray diffraction after six years, high quantities of CH and C-S-H were observed in the interior of the samples, although these phases were completely absent from in surface of the samples and replaced by gypsum and aragonite. Odler [25] agrees with Mehta that gypsum formation leads to surface spalling and not necessarily expansion of concrete. Odler points out that since gypsum is usually observed in the outer concrete

PAGE 28

16 surfaces, it might therefore serve as a barrier between the sulfate environment and the core of the concrete. 1.3.3 Expansion due to Monosulfoaluminate Conversion If monosulfoaluminate was formed during initial curing, it becomes unstable once the concrete comes in contact with a source of sulfate ions. Sulfate ions diffuse through the cracks and pores in concrete and react with monosulfoaluminate to form ettringite [2]. 4CaOAl2O312H2O + 2CaSO42H2O + 16H2O 3CaOAl2O33CaSO432H2O This reaction results in a solid volume increase that creates internal stresses in the concrete matrix. These internal stresses lead to cracking and spalling of concrete, accelerating the rate of sulfate attack by providing an easier passa ge for the sulfate ions into the interior of the concrete [2]. 1.4 Review of Previous Research The question of the influence of C3S content of cement on concrete durability has occupied researchers for many years. As early as 1979, Mehta et al. [26] studied the behavior of alite cements in sulfate environments. At the end of the study, the researchers concluded that alite cements exhibited poor durability in sulfate environment as they experienced severe spalling and decrease in strength. The samples used for evaluation of compressive strength and expansion were stored in a 5 % sodium sulfate solution, while some of the concrete prisms were placed in a sulfate solution (5 % Na2SO4 + 5 % MgSO4) for x-ray diffraction analysis. The researchers evaluated compressive strength and expansion of concrete and mortar prepared with pure alite, and

PAGE 29

17 alite with 3 % and 6 % gypsum additions, and concrete prepared with Types I/II, III, and V cements. Alite cements exhibited higher early strengths than Portland cements, but at later ages (28 and 90 days), mortars prepared with Type I/II cement possessed higher strengths. Specimens prepared with alite a nd 6 % gypsum experienced the largest drop in strength. Unfortunately, all the mixes had a variable water/cement ration, which would result in their variable porosity, thus influencing compressive strength results. The researchers stated that the expansion of alite cements was “exceptionally low”, however, expansion was monitored only for 75 days. After six years of sulfate exposure, x-ray diffraction analysis of the samples prepared with alite cements stored in a sodium sulfate-magnesium sulfate solution revealed the presence of high quantities of CH and C-S-H in the interior of the samples, although it seems that no quantification or semi-quantification of these phases have been performed. CH and C-S-H phases were completely absent from in surface of the samples and replaced by gypsum and aragonite. These XRD results should not be related to the expansion results and compressive strength data because samples were stored in different solutions. Reaction products formed during ma gnesium sulfate attack are different from the attack products in sulfate solution. More recently, Tian and Cohen [19] also studied pastes prepared with two types of C3S: one type (alite) consisted of monoclinic crystals with some impurities, and the other type was pure triclinic C3S. Paste specimens were prepared with alite, while C3S was used to prepare mortar samples. Both paste and mortar samples had a water/cement ratio of 0.48. Alite paste samples were stored in 5 % sodium sulfate solution and in limewater. The samples store in sodium sulfate solution had an induction period of 360

PAGE 30

18 days, after which they started to expand reaching 0.1 % at 480 days. The authors referred to this as “significant expansion.” It s hould be noted that from the plot accompanying these results it seems that all the samples experienced shrinkage and that their length change at 360 days was approximately –0.01 %. This is somewhat puzzling, considering the samples were supposed to be stored in solution. C3S mortar samples had a dormant period of about 40 days. After 40 days, the expansion rate of the C3S samples increased dramatically and an expansion of 1.05 % was reached at 230 days. It does not seem that either the C3S bars stored in limewater or the C3S bars stored in sodium sulfate solution experienced any negative change in length. Tian and Cohen [19] mention that they had only a limited amount of alite, which might explain why they chose to prepare paste samples instead of mortar. But it is still unclear why mortar samples were prepared with C3S and not paste. For comparative purposes it would have made more sense to make paste specimens in both cases. Mortar has higher porosity that paste, which would account for the shorter induction period and higher expansion with the C3S samples. X-ray diffraction analysis showed that gypsum was the main phase formed in all the specimens. The researchers concluded that although the mechanism of gypsum formation in concretes exposed to sulfate attack is unclear, the formation of gypsum seems to be expansive and causes distress. It is not surprising that both Mehta [26] and Tian et al. [19] found gypsum to be the main product formed, since their alite cements had no source of Al3+ ions necessary for ettringite formation. Irassar et al. [7] evaluated the performance of two Type V cements with different C3S content (40 % and 74 %) in a pH-controlled sulfate environment (0.352 M Na2SO4).

PAGE 31

19 High C3S cement also had a high C3S/ C2S ratio and a C3A content of 1 %. The C3A content of low C3S cement was zero. The researchers measured expansion, sulfate consumption, flexural and compressive strengths of the mortars prepared with these cements as well as their blends with 10 % and 20 % limestone filler (85 % CaCO3) replacement and 20 % and 40 % natural pozzolana replacement. It was concluded that high C3S cement had poor resistance to sulfate attack. Cement with high C3S content experienced a 0.350 % expansion at 360 days, which was an order of magnitude higher than that of the low C3S cement (0.035 %). Addition of limestone filler increased expansion of both cements, whereas addition of pozzolanic material reduced expansion. Compressive strength of the low C3S cement mortars was higher than that of the high C3S cement mortars. Addition of pozzolana improved compressive strength, while addition of limestone filler lead to strength loss. X-ray diffraction analysis revealed the presence of relatively small gypsum peaks in the low C3S cement mortar, but no ettringite peaks were observed. Gypsum and secondary ettringite were found in high C3S cement mortars. Irassar et al. suggest that secondary ettringite in the high C3S cement mortar formed due to interaction between sulfate ions and ferroaluminate hydrates. The researchers attribute the differences in durability of the two cements to their different C3S / C2S ratios. The increase in this ratio leads to the increased production of CH, which reacts with sulfate ions to produce gypsum. The authors speculate that gypsum forms crystals in the transition zone at the onset of sulfate attack. They suggest that this gypsum formation creates a favorable environment for the formation of

PAGE 32

20 expansive ettringite from calcium aluminoferrite at later ages. Since the high C3S cement had no C3A, and ettringite that was observed by XRD, it is logical to conclude that it was formed from the ferroaluminates, although this type of ettringite is though to be not expansive. Based on the behavior of the low C3S cement, researchers concluded that gypsum causes expansion only after a certain percentage of it has formed. High gypsum content in mortar also provides a favorable environment for formation of expansive secondary ettringite. Although the researchers do not reflect on the mechanisms of these reactions, they state that high amounts of calcium hydroxide are responsible for the poor performance of the high C3S cement. In an earlier paper, Gonzalez and Irassar [20] observed in the SEM that in mortars prepared with high C3S cements (60 % and 74 %) and no C3A gypsum precipitated in massive deposits in the air voids and in the transition zone around the aggregates at 360 days. They attributed this phenomenon to higher quantities of CH in the transition zone and its higher porosity. In the low C3S mortars (40 % and 50 %) gypsum was dispersed throughout the sample. At 720 days, ettringite was observed throughout the sample in high C3S mortars, while in low C3S mortars ettringite was located only in pores. The EDX analysis of the ettringite showed the presence of 2 to 10 % ferrite in its structure. The researchers believe that the deposition of gypsum in the transition zone was the cause of high expansion of these two mortars. Gonzalez and Irassar suggest that once gypsum filled all the available voids in the C-S-H gel around the aggregates, the growing gypsum crystals started to apply crystallization pressure on the C-S-H gel, producing cracks in the structure. The authors suggest that after a certain amount of gypsum forms

PAGE 33

21 and it becomes a dominant phase, C-S-H gel loses cohesion, and expansive ettringite starts to form. In conclusion, mortars with low C3S content (40 % and 50 %) did not experience high expansion. These samples had less gypsum a nd gypsum crystals were of smaller size and dispersed throughout the paste. Cao et al. [27] investigated resistance to sulfate attack of Portland cements with similar fineness and different chemical compositions. They discovered that cement with low C3S (39%) and low C3A content (4.6%) had the best performance in sulfate environments. It exhibited the lowest comp ressive strength drop and lowest expansion. The strength reduction for the all other cements was very similar. Cement with the highest C3A content (6.6%) had the highest expansion, although its C3S content was moderate (51 %). It was observed under the SEM, that in the cements with higher C3S content gypsum tended to form in large veins that were parallel to the surfaces exposed to solution. However, in the low C3S cement gypsum was located throughout the sample. Ettringite was found in all cements. The authors concluded that lowering the C3S and C3A content of cement improves its resistance to sulfate attack. Monteiro and Kurtis [28] confirm the findi ngs presented by Cao et al. [27]. After analyzing the results of a long-term study initiated by the U. S. Bureau of Reclamation, in which concrete cylinders were exposed to 2.1 % Na2SO4 solution for 40 years, Monteiro and Kurtis [28] concluded that low w/c ratio and low C3A content are of primary importance in delaying concrete deterioration due to sulfate attack. They also observed that concretes prepared with cements containing high amount of C3S failed much earlier than concretes with similar C3A amounts, but lower C3S content. Unfortunately, length

