1 STUDY OF CEMENT EPOXY INTERFACES, ACCELERATED TESTING, AND SURFACE MODIFICATION By ANDREW STEWART A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2012
2 2012 Andrew Stewart
3 This dissertation is dedicated to my parents, Dorothy and Marty, and my girlfriend, Jennifer. I would not have been able to complete my body of work without y our consistent love and support.
4 ACKNOWLEDGMENTS I would like to take a moment to thank my encouraging family and inspiring girlfriend. The completion of this body of work would not have been possible without their continued support, reassurance, and unconditional love. My thesis advisor, Dr. Elliot Do uglas is a great mentor, and I am grateful to have had the opportunity to work under his guidance and direction. I would also like to thank Dr. Batich, Dr. Brennan, Dr. Craciun, Dr. Hamilton, and Dr. Mecholsky, for their suggestions and support in serving as members of my committee In addition, I am appreciative of Dr. Lambert and Dr. Gervais for their guidance and assistance during my brief stay in Paris. Without Dr. wo rk would not have been possible. Eric Lambers and Gary Scheiffele were also very helpful in acquiring XPS and FTIR spectra. Finally, I would like to thank the undergraduate students, Brett Schlosser and Matt Faatz, who helped me with my experiments in my final year.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ .......... 9 LIST OF ABBREVIATIONS ................................ ................................ ........................... 13 ABSTRACT ................................ ................................ ................................ ................... 15 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 17 Background ................................ ................................ ................................ ............. 17 Civil Vs Chemistry Per spective ................................ ................................ ............... 17 Goals of This Research ................................ ................................ .......................... 19 2 ACCELERATED DEGRADATION OF EPOXY AND FRPS ................................ .... 20 What is Cement? ................................ ................................ ................................ .... 20 Cement Chemistry and Nomenclature ................................ ............................. 21 Hydration Products ................................ ................................ ........................... 22 Modeling of CSH ................................ ................................ .............................. 22 Accelerated Testing of Concrete ................................ ................................ ............. 26 What is Epoxy? ................................ ................................ ................................ ....... 28 Effect of Temperature ................................ ................................ ....................... 29 Effects of Water ................................ ................................ ................................ 32 Accelerated Testing of Epoxies ................................ ................................ .............. 33 Change in Properties ................................ ................................ ........................ 33 Degradation Mechanisms of Epoxies ................................ ............................... 40 Accelerated Te sting of FRPs ................................ ................................ .................. 49 Change in Properties ................................ ................................ ........................ 49 Change in Properties of FRPs with Concrete ................................ ................... 52 Degradation Mechanisms of FRPs ................................ ................................ ... 59 Summary of Accelerated Testing of Epoxy and FRP Systems ......................... 64 3 NATURE OF THE CONCRETE EPOXY INTERFACE ................................ ........... 66 Motivation ................................ ................................ ................................ ............... 66 Theories of Adhesion ................................ ................................ ....................... 66 Hydrogen Bonding ................................ ................................ ............................ 67 Experimental Procedure ................................ ................................ ......................... 69 FTIR Theory and Technique ................................ ................................ ............. 70
6 NMR Theory and Technique ................................ ................................ ............ 72 Results and Discussion ................................ ................................ ........................... 75 FTIR ................................ ................................ ................................ ................. 75 NMR ................................ ................................ ................................ ................. 80 2D NMR ................................ ................................ ................................ ............ 91 Summary ................................ ................................ ................................ ................ 96 4 EPOXY DEGRADATION BY ENVIRONMENTAL EXPOSURE .............................. 98 Background ................................ ................................ ................................ ............. 98 Experimental Procedure ................................ ................................ ......................... 99 Results and Discussion ................................ ................................ ......................... 101 Change in Mechanical Properties of the Model Epoxy System During Hygrothermal Exposure ................................ ................................ .............. 101 Change in Mechanical Properties of the Model Epoxy System During UV and Water Exposure ................................ ................................ ................... 107 Change in Mechanical Properties of the Commercial Epoxy System ............. 110 Diffusion of Water Into the Model Epoxy System ................................ ........... 112 IR Characterization of Degradation ................................ ................................ 