PAGE 34

22 change was the only parameter evaluated in this study and cement composition was determined based on the Bogue calculation alone. Ferraris et al. [29] evaluated the effect of C3A content of cement on expansion of mortars placed in sodium sulfate solutions of variable pH. Three original cements with 2.7, 11.1, and 12.8 % C3A were mixed to produce blended cements with C3A contents of 4, 6, and 8 %. Both original and blended cements were used to produce mortar mixes conforming to the ASTM C 109 that were stored in different sodium sulfate solutions. The sulfate concentration in the solutions varied from 0 to 10 % by weight. The pH of the solutions was controlled at 7, 9, and 11. Some bars were placed in the solution pH of which was not controlled, but the solution was changed every time measurements were taken according to ASTM C 1012. It was noticed that expansion of the bars in solution with controlled pH proceeded at a high rate compared to the bars stored in the solution with uncontrolled pH. This was as expected since high controlled pH solution had a higher concentration of sulfate ions. This led Ferraris to conclude that sulfate concentration has an effect on expansion rate of mortars. X-ray imaging was performed on mortars at different ages in order to identify the phases formed during sulfate attack. The samples taken from the outer 3 mm of the cross-section were considered as “outer’ samples, from the next 3 mm as “inner”, and from the center of the cross-section as “core” samples. There was a high concentration of sulfur observed up to approximately 2 mm from the surface compared to the control samples stored in saturated calcium hydroxide solution. This corresponds to High concentration of sodium levels in the calcium silicate hydrate gel were observed in the outer 200 microns by using the quantitative energy dispersive X-ray spectrometry.

PAGE 35

23 However, the presence of sodium in the C-S-H was not observed beyond 200 microns. It was noticed that the outer C-S-H was depleted in calcium compared to the control samples. However, calcium depletion of the mortars stored in sulfate solutions decreased with depth. Ettringite, gypsum, and monosulfate were observed in the outer layer of the samples after three days. The inner part and core of the samples contained calcium hydroxide, monosulfate, and traces of ettringite. Analysis performed at 28 days indicated the presence of gypsum and ettringite throughout the samples. The amount of monosulfate, however, appeared to decrease. The trend for all ages seems to indicate that the amount of ettringite and gypsum present in the samples increases with age and the amount of calcium hydroxide decreases. This is relevant for all depths of the crosssection, although the changes are most noticeable in the outer portion of the samples [16].

PAGE 36

24 CHAPTER 2. EXPERIMENTAL METHODS 2.1 Chemicals All the chemicals used in this study were ASC reagent grade chemicals obtained from Fisher Scientific, unless otherwise stated. 2.2 C3S (Alite), C3A, C4AF, and MgO Monoclinic alite, cubic and orthorhombic tricalcium aluminate, tetracalcium aluminoferrite and magnesium oxide we re furnished by Construction Technology Laboratories. Alite was prepared with Al3+ and Mg2+ ion substitutions to stabilize the monoclinic crystal structure at room temperature. Blaine fineness of alite used for mortar preparation was 3800 cm2/g, same as the cements in this study. Cubic and orthorhombic C3A, C4AF, and MgO had particles finer than 45 microns. 2.3 Cement 2.3.1 Blaine Fineness and Particle Size Distribution Cement fineness was determined according to the ASTM C 204-00 “Standard Test Method for Fineness of Hydraulic Cement by Air-Permeability Apparatus.” The Blaine air-permeability apparatus, obtaine d from Humboldt Manufacturing Co., was

PAGE 37

25 calibrated prior to use with the three separately prepared beds of standard cement SRM 114, purchased from the National Institute of Standards and Technology. Particle size distribution of cements was analyzed with a Malvern Mastersizer 2000 laser particle size analyzer. The results are displayed in Figures 1-5. 2.3.2 Oxide Chemical Composition An external laboratory performed the analysis of the oxide chemical composition of cements. Cements were fused at 1000C with Li2B4O7 and analyzed by x-ray fluorescence spectrometry with accordance to the precision and accuracy requirements of the ASTM C 114-99 “Standard Test Method for Chemical Analysis of Hydraulic Cement.” The results are listed in Table 4. 2.3.3 Mineralogical Composition 2.3.3.1 Bogue Calculation ASTM C 150-00 “Standard Specification for Portland Cement” provides the guidelines for using the Bogue formulas to determine the theoretical mineralogical cement composition. The results of the Bogue calculations are presented in Table 5. 2.3.3.2 Internal Standard Method Calibration curves for cubic and orthorhombic C3A, C4AF, and MgO were prepared according the ASTM Standard C 1365-98 “Standard Test Method for Determination of the Proportion of Phases in Portland Cement and Portland-Cement Clinker Using X-ray Powder Diffraction Analys is.” The x-ray scans were collected on a Phillips X’Pert PW3040 Pro diffractometer with Cu K radiation. The samples were scanned from 2 of 5 to 60 degrees with a step size of 0.02 degrees per step and counting time of 4 seconds per step. The tension and current was set at 45 kV and 40 mA.

PAGE 38

26 Divergence slit was fixed at 1 receiving slit had a height of 0.2 mm, and anti-scatter slit was fixed at 1 Collected patterns were analyzed with Profit software to determine selected peak areas. Measuring the peak area, rather than peak height provides more reliable data [30]. First, background was manually determined for each scan. Then, the program automatically searched for peaks. At this point, the user determined if all the necessary peaks have been identified. If the peak search routine had overlooked some peaks, they were specified manually. The 24.3 peak of C4AF, 27.4 and 36.1 peaks of TiO2 and 42.9 peak of MgO had to be corrected for the K -2 contribution by inserting a 0.5 in the 1/ 2 column for the appropriate peak. After this, the profile fitting command was executed, and the peak areas were calculated. The 36.1 peak of TiO2 was used in construction of all the calibration curves. The calibration curve for alite was prepared by grinding 10 g of C3S in a Norton ceramic jar mill operated at 150 RPM. The jar had a volume of approximately 300 ml and was filled with 68 cylindrical beads. Alite was ground for 90 min ethylene glycol. After grinding, 0.768 g of alite was mixed with 0.0853 g of TiO2, so that the total sample weight was 0.8533 g. TiO2 was obtained from Aldrich Chemical Company. The TiO2 powder had particles smaller than 5 microns and had a purity of at least 99.9%. This sample containing 90 % of alite and 10 % of TiO2 was called the “100% sample.” Another sample, called the “50 % sample”, was prepared by mixing 0.384 g of alite with 0.384 g of CaF2 and 0.0853 g of TiO2. Three 100 % and three 50 % samples were weighed out. The samples were placed in a vial and mixed with a spatula for 5 minutes to achieve homogeneity. Two drops of cyclohexane were added to each sample before mixing to improve the blending of the material. After mixing, samples were loaded into

PAGE 39

27 an aluminum sample holder and placed in the diffractometer. The area of 30.1 alite peak was divided by the area of the TiO2 peak at 36.1 and the obtained area ratio was plotted vs. the C3S content of the samples, excluding the amount of TiO2 added. The calibration curves for C3S, C3A, C4AF, and MgO are shown in Figures 9, A-1 A-2, and A-3 respectively. 2.3.3.3 X-Ray Powder Diffraction Analysis X-ray powder diffractometry was used to determine the mineralogical composition of cements in this study. The following procedure was employed to prepare samples for XRD analysis. Approximately 10 g of as-received cement was ground in a Norton ceramic jar as described above. Cement was ground for 30 mi n ethylene glycol. After grinding, most of the ground cement was placed in a ceramic crucible and ignited for 30 minutes at 500 C in a Fisher Scientific Isotemp oven in order to convert any gypsum or bassanite that might be present in cement to anhydrite. This was done to simplify x-ray diffraction analysis, as cements already suffer from significant peak overlap [30]. After cooling down to ambient temperature, 0.8533 g of cement with TiO2 was weighed out, the weight of TiO2 being 10 % of the total sample weight. The sample was placed in a vial and mixed with a spatula for 5 minutes to achieve homogeneity. TiO2 was added to this sample because this sample was used to verify the results of Rietveld refinement through the calibration curves. Another 4.88 g was taken from the ignited sample and mixed with 0.12 g of TiO2, which constituted 2.4 % of the total sample weight. This 5 g sample was used for the salicylic acid/methanol extraction (SAM) [30, 31]. It was estimated using the Bogue

PAGE 40

28 calculation that a 2.4 % addition of TiO2 before the extraction would result in approximately 10 % of TiO2 after the extraction. Larger amounts of TiO2 would be undesirable, since they might dilute the sample and make it impossible to see the minerals present in small quantities (less than 1 %). Salicylic acid/methanol extraction was used to determine the crystal structure of C3A present in cement. SAM extraction dissolves calcium silicates and free lime, leaving a residue of aluminates, ferrites, and minor phases, such as periclase, carbonates, alkali sulfates, and double alkali sulfates [30, 31]. To prepare this extraction, 20 g of salicylic acid was added to 300 ml of methanol that was mixing on a stirring plate. Then, 5 g of cement and TiO2 were added and mixed for 2 hours in a stoppered flask. The suspension was then vacuum filtered through a Polypro membrane 0.45-micron disc filter. The residue collected on the filter was washed with 100 ml of methanol, and then dried at 100 C for 30 minutes. It was possible to relate the amounts determined after the extraction back to the original cement composition by adding TiO2 before the extraction, although the results could not be considered reliable since it was impossible to determine how much of the extracted sample was lost in the process. The portion of the ground cement that was not ignited was analyzed via x-ray powder diffractometry. X’Pert Plus, a Riet veld-refinement software, was used for quantitative analysis of these results. It should be noted that only seven crystal structures can be refined at the same time with this particular software. The patterns were analyzed in the semi-automatic mode. The following parameters were refined simultaneously: background parameters 1, 2, 3, 4, and 6, zero shift, and scale factors for each phase.

PAGE 41

29 Then, lattice parameters were refined one phase at a time for each phase. Calculated weight fractions were checked to see which phases constituted less then 5 weight percent of the sample. For these phases, no further refinement was performed because it was observed that attempting to refine half width parameters for these phases generated an error in the program. For the phases that constituted more than 5 weight percent of the sample the half width parameter W was refined. For samples containing alite, V and U parameters as well as preferred orientation were refined for alite. 2.4 Sand All the sand used in this study was obtained from the U.S. Silica Company and conforms to the ASTM C 778-00. 2.5 Mortar Durability Tests Two aspects of durability were investigated in this study: compressive strength in lime and sodium sulfate solutions and length change in sodium sulfate solution. Mortar cubes were prepared to assess compressive strength and mortar bars were used to determine length change. 2.5.1 Mortar Cubes Mortar cubes were mixed in accordance to the ASTM C 305-99 “Standard Practice for Mechanical Mixing of Hydraulic Cement Pastes and Mortars of Plastic Consistency” and molded and tested according to the ASTM C 109-99 “Standard Test Method for Compressive Strength of Hydraulic Cement Mortars.”

PAGE 42

30 Thirty-nine cubes were prepared for each of the as-received cements. The mix proportions are listed in Table 1. Three cubes were tested at 1 day, then 18 cubes were placed in saturated calcium hydroxide solution and the other 18 cubes were placed in 5 weight percent sodium sulfate solution. Sodium sulfate solution was prepared according to the ASTM C 1012. Compressive strength of cubes stored in both solutions was tested at 3, 7, 28, 90, 120, and 180 days according to the ASTM C 109 using the MTS 809 Axial/Torsional Test System. Sodium sulfate solution was replaced every time the cubes were tested for strength. Before the cubes/bars were placed in the new solution, its pH was measured with a pH meter to ensure that the pH was between 6.0 and 8.0. After the bars were placed in solution, the pH changes were not monitored. Table 1. Mix Proportions for As-Received Cements 9 cubes Cement, g 740 Sand, g 2035 Water, ml 359 In addition to the cubes prepared with as-received cements, 9 cubes were prepared for each cement with additions of C3S and C3A. The mix proportions are listed in Table 2. Alite was added to cement to bring its total C3S content up to 69 % based on the Bogue mineralogical content. C3A was added to keep the amount C3A same as in the asreceived cement, again based on the Bogue calculations. Alite and tricalcium aluminate were added to dry cement and mixed in beaker with a spatula several minutes until homogeneous appearance.

PAGE 43

31 Table 2. Mix Proportions for C3S-Doped Cements C D2 E P Cement, g 531.52 636.04 486.49 392.96 C3S, g 194.35 98.19 237.51 321.11 C3A, g 14.14 5.77 16.00 25.92 Sand, g 2035 2035 2035 2035 Water, ml 359 359 359 359 Cubes were mixed in the same manner as described above. The alite-doped cubes were stored in the 5 % sodium sulfate solution and tested for compressive strength at 28, 90, and 120 days. Their solution was changed at the same ages, as the as-received cement cubes, i.e. 3, 7, 28, 90 days. 2.5.2 Mortar Bar Preparation Mortar bars were prepared according to the ASTM C 1012-95a “Standard Test Method for Length Change of Hydraulic-Cement Mortars Exposed to a Sulfate Solution” and ASTM C 490-00a “Standard Practice for Use of Apparatus for Determination of Length Change of Hardened Cement Paste, Mortar, and Concrete.” After the initial curing in saturated calcium hydroxide solution, mortar bars were exposed to the sodium sulfate solution. Their length change was measured at 7, 14, 21, 28, 56, 91, 105, 120, 150, 180 days, and then every 15 days after that. Sodium sulfate solution was replaced every time the bars were measured. 2.5.3 X-Ray Powder Diffraction Analysis of Mortar Bars At the age of 360 days, mortar bars were broken with a hammer and a screwdriver. Two inches from the middle were taken out and broken in half. One half

PAGE 44

32 was soaked in acetone for an hour, ground in mortar as described above, mixed with TiO2 and placed in a diffractometer. The outside surfaces of the other half were sawed off with a hacksaw at a thickness of 1 mm, ground, mixed with TiO2 and analyzed with XRD as described in section 2.3.3.2. 2.5.4 Scanning Electron Microscopy When the mortar bars were broken after 360 days. A cross-section of the bar was cut with a Buehler Isomet slow speed saw. The thickness of the cross-section was approximately 5 mm, and one of the faces of the cross-section was a fractured surface exposing the morphology of the hydration products. The cross-section was submerged in acetone overnight to stop the hydration and to dry the sample by displacing the water with acetone. After the over-night soaking in acetone, the cross-sections were placed in a vacuum desiccator under a vacuum of 25 in of Hg. The pieces from the top of the bars were mounted onto the sample holder with double-stick tape and the cross-sections were attached to the sample holder with copper tape. Then the samples were placed in a Hummer 6.2 sputter coater and coated with 40 nm of AuPd. They were imaged in a Hitachi S-350N variable pressure scanning electron microscope. Energy dispersive spectroscopy (EDS) was performed in a spot mode with a Princeton Gamma Tech prism light element detector. The working distance was 15 mm with zero degree tilt. The software used to analyze EDS data was Princeton Gamma Tech IMIX.

PAGE 45

33 CHAPTER 3. EXPERIMENTAL RESULTS AND DISCUSSION 3.1 Physical Characteristics of As-Received Cements 3.1.1 Blaine Fineness Fineness is an important parameter in determining the reactivity of cement. An increase in fineness corresponds to an increased surface area of cement that would be in contact with water during cement hydration, thus increasing the rate of hydration. Previous work [25, 32] has shown that higher cement fineness results in higher expansion of concrete exposed to sulfate environment. Based on these considerations, it was decided to eliminate fineness as a variable in this study. Therefore, all the cements in this study were specifically chosen because of their variable C3S content and similar fineness. Cement D was added for the sole purpose of illustrating the effect of fineness on the susceptibility of a high C3S cement to sulfate attack. Table 3 lists Blaine fineness values obtained for the as-received cements. D cement has the lowest fineness, while the rest of the cements have very similar fineness. Even though cement D2 has the highest fineness and cement E has the lowest fineness among the four cements, the difference between D2 and E is not significant.

PAGE 46

34 Table 3. Blaine Fineness Cement D C D2 E P Blaine Fineness (cm2/g) 3560 3840 3880 3800 3820 3.1.2 Particle Size Distribution Particle size distribution is closely relate d to fineness. Figure 1 shows a plot of the particle size distribution for all the cements presented as cumulative percent passing. 0 20 40 60 80 100 120 0.1 1 10 100 1000Particle size (microns)Cumulative % passing C D2 E P Figure 1. Particle Size Distribution for As-Received Cements Figures 2 through 5 display the particle size distribution as cumulative percent passing as well as individual percent retained for each cement. It was not deemed necessary to determine the particle size distribution for cement D. When considering the results for particle size distribution expressed as cumulative percent passing, the cements can be divided into two groups: E, P, and C, D2. Particle size distribution of cement E is very similar to that of P, and the fineness of C cement is very similar to the fineness of cement D2. C cement has the largest amount of particles

PAGE 47

35 less than 10 microns (55 %), while D2 cement has slightly smaller amounts (51 %). Both E and P cements have 40 % of particles that are smaller than 10 microns. 0 1 2 3 4 5 6 0.11101001000 Particle size (microns)Individual % retaine d 0 20 40 60 80 100 120 0.1 1 10 100 1000 Particle size (microns)Cumulative % passin g Figure 2. Particle Size Distribution for Cement C 0 1 2 3 4 5 6 0.11101001000 Particle size (microns)Individual % retaine d 0 20 40 60 80 100 120 0.1 1 10 100 1000 Particle size (microns)Cumulative % passin g Figure 3. Particle Size Distribution for Cement D2 Based on the individual percent-retained graphs, C has an average particle size of 10microns, D2 cement has an average particle size of 15 microns, and E and P cements have an average particles size of approximately 20 microns. In C, D2, and P cements, this particle size occurs between 4.34 and 4.4 %, while in E cement the mode particle size

PAGE 48

36 comprises 5.03 % of the cement. Overall, similar reactivity of C, D2, and E, P cements is based their similar fineness. 0 1 2 3 4 5 6 0.11101001000 Particle size (microns)Individual % retaine d 0 20 40 60 80 100 120 0.1 1 10 100 1000 Particle size (microns)Cumulative % passin g Figure 4. Particle Size Distribution for Cement E 0 1 2 3 4 5 6 0.11101001000 Particle size (microns)Individual % retaine d 0 20 40 60 80 100 120 0.1 1 10 100 1000 Particle size (microns)Cumulative % passin g Figure 5. Particle Size Distribution for Cement P 3.2 Oxide Chemical Content Oxide chemical composition of as-received cements was determined with x-ray fluorescence spectroscopy. The results are listed in Table 4. D2 cement has the highest

PAGE 49

37 amount of MgO followed by C cement, although their MgO is well below the 6 % limit established in the ASTM C 150-00. P and C cements have the higher SO3 content than D2 and E. P cement also has the highest amount of equivalent alkalies, which marginally exceeds the limit established by ASTM C 150-00. D2 cement has the highest level of free CaO. Table 4. Oxide Chemical Composition of As-received Cements Cement D C D2 E P Analyte SiO2 (%) 21.1 20.52 20.55 21.15 20.78 Al2O3 (%) 3.52 4.92 4.4 4.78 5.47 Fe2O3 (%) 3.47 3.7 3.61 3.76 4.15 CaO (%) 65.97 64.31 64.6 64.41 63.14 MgO (%) 1.37 1.71 2.47 0.95 0.85 SO3 (%) 2.57 2.81 2.54 2.58 2.88 Na2O (%) 0.03 0.01 0.03 0.18 0.26 K2O (%) 0.62 0.41 0.54 0.34 0.6 TiO2 (%) 0.16 0.27 0.22 0.33 0.32 P2O5 (%) 0.06 0.03 0.05 0.07 0.18 Mn2O3 (%) 0.05 0.04 0.05 0.03 0.03 SrO (%) 0.1 0.04 0.02 0.12 0.05 Cr2O3 (%) <0.01 <0.01 0.02 <0.01 0.02 ZnO (%) 0.05 <0.01 0.03 0.02 0.02 L.O.I. (950 C) (%) 0.74 1.08 0.99 1.15 1.3 Total (%) 99.81 99.83 100.12 99.84 100.04 Alkalies as Na2O (%) 0.43 0.27 0.39 0.4 0.65 Free CaO (%) 1.01 0.92 2.31 1.05 0.44 3.3 Mineralogical Composition 3.3.1 Bogue Mineralogical Content Based on the chemical oxide analysis and Bogue calculation, all the cements are ASTM Type I portland cements. The C3S content of cements varies from 48 % in

PAGE 50

38 cement P to 72 % in cement D. As the C3S content of the cement increases, so does its C3S/C2S ratio. P cement has the lowest C3S/C2S ratio of 2.1 and D cement has the highest ratio of 12. D cement has the lowest C3A content of 3 %. C and P cements have C3A content of 7 %. The C3A content of D2 and E cements is slightly lower than that of C and P (6 %). All the cements have a C4AF content of 11 %, except for cement P, which has 13 % of C4AF. Table 5. Bogue Mineralogical Content Cement Compound C D D2 E P C3S (%) 60 72 65 57 48 C2S (%) 14 6 10 18 23 C3A (%) 7 3 6 6 7 C4AF (%) 11 11 11 11 13 C3S/C2S 4.3 12 6.5 3.2 2.1 3.3.2 Internal Standard Method The C3S calibration curve was determined using a laboratory prepared sample of monoclinic alite. Monoclinic alite has three polymorphs: M1, M2, and M3 [4]. These polymorphs have some differences in their powder x-ray diffraction patterns. A visual identification method based on comparing the different peak shapes of the three polymorphs is described by Courtial [33]. This method was used to determine which monoclinic polymorphs were present in the sample. After observing the five angular windows recommended by Courtial, (Figure 6), it was concluded that the laboratory prepared alite consisted mostly of the M3 polymorph. However, the shape of the peak in the 51-53 2 angular range suggests that the sample has small amounts of the M1 polymorph.

PAGE 51

39 0 50 100 150 200 250 300 350 400 450 500 24.52525.52626.5 Position (deg 2 Theta)Intensity (counts ) 0 50 100 150 200 250 300 350 400 450 500 26.52727.52828.5 Position (deg 2 Theta)Intensity (counts ) 0 500 1000 1500 2000 2500 3000 31.53232.53333.5 Position (deg 2 Theta)Intensity (counts ) 0 50 100 150 200 250 300 350 400 450 500 3636.53737.538 Position (deg 2 Theta)Intensity (counts ) 0 200 400 600 800 1000 1200 5151.55252.553 Position (deg 2 Theta)Intensity (counts ) Figure 6. Five Characteristic Angles of the Laboratory-Prepared Alite Used for the Monoclinic Polymorph Identification

PAGE 52

40 It was not possible to determine the proportions of M1 and M3 polymorphs in the sample because inserting two phases in the program that have the same chemical formula and crystal system with only minor differences in the structure causes failure in the Rietveld refinement [4]. This is a drawb ack of the Rietveld refinement method, and not of the software used in this project. Figures 7 and 8 show that the predominant form of the C3S in the cements is monoclinic, most likely a combination of M3 and M1, although it is clear that each cement had a different combination of the alite polymorphs. Figure 7. Characteristic Angles Used for the Monoclinic Alite Polymorph Identification

PAGE 53

41 Figure 8. Characteristic Angles Used for the Monoclinic Alite Polymorph Identification Since the predominant form of alite in the cements used in this investigation was M3 in combination with M1, the obtained monoclinic alite was considered to be of acceptable purity for the construction of the C3S calibration curve, which is presented in Figure 9.

PAGE 54

42 y = 0.023x R2 = 0.9828 0 0.5 1 1.5 2 2.5 3 020406080100120Weight % of sampleArea ratio Figure 9. Established C3S Calibration Curve Tricalcium aluminate is commonly found in portland cement in cubic or orthorhombic form. Before the C3A amounts could be determined for the cements used in this study, it was necessary to determine the form of C3A in these cements. Tricalcium aluminate comprises only a small fraction of cement compared to the silicate compounds, which obscure the aluminate peaks. The highest peaks for C3S, C2S, and C3A are located between 2 of 30 and 35 [3]. In order to eliminate the contribution from the silicate phases, cements were treated with a salicylic acid/methanol solution. This procedure [30, 31], also referred to as a SAM extraction, dissolves tricalcium and dicalcium silicates. Based on the x-ray diffraction patterns of th e extracted residues containing tricalcium aluminate, calcium aluminoferrite, anhydrite, and minor phases, it was determined that cements used in this investigation did not contain any orthorhombic C3A. Figure 10 shows the XRD scans of the pure cubic and orthorhombic C3A, as well as scans of extraction residues of C, D2, E, and P cements. Although the XRD patterns of cubic and

PAGE 55

43 orthorhombic C3A are very similar, orthorhombic C3A does not have a peak at 2 of 21.8 It is clear from Figure 10 that the peak at 2 of 21.8 is present in all the cements, which confirms the presence of cubic C3A. Figure 10. Angular Range (2 = 18-24 ) Confirming the Presence of Cubic C3A in Cements As can be seen in Figure 11, cubic C3A has only one peak at 2 of 33.3 while orthorhombic C3A has two peaks in this angular range, one at 2 of 32.9 and another at 2 of 33.2 The 2 of 32.9 peak is absent from the cement patterns, which confirms the absence of orthorhombic C3A. Calibration curves for cubic C3A, C4AF, and MgO prepared per ASTM C 1365 are listed in Appendix A.

PAGE 56

44 Figure 11. Angular Range (2 = 29-36 ) Confirming the Absence of Orthorhombic C3A in Cements Table 6 presents the amounts of C3S, C3A, C4AF, and MgO in as-received cements based on the internal standard method. P cement has the lowest C3S content, followed by E cement, which is in accordance with the Bogue results. However, based on the calibration curves, C cement has the highest C3S content, followed by D2. Numerically, only D2 and E cements had C3S contents similar to those based on the Bogue calculation. As far as the C3A content, only P cement had a value similar to the Bogue. The rest of the cements had a much lower C3A content than that determined by

PAGE 57

45 the Bogue calculation. P cement has the highest C3A content, followed by E, C, and D2. Cements D2, E, and P had a similar C4AF content to that determined by Bogue. C4AF content of C cement was higher when calculated by the internal standard method compared to the Bogue method. P and C cements had a similar C4AF content, which was higher than that of D2 and E cements. Table 6. Phase Composition of As-Received Cements Based on the Internal Standard Method Cement Compound C D2 E P C3S (%) 70 63 58 55 Cubic C3A (ASTM C 1365) (%) 3 3 4 6 C4AF (ASTM C 1365) (%) 14 11 10 15 MgO (ASTM C 1365) (%) 1 2 0 0 The advantage of using the calibrations curves of the internal standard method is that it allows for one phase to be quantified without determining the amounts of the other phases present in the sample [34]. It is ideal to use a material for building the calibration curves that has the exact chemical composition and crystal structure, as it is present in the samples to be analyzed. However, this is often impossible when dealing with cement samples due to numerous possible substitutional ions and crystal system polymorphs. Preferred orientation as well as peak overlap is also a concern when using the internal standard method [34]. Care was taken to sel ect peaks free from preferred orientation and with minimal overlap from other phases.

PAGE 58

46 Usually, the values obtained by Bogue calculation do not agree with those obtained by XRD because the Bogue calculation is based on several assumptions that are not true in reality. Bogue formulas assume that the composition of all phases is constant, while all the phases can incorporate a large number of substitutional ions. The calculation also assumes chemical equilibrium, which is hardly ever the case in a kiln. [3]. However, even though the Bogue calculation is an approximation of the actual phase composition, it is still widely used in the cement industry and is required by the ASTM C 150-00; therefore, the Bogue mineralogical content was used in this study. 3.3.3 Rietveld Refinement To date, only two structures have been determined for monoclinic alite: M3 superlattice structure [35] and M3 sublattice structure [36]. A superlattice structure is defined as “ a structure in which the atoms are in positions that deviate only slightly from those of the simple parent structure, at least one set of equivalent sites of the parent structure being split into two or more sets of equivalent sites in the superlattice [37].” In other words, it has larger dimensions than a parent structure, which is also sometimes referred to as a sublattice of the superlattice structure [38]. After several initial refinements, Nishi superstructure was used for quantification because it provided a better fit for all the cements. This structure seemed to be more flexible, which could be attributed to a much higher number of atomic positions (226 in the Nishi structure compared to 21 in the Mumme structure). Figures 12 and 13 illustrate the differences between Mumme and Nishi structures.

PAGE 59

47 Figure 12. Mumme Sublattice Structure of the Monoclinic Alite Figure 13. Nishi Superlattice Structure of the Monoclinic Alite As it was mentioned previously, no orthorhombic C3A was present in the cements used in this study. Therefore, only cubic C3A structure was used in the Rietveld refinement.

PAGE 60

48 Figure 14 shows the final Rietveld refinement profile for cement C. Rietveld refinement profiles for the other cements are listed in Appendix B. The top plot shows the collected XRD pattern (red) and calculated pattern (blue). Vertical lines correspond to the peak positions of different phases, one color per phase. Below, is a difference plot between the collected and the calculated patterns. Figure 14. Rietveld Refinement for As-Received Cement C

PAGE 61

49 E cement has the highest amount of the total calcium sulfates, 3.6 % as determined by Rietveld analysis. Although C cement has the second highest total alkali content, 2.8 %, almost half of it is pres ent in the form of insoluble anhydrite (Ca2SO4). As it was mentioned in section 1.2, solubility of this form of anhydrite, as implied by its name, is extremely low and should be excluded from the analysis. Therefore, the actual amount of calcium sulfates participating in hydration in cement C is only 1.5 %, which is very similar to the total calcium sulfate content of D2 and P cements. The similarity between cement C and D2 is further promoted by the fact that in both of them calcium sulfate is present in the form of bassanite, which is more soluble than gypsum. P cement has the lowest C3S and the highest C3A content, while C cement has the highest C3S and the lowest C3A. C cement has the highest C3S/C2S ratio, followed by D2. E and P cements have the same C3S/C2S ratio. This is not in agreement with the Bogue results. But, as discussed previously, Bogue values are very approximate, and since C3S, C3A, C4AF, and MgO amounts agree with those obtained by the internal standard method, there is no reason to doubt the validity of the Rietveld refinement results. In the case of C3A, there are only minor variations between the internal standard method and Rietveld results. The variance in the C3S content is not significant either. This variance could be attributed to the fact that even though the C3S is present in the monoclinic form in all the cements, the combination of the monoclinic polymorphs is different in each cement. Figure 8 presents two of several angular windows that are commonly used to identify the structure of C3S [3, 33]. This figure clearly illustrates the differences between the alite structures in the cements.

PAGE 62

50 Table 7. Rietveld Refinement Result s for the Ground As-Received Cements Cement Compound C D2 E P C3S (%) 67 61 54 53 -C2S (%) 15 19 25 23 Cubic C3A (%) 2 3 4 9 C4AF (%) 14 12 13 11 Gypsum (Ca2SO42H2O) (%) --2.0 1.1 Bassanite (Ca2SO40.5H2O) (%) 1.5 1.6 1.6 0.7 Insoluble Anhydrite (Ca2SO4) (%) 1.3 ---Magnesite (MgCO3) (%) ---1.8 Periclase (MgO) (%) 0.6 1.8 --Dolomite (CaMg(CO3)2) (%) ---0.8 Portlandite (Ca(OH)2) (%) -1.2 --C3S/C2S 4.5 3.2 2.2 2.3 Total Ca2SO4 (%) 2.8 1.6 3.6 1.8 The advantage of the Rietveld refinement is that it uses the complete diffraction pattern rather than just one peak. Crystal structures are refined to fit each individual pattern, and there is no need for an internal standard that might obscure some of the peaks and complicate the pattern. However, Rietveld refinement assumes that all the phases add up to a 100 %, so a complete list of all the crystal structures should be inserted into the program.

PAGE 63

51 3.4 Optical Microscopy All the clinkers were stained with ammonium nitrate solution and then with potassium hydroxide solution [39]. The first st aining colored alite grains dark brown to blue and belite grains light brown. The second staining turned C3A a bluish gray color, making it more distinctive inside the matrix. The C4AF was unaffected by either solution and can be observed as the white portion of the matrix. Figure 15. Reflected Light Microscopy Images of Clinker C Typical appearance of the alite crystals in C clinker is illustrated in Figure 15a-c. Although C clinker has fairly large alite crystals, some 70 to 80 microns long (Figure 15a), some very small crystals were observed in this clinker as well (Figure 15d). Both

PAGE 64

52 euhedral prismatic crystals and subhedral alite crystals were observed in this clinker (Figure 15 a, c and b, d respectively). Alite crystals in C exhibit twinning. Figure 16. Reflected Light Microscopy Images of Clinker D2 Alite crystals in the D2 clinker are much smaller than in C (Figure 16b illustrates the typical size of alite grains in D2), although alite grains appear to be much larger around the belite clusters (Figure 16a,c). Crystals exhibit twinning and are predominantly subhedral. Clear zoning inside alite crystals is depicted in Figure 16d and is typical for the D2 clinker. This appearance of alite in commercial clinkers has been studied by other scientists [1]. It was concluded that the core of the alite grains usually consists of the M1 polymorph, while the outside is an M3. The difference in color has been attributed to the higher concentration of impurities in the core.

PAGE 65

53 Figure 17. Reflected Light Microscopy Images of Clinker E Typical size of alite crystals in E clinker is illustrated in Figure 17a and 17d. Crystals are predominantly subhedral; most of them are zoned and exhibit twining (Figure 17d). P clinker also has subhedral alite crystals that are similar in size to alite crystals in E (Figure 18). Most of the crystals are twined and do not exhibit zoning.

PAGE 66

54 Figure 18. Reflected Light Microscopy Images of Clinker P Overall, clinker C has the largest alite grains, followed closely by E and P. D2 clinker has the finest C3S crystals; therefore, it would have the most reactive C3S. Alite in cements E, and P was colored light blue and in C it was stained dark brown to dark blue. In D2, most of the alite was stained light blue, although in some regions alite crystals have a darker color. This difference in color between clinkers could be attributed to the difference in amount or type of substitutional atoms in the alite of C clinker compared to the other clinkers [1]. There was no belite observed in the C clinker, except for the small inclusions inside the C3S grains illustrated in Figure 15. These inclusions were observed in other clinkers as well. In D2 clinker, belite is very localized, forming streaks and clusters

PAGE 67

55 (Figure 16). Belite in E clinker was also observed in clusters, but they were much larger than the belite clusters in D2 (Figure 17b). The matrix inside these clusters in E is for the most part completely devoid of any C3A, although C3A was observed in the areas where alite grains were intermixed with belite grains. Belite in P clinker is well dispersed throughout the sample and has larger grains than belite in other clinkers. Figure 18 images a, c, d illustrate the average size of belite and alite grains in clinker P. It should be noted, that the size of C3A grains in images a and d are uncharacteristically large for P clinker. Belite crystals in all the clinkers, except for C, are rounded, with well-defined lamellar structures. The absence of belite in C implies that no strength gain should be expected at later ages, after the C3S hydration has completed. All the clinkers have a well-defined matrix. P clinker has the smallest average size of C3A crystals, which implies that C3A in P is highly reactive compared to the C3A in other clinkers. Figure 18b displays the average grain size of C3A in this clinker, although it does include some unusually large belite particles. C3A particles in C, D2, and E have similar size and, therefore, similar reactivity. 3.5 Durability Durability of cements was determined in this study by measuring length change of mortar bars as well as compressive strength of mortar cubes in sodium sulfate environment. These are standard tests used in industry and specified by the American Society of Testing and Materials. Further, ettringite formation in mortar, as well as formation of other phases, was observed through powder x-ray diffractometry and scanning electron microscopy.

PAGE 68

56 3.5.1 Length Change of Mortar Bars in Na2SO4 Solution Behavior of mortar or concrete bars in a sulfate solution is generally divided into two stages. The first stage, also referred to as the induction period, is characterized by relatively minor changes in length. In the beginning of the second stage, expansion increases dramatically and the same rate of expansion is maintained until failure [40]. Figure 19 displays the expansion of the mortar bars prepared with as-received cements. P cement had the shortest induction period of approximately 105 days. The induction period for cements C and D2 lasted for 120 days, while for cement E it was 270 days. Since D cement expanded so little, the end of its induction period could not be determined. P bars exhibited the highest overall expansion. At 270 days, these bars expanded by 0.89 %. They could not be measured after the age of 270 days because they have expanded beyond the capabilities of the comparator. C bars had the second highest overall expansion, and the expansion of D2 bars was only slightly less. At 270 days, C bars expanded by 0.61%, and expansion of D2 bars at that age was 0.52 %. D2 bars could not be measured after the age of 300 days, at which time their expansion was 0.66 %. C bars were measured one more time at 330 days, and their expansion was recorded to be almost 1 %. After 330 days, C bars became too large to fit in the comparator. D bars had the lowest overall expansion. At 360 days, their length change was only 0.09 %, and at 270 days expansion of D bars was 0.05%. E bars had the second lowest overall expansion, which was 0.10 % at 270 days and 0.21 % at 360 days.

PAGE 69

57 0.000 0.200 0.400 0.600 0.800 1.000 1.200 050100150200250300350400 Time (days)% Expansion C D D2 E P Figure 19. Length Change of Bars Prepared with As-Received Cements It is obvious from Figure 19 that these cements can be divided into four groups according to their expansion behavior: P (highest expansion), C and D2 (high expansion), E (low expansion), and D (virtua lly no expansion). High rapid expansion of P cement is readily explained by its high C3A content, 9 % according to the Rietveld refinement analysis compared to 3 1 % for cement C, D2, and E. Excluding P, it is interesting to note that expansion is following the fineness trends. It seems that the coarser the cement, the more durable it is in a sodium sulfate environment. D cement has the lowest Blaine fineness and it also has the lowest expansion, in spite of its high C3S content suggested by Bogue (Table 5). This phenomenon could be attributed to the fact that a larger particle size would result in the slower hydration of C3S and, therefore, a slower release of calcium hydroxide. High concentrations of calcium hydroxide lower C3A solubility and change the mechanism of ettringite formation. If the CH concentrations are low, C3A would react at a normal rate and ettringite would be formed by a through-solution mechanism in D bars. Delayed hydration of tricalcium silicate,

PAGE 70

58 together with small amount of tricalcium aluminate present in this cement (3 % Bogue), probably allowed for complete hydration of C3A while the mortar was still plastic. Thus, any volumetric changes caused by ettringite formation would have been easily accommodated. Since the other cements had very similar fineness and similar C3A content, their expansion could not be attributed to these parameters. Cements C and D2 have a similar C3S content, which is much higher than that of E. C cement had the highest C3S content out of all the cements (67 % Rietveld), which explains its high expansion rate. D2 cement had the second highest C3S content (61 %) and its overall expansion was only slightly lower than that of C cement. Low expansion of cement E can be explained by its low C3S content (54 %) as well as its low C3A content (4 %). Table 8. Length Change of Mortar Bars Prepared with As-Received Cements at 300 Days Mortar Bars C D2 E Expansion (%) 0.772 0.660 0.130 C3S (% Rietveld) 67 61 54 To confirm the hypothesis that expansion behavior of C, D2, and E cements is related to their C3S content, expansion of the mortar bars at 300 days was plotted against the C3S content as well as against the C3S/ C2S ratio. The correlation between expansion and C3S content is stronger than between expansion and C3S/ C2S ratio.

PAGE 71

59 y = 0.0501x 2.5194 R2 = 0.9039 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 40455055606570 C3S content% Length change Figure 20. Expansion of Mortar Bars Prepared with As-Received Cements at 300 Days vs. C3S Content of Cements Determined by Rietveld Refinement y = 0.2697x 0.3693 R2 = 0.8226 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 012345 C3S/C2S% Length change Figure 21. Expansion of Mortar Bars Prepared with As-Received Cements at 300 Days vs. C3S/ C2S Ratio of Cements Determined by Rietveld Refinement

PAGE 72

60 Table 9. Compound Composition of As-Received and Doped Cements Based on Bogue, Internal Standard, and Rietveld Refinement Methods Cement Compound Method of determination C C69D D2 D269D E E69D P P69D Bogue 60 69 65 69 57 69 48 69 Internal standard 70 77 63 67 58 70 55 73 C3S (%) Rietveld 67 74 61 66 54 68 53 72 Bogue 7 7 6 6 6 6 7 7 Internal standard 3 4 3 3 4 5 6 7 C3A (%) Rietveld 2 3 3 3 4 5 9 8 In order to isolate the effect of C3S and eliminate variability between cements, the C3S content of as-received cements was increased to 69 % C3S based on the Bogue calculation, and expansion behavior of doped cements was compared to the expansion behavior of as-received cements. Figures 22-26 show expansion plots of doped cements compared to the as-received cements. All the doped cements showed higher expansion than as-received cements except for D2-69D. The C3S content of D2 cement was only increased by 4 % Bogue (5 % Rietveld). Evidently, this amount is not enough bring about significant expansion at 150 days. It should also be noted that C3S in D2 cement is very reactive, as illustrated in section 3.4. It is possible that C3S particles in the monoclinic alite used to dope the cements are much larger than the alite in D2, which might delay the expansion of D2-69D bars.

PAGE 73

61 0.000 0.200 0.400 0.600 0.800 1.000 1.200 1.400 1.600 1.800 2.000 050100150200250300350400 Time (days)% Expansion C C-69D Figure 22. Expansion of As-Received Ce ment C and Doped Cement C-69D 0.000 0.200 0.400 0.600 0.800 1.000 1.200 1.400 1.600 1.800 2.000 0100200300400 Time (days)% Expansion D2 D2-69D Figure 23. Expansion of As-Received Ceme nt D2 and Doped Cement D2-69D Addition of C3S shortened the induction period of C, E, and P cements. Induction period for bars prepared with as-received cement C lasted for approximately 120 days, while for C-69D bars it was 105 days. The acceleration of expansion for E and P bars

PAGE 74

62 was much more dramatic. The end of the induction period for E bars occurred around 270 days, while for E-69D bars it was only 120 days. Induction period of P bars was approximately 105 days, while for P-69D bars it lasted for less than 56 days. 0.000 0.200 0.400 0.600 0.800 1.000 1.200 1.400 1.600 1.800 2.000 050100150200250300350400 Time (days)% Expansion E E-69D Figure 24. Expansion of As-Received Cement E and Doped Cement E-69D 0.000 0.200 0.400 0.600 0.800 1.000 1.200 1.400 1.600 1.800 2.000 050100150200250300 Time (days)% Expansion P P-69D Figure 25. Expansion of As-Received Cement P and Doped Cement P-69D It is obvious that increasing C3S content decreased cements sulfate durability. E cement showed low expansions in the as-received condition, but after doping its expansion was increased by on order of magnitude. The effect of increasing the C3S

PAGE 75

63 content was even more pronounced in cement P because of its high C3A content (9 %). Even when the C3S content of the doped cements was recalculated based on the Rietveld refinement results for the as-received cements, P cement had a slightly lower C3S content than C cement. However, P-69D bars expanded a lot more than C-69D bars, which can only be attributed to the higher C3A content of P-69D cement. 0.000 0.200 0.400 0.600 0.800 1.000 1.200 1.400 1.600 1.800 2.000 050100150200250 Time (days)% Expansion C-69D D2-69D E-69D P-69D Figure 26. Expansion of Doped Cements 3.5.2 Compressive Strength of Mortar Cubes Compressive strength of mortar cubes stored in saturated lime solution is displayed in Figure 27. E cement has the highest compressive strength of all the cements at 180 days. Compressive strength of C cubes is almost identical to that of E cubes at 180 days, and P and D2 cubes have the lower compressive strengths. The high strength of E cement can be attributed to its high C2S content. Generally, early-age compressive strength, up to 28 days, is controlled by the C3S content of cement, while later-age strength development is governed by C2S content [1]. Although P cement has a similar

PAGE 76

64 amount of C2S, it has larger belite grains than E clinker and a higher alkali content, which can explain its low compressive strength. 0 1000 2000 3000 4000 5000 6000 7000 8000 050100150200 Time (days)Compressive strength (ps i C D2 E P Figure 27. Compressive Strength of Mortar Cubes Prepared with As-Received Cements Stored in Saturated Lime Solution Figures 28-31 illustrate the strength deterioration of mortar cubes prepared with as-received cements in sodium sulfate solution. While all the cubes experienced a drop in strength when placed in the sodium sulfate solution, C cubes were affected the most. At early ages CS cubes stored in sulfate exhi bited higher strengths than C cubes stored in lime. However, after 28 days their strength started to drop and at 180 days CS cubes experienced a strength decrease of almost 1200 psi.

PAGE 77

65 0 1000 2000 3000 4000 5000 6000 7000 8000 050100150200 Time (days)Compressive strength (psi ) C CS Figure 28. Compressive Strength of Mortar Cubes Prepared with Cement C, Stored in Lime (C) and Sodium Sulfate (CS) Solutions 0 1000 2000 3000 4000 5000 6000 7000 050100150200 Time (days)Compressive strength (psi) D2 D2S Figure 29. Compressive Strength of Mortar Cubes Prepared with Cement D2, Stored in Lime (D2) and Sodium Sulfate (D2S) Solutions

PAGE 78

66 0 1000 2000 3000 4000 5000 6000 7000 8000 050100150200 Time (days)Compressive strength (psi) E ES Figure 30. Compressive Strength of Mortar Cubes Prepared with Cement E Stored in Lime (E) and Sodium Sulfate (ES) Solutions 0 1000 2000 3000 4000 5000 6000 7000 8000 050100150200 Time (days)Compressive strength (psi ) P PS Figure 31. Compressive Strength of Mortar Cubes Prepared with Cement P Stored in Lime (P) and Sodium Sulfate (PS) Solutions

PAGE 79

67 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 050100150200 Time (days)Compressive strength (psi) CS D2S ES PS Figure 32. Compressive Strength of Mortar Cubes Prepared with As-Received Cements Stored in Sodium Sulfate Solution Compressive strengths of doped PS-69D cubes at 28 days was almost identical to the strength of PS cubes, although after 28 days the strength of PS-69D started to decline. Compared to their 28-day strength, PS-69D cubes experienced a compressive strength decrease of 1000 psi, and at 120 days the strength of PS-69D cubes was lower than the strength of PS cubes by 1500 psi. The negative effect of increasing the C3S content of cement was much more pronounced in cement P because of its high C3S content. As it was proposed in section 3.5.1, increasing C3S decreases the C3A hydration rate, thus making P cement particularly vulnerable to rapid sulfate attack.

PAGE 80

68 0 1000 2000 3000 4000 5000 6000 7000 8000 050100150200 Time (days)Compressive strength (psi) PS PS-69D Figure 33. Compressive Strength of Mortar Cubes Prepared with As-Received Cement P and Doped Cement P-69D St ored in Sodium Sulfate Solution 3.5.3 X-Ray Diffraction Analysis of Mortar Bars The XRD analysis performed on the bulk sections of bars prepared with asreceived cements shows that for cements with similar C3A content (C, D2, and E) both the amount of ettringite and the amount of gypsum seem to increase with increasing C3S content of cement and the observed expansion of mortar bars. Table 10 and Figure 34 illustrate that C bars, which experienced the highest overall expansion and had the highest C3S content, contained the highest amounts of ettringite and gypsum out of these three cements. Although P bars contained higher amounts of gypsum than C bars and ettringite content of P bars was second highest overall, it should be noted that P bars had a much higher C3A content than the rest of the as-received bars and therefore should not be compared to them directly.

PAGE 81

69 Table 10. Expansion and Ettringite Intensity Ratios (Bulk) for Mortar Bars Prepared with As-Received Cements Mortar Bars C D2 E P Age of Expansion Measurement (days) 330 300 360 270 Expansion (%) 0.981 0.660 0.209 0.887 Ettringite Intensity Ratio 0.44 0.32 0.20 0.38 C3S (% Rietveld) 67 61 54 53 C3A (% Rietveld) 2 3 4 9 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 Ettringite GypsumIntensity Ratio C-360days D2-360days E-360days P-360days Figure 34. XRD of Bulk Sections of Mortar Bars Prepared with As-Received Cements Analysis of the outside surfaces of the P-69D bar showed that it contained higher amounts of ettringite and almost twice the amount of calcium hydroxide found in P bars.

PAGE 82

70 A significant quantity of gypsum was observed in the surfaces of P-69D bars, while this phase was present in the surfaces of P bars only in minute amounts. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 EttringiteGypsumCHIntensity Ratio P-360 days P-69D-160 days Figure 35. XRD of Outside Surfaces of Mortar Bars Prepared with As-Received P Cement and Doped P-69D Cement The x-ray diffraction analysis of the mortar bars confirms that an increase in the C3S content of cements, for the C3A content of approximately 4 %, results in a higher ettringite formation in mortar. Increasing the C3S content at higher C3A values (approximately 9 %) resulted in the increased formation of ettringite, gypsum, and calcium hydroxide in the outside surfaces of the bar. 3.5.4 SEM Scanning electron microscopy observations of the as-received bars revealed that spherulites were present in the cross-sections of all the bars that experienced high expansion, namely C, D2, and P. It is not clear whether these ettringite formations are

PAGE 83

71 responsible for high expansions of these cements or, as Famy and Taylor [41] suggest, just a product of ettringite recrystallization in existing voids. Figure 36. Ettringite Spherulite in the C Bar Cross-Section As it can be seen in Figure 36, the pores of C bars were populated with large clusters of ettringite spherulites of rather significant size. They were very similar in size and number to those observed in P (Figure 46), but much larger in diameter and much more numerous than the spherulites in D2 bars. In fact, the diameter of the spherulites in C and P bars was 2 to 4 times larger than their diameter in D2 (Figure 45). Spherulites in P-69D bar varied in size from having similar diameter to spherulites in D2 to larger than spherulites in C (Figure 47), although P-69D bars were analyzed at a much younger age (160 days compared to 360 days for other bars).

PAGE 84

72 Figure 37. Ettringite Crystals in the Cross-Section of Bar P-69D Figure 38. Needle-Like Ettringite Crystals in the Cross-Section of Bar E

PAGE 85

73 While in other samples ettringite crystals were disjointed, forming only local clusters, in P-69D ettringite crystals seemed to form an interconnected web of crystals that was covering the whole sample (Figure 37). E bars generally had prismatic crystals as illustrated in Figure 38. In conclusion, it was observed under the SEM that high C3S content in cement seems to affect ettringite morphology. In bars prepared with high C3S cements and experiencing high expansion, ettringite crystals were generally much finer than in cements with low C3S. In high C3S cements ettringite usually occurred in the form of spherulites comprised of long needle-like crystals, while in low C3S cement ettringite generally had prismatic crystals. In the cross-section of P-69D bar, which experienced the highest degree of expansion, ettringite was observed forming an interconnected “web” of crystals that possibly originated from the spherulites.

PAGE 86

74 CHAPTER 4 CONCLUSIONS AND RECOMMENDATIONS It was concluded at the end of this study that cements with high tricalcium silicate content generally have poor durability in sodium sulfate environment. All the cements experienced higher expansion with increased C3S content. Not all the cements with increased C3S content started to show a drop in compressive strength at 120 days, although it is believed that testing at further ages would demonstrate strength deterioration. Expansion rate and strength decline of a high C3S and high C3A cement was particularly rapid. High C3S content combined with high C3A content is particularly detrimental to cement’s resistance to sodium sulfate attack. It is suggested that the deleterious action of tricalcium silicate is due to its effect on tricalcium aluminate hydration. A high C3S content of cement corresponds to high calcium hydroxide concentration in the pore solution during hydration. High calcium hydroxide concentrations lower C3A solubility, forcing ettringite to form topochemically around the C3A grains. Thus, there is still some unhydrated C3A remaining after the paste had hardened. When this cement is placed in a sulfate environment, the interaction of sulfates with unreacted C3A would result in ettringite formation in the hardened paste and disruption of the concrete matrix.

PAGE 87

75 It is suggested that a limit should be placed on C3S content of sulfate-resistant cements together with the existing limit on C3A. Further investigations on the impact of high C3S content on the early-age ettringite formation and morphology as well as the impact of fineness on durability of high C3S cements are recommended.

PAGE 88

76 REFERENCES 1. Hewlett, Peter C, ed. (1998). Lea’s Chemistry of Cement and Concrete, 4th ed. John Wiley & Sons, Inc., New York, NY. 2. Mindess, S., Young, J.F. (1981). Concrete. Prentice-Hall, Inc., Englewood Cliffs, N.J., pp. 548-551. 3. Taylor, H. F. W. (1997 ). Cement Chemistry, 2nd ed Thomas Telford Publishing, London, UK. 4. Taylor, J. C., Hinczak, I., Matulis, C. E. (2000). “Rietveld Full-Profile Quantification of Portland Cement Clinker: The Importance of Including a Full Crystallography of the Major Phase Polymorphs.” Powder Diffraction Vol. 15, No. 1, pp. 7-18. 5. Odler, I., Wonnemann, R. (1983). “Effect of Alkalies on Portland Cement Hydration: I. Alkali Oxides Incorporated into the Crystalline Lattice of Clinker Minerals.” Cement and Concrete Research Vol. 13, pp.477-482. 6. Odler, I., Wonnemann, R. (1983). “Effect of Alkalies on Portland Cement Hydration: II. Alkalies Present in Form of Sulfates.” Cement and Concrete Research Vol. 13, pp.771-777. 7. Irassar, E. F., Gonzlez, M., and Rahhal, V. (2000). “Sulfate Resistance of Type V Cements with Limestone Filler and Natural Pozzolana.” Cement and Concrete Composites Vol. 22, pp. 361-368. 8. Skalny, J., Marchand, J, Odler, I. (2002). Sulfate Attack on Concrete. Spon Press, New York, NY.

PAGE 89

77 9. Deng, M., Tang, M. (1994). “Formation and Expansion of Ettringite Crystals.” Cement and Concrete Research Vol. 24, pp. 119-126. 10. Ogawa, K., Roy, D. M. (1981). “C4A3 S Hydration, Ettringite Formation, and Its Expansion Mechanism: I. Expansion; Ettringite Stability.” Cement and Concrete Research Vol. 11, pp. 741-750. 11. Ogawa, K., Roy, D. M. (1982). “C4A3 S Hydration, Ettringite Formation, and Its Expansion Mechanism: II. Microstructural Observation of Expansion”. Cement and Concrete Research Vol. 12, pp.101-109. 12. Mehta, P. K. (1973). “Mechanism of Expansion Associated with Ettringite Formation”. Cement and Concrete Research Vol. 3, pp.1-6. 13. Mehta, P. K. (1976). “Scanning Elect ron Micrographic Studies of Ettringite Formation”. Cement and Concrete Research Vol. 6, No. 2, pp.169-182. 14. Mehta, P.K. (1992). “Sulfate Attack on Concrete -A Critical Review.” Materials Science of Concrete. J. Skalny, ed. American Ceramic Society, Westerville, OH. Vol. 3, pp. 105-130. 15. Yang, S., Zhongzi, X., and Mingshu, T. (1996). “The Process of Sulfate Attack on Cement Mortars.” Advanced Cement Based Materials Vol. 4, pp. 1-5. 16. Odler, I., Colan-Subauste, J. (1999). “Investigations on Cement Expansion Associated with Ettringite Formation.” Cement and Concrete Research Vol. 29, pp.731-735. 17. Yang, S., Zhongzi, X., and Mingshu, T. (1996). “A Reply to Discussion by B. Mather of the Paper “The Process of Sulfate Attack on Cement Mortars.” Advanced Cement Based Materials Vol. 5, pp. 111-112. 18. Tian, B., Cohen, M. D. (2000). “Expansion of Alite Paste Caused by Gypsum Formation during Sulfate Attack.” Journal of Materials in Civil Engineering February 2000, pp. 24-25.

PAGE 90

78 19. Tian, B., and Cohen, M. D. (2000). “Does Gypsum Formation during Sulfate Attack on Concrete Lead to Expansion?” Cement and Concrete Research 30, pp. 117-123. 20. Gonzalez, M. A., and Irassar E. F. (1997). “Ettringite Formation in Low C3A Portland Cement Exposed to Sodium Sulfate Solution.” Cement and Concrete Research Vol.27, pp. 1061-1072. 21. Hime, W. G. and Mather, B. (1999). ““Sulfate Attack” or Is It?” Cement and Concrete Research Vol.29, pp. 789-791. 22. Mehta, P. K. (1983). “Mechanism of Sulfate Attack on Portland Cement Concrete – Another Look”. Cement and Concrete Research Vol. 13, pp.401-406. 23. Ping, X., Beaudoin, J. J. (1992). “Mechanism of Sulfate Expansion. I. Thermodynamic Principle of Crystallization Pressure”. Cement and Concrete Research Vol. 22, pp.631-640. 24. Ping, X., Beaudoin, J. J. (1992). “Mechanism of Sulfate Expansion. II. Validation of Thermodynamic Theory”. Cement and Concrete Research Vol. 22, pp.845-854. 25. Odler, I. (1991). “Expansive Reactions in Concrete.” Materials Science of Concrete. J. Skalny, ed. American Ceramic Society, Westerville, OH. Vol. 2, pp. 221-247. 26. Mehta, P. K., Pirtz, D., Polivka M. (1979) “Properties of Alite Cements.” Cement and Concrete Research Vol. 9, pp. 439-450. 27. Cao, H. T., Bucea, L., Ray, A., Yozghatlian, S. (1997). “The Effect of Cement Composition and pH of Environment on Sulfate Resistance of Portland Cements and Blended Cements.” Cement and Concrete Composites Vol.19, pp. 161-171. 28. Monteiro, P. J., Kurtis, K. E. (2003). “Time to Failure for Concrete Exposed to Severe Sulfate Attack.” Cement and Concrete Research Vol. 33, pp. 987-993.

PAGE 91

79 29. Ferraris, C. F., Clifton, J. R., Stutzman, P. E., Garboczi E. J. (1997). "Mechanisms of Degradation of Portland Cement-Based Systems by Sulfate Attack." Proc. of MRS Nov. 1995, Mechanisms of Chemical Degradation of Cement-Based Systems Ed. K.L. Scrivener and J.F Young, pp. 185-192. 30. Stutzman, P. E. (1996). "Guide for X -ray Powder Diffraction Analysis of Portland Cement and Clinker." National Institute of Standards and Technology Internal Report 575. 31. Gutteridge, W. A. (1979). "On the Dissolution of the Interstitial Phases in Portland Cement." Cement and Concrete Research, Vol. 9, pp. 319-324. 32. Arenes-Oliva, G. A. (2002). Influence of the Variation of Cement Source on Fresh and Hardened Concrete Properties Master’s Thesis, University of South Florida. 33. Courtial, M., de Noirfontaine, M. -N., Ga secki, G., Signes-Frehel, M. (2003). “Polymorphism of Tricalcium Silicate in Portland Cement: A Fast Visual Identification of Structure and Superstructure.” Powder Diffraction Vol. 18, No. 1, pp. 7-15. 34. Jenkins, R., Snyder, R. L. (1996). Introduction to X-ray Powder Diffraction. John Wiley & Sons, Inc., New York, NY. 35. Nishi, F., Takeuchi, Y., Maki, I. (1985). “Tricalcium silicate, Ca3O(SiO4): The Monoclinic Superstructure.” Zeitschrift Fur Kristallographie Vol. 172, No. 3-4, pp. 297-314. 36. Mumme, W. G. (1995). “Crystal Structur e of Tricalcium Silicate from a Portland Cement Clinker and Its Application to Quantitative XRD Analysis.” Neues Jahrbuch Fur Mineralogie-Monatshefte No. 4, pp. 145-160. 37. McKie, D. and McKie, C. (1974). Crystalline Solids. John Wiley & Sons, Inc., New York, NY. 38. Giacovazzo, C. (2002). Fundamentals of crystallography, 2nd ed. Oxford University Press, New York, NY.

PAGE 92

80 39. Campbell, D. H. (1986). Microscopical Examination and Interpretation of Portland Cement and Clinker. Construction Technology Laboratories, Skokie, IL. 40. Santhanam, M., Cohen, M. D., Olek, J. (2002) “Mechanism of Sulfate Attack: A Fresh Look. Part 1: Summary of Experimental Results.” Cement and Concrete Research Vol. 32, pp. 915-921. 41. Famy, C., Taylor, H. F. W. (2001). “E ttringite in Hydration of Portland Cement Concrete and Its Occurrence in Mature Concretes.” ACI Materials Journal Vol. 98, No. 4, pp. 350-356.

PAGE 93

81 APPENDICES

PAGE 94

82 Appendix A. Calibration Curves y = 0.0426x R2 = 0.9651 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 010203040506070Weight % of sampleArea ratio Figure 39. Calibration Curve for Cubic C3A y = 0.0212x R2 = 0.9674 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 010203040506070Weight % of sampleArea ratio Figure 40. Calibration Curve for C4AF

PAGE 95

83 Appendix A. (Continued) y = 0.2026x R2 = 0.9292 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 051015202530Weight % of sampleArea ratio Figure 41. Calibration Curve for MgO

PAGE 96

84 Appendix B. Rietveld Refinement Profiles Figure 42. Rietveld Refinement for As-Received Cement D2

PAGE 97

85 Appendix B. (Continued) Figure 43. Rietveld Refinement for As-Received Cement E

PAGE 98

86 Appendix B. (Continued) Figure 44. Rietveld Refinement for As-Received Cement P

PAGE 99

87 Appendix C. SEM Images Figure 45. Ettringite Spherulite in the D2 Bar Cross-Section Figure 46. Ettringite Spherulite in the P Bar Cross-Section

PAGE 100

88 Appendix C. (Continued) Figure 47. Ettringite Spherulite in the P-69D Bar Cross-Section