116 Summary ................................ ................................ ................................ .............. 121 5 MODIFICATION OF CEMENT PASTE SURFACES WITH SILANE COUPLING AGENTS ................................ ................................ ................................ ............... 122 Background ................................ ................................ ................................ ........... 122 Experimental Procedure ................................ ................................ ....................... 127 Results and Discussion ................................ ................................ ......................... 130 AFM Meas urements ................................ ................................ ....................... 130 Contact Angle Measurements ................................ ................................ ........ 131 XPS ................................ ................................ ................................ ................ 135 Summary ................................ ................................ ................................ .............. 143 6 GENERAL CONCLUSIONS AN D FUTURE WORK ................................ ............. 145 APPENDIX A MECHANICAL PROPERTIES OF THE MODEL EPOXY SYSTEM DURING HYGROTHERMAL EXPOSURE ................................ ................................ ........... 148 B TYPICAL LOAD DISPLACEMENT CURVES FOR HYGROTHERMALLY EXPOSED SAMPLES AFTER 8 WEEKS ................................ ............................. 149 C MECHANICAL PROPERTIES OF THE MODEL EPOXY SYSTEM DURING UV EXPOSURE ................................ ................................ ................................ .......... 153 D MECHANICAL PROPERTIES OF SI KADUR 300 DURING HYGROTHERMAL EXPOSURE ................................ ................................ ................................ .......... 154
7 E MANOVA RESULTS FOR MODULUS VALUES FOR THE MODEL EPOXY SYSTEM WITH HYGROTHERMAL EXPOSURE ................................ ................. 155 LIST OF REFERENCES ................................ ................................ ............................. 156 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 170
8 LIST OF TABLES Table page 2 1 Cement composition abbreviations ................................ ................................ ..... 21 2 2 Change in Tg of DGEB A cured with different curing agents ............................... 31 2 3 Summary of changes in properties for accelerated testing of epoxy systems .... 39 2 4 Summary of mechanisms of degradation for accelerated testing of epoxy systems ................................ ................................ ................................ .............. 47 2 5 Summary of changes in properties for accelerated testing of FRP systems ....... 56 2 6 Summary of mechanisms of degradation for accelerated testing of FRP systems ................................ ................................ ................................ .............. 63 2 7 Summary of property changes and mechanisms for epoxy and FRP systems ... 65 3 1 Assignment of obser ved peaks for cured cement paste ................................ ..... 76 4 1 Summary of changes in properties for the model epoxy system after hygrothermal exposure ................................ ................................ ..................... 102 4 2 Diffusion coefficients at different water temperatures for the model epoxy system ................................ ................................ ................................ .............. 115 4 3 IR peak assignment of the absorption bands in the model epoxy system ........ 116 4 4 Area comparisons of water absorption peaks relative to the phenyl ring for exposure at various water temperatures ................................ .......................... 118 4 5 Area compa risons of various absorption peaks relative to the phenyl ring for exposure to UV radiation with and without water ................................ .............. 121 5 1 Measurem ent of contact angles using different techniques on various concrete sample types ................................ ................................ ...................... 132 5 2 Apparent surface composition and binding e nergies of treated and untreated cement pastes ................................ ................................ ................................ .. 137 5 3 Binding energy positions and relative amounts of bridging and non bridging atoms ................................ ................................ ................................ ................ 137
9 LIST OF FIGURES Figure page 2 1 Simplified model for hydrated Portland cement proposed by Feldman and Sereda ................................ ................................ ................................ ................ 24 2 2 Jennings model for LD and HD CSH formed after drying ................................ ... 24 2 3 (A) TEM image of clusters of CSH ( B) the modecular model of CSH ................. 25 2 4 Reaction between an epoxide and primary amine ................................ .............. 28 2 5 DGEBA and Jeffamine D230 chemical structures ................................ .............. 31 2 6 Transport of water through nanopore network in epoxy proposed by Soles ....... 32 2 7 DSC measurement of a highly crosslinked epoxy during hygrothermal ................................ ................................ ...................... 35 2 8 Evolution of tan ( ) as a function of temperature for an epoxy with and without water ................................ ................................ ................................ ...... 36 2 9 2D and 3D AFM images of samples exposed to diffe rent environmental conditions ................................ ................................ ................................ ........... 38 2 10 Asynchronou s 2D correlation IR spectra of water in epoxy in the spectral range 2800 3700 cm 1 ................................ ................................ ....................... 43 2 11 ATR IR spectra of various stoichiometries of epoxy and amine curing agent m .............. 45 2 12 Hydrolysis degradation mechanism of DDA cured epoxy ................................ ... 46 2 13 ........................... 46 2 14 Change in pull off strength of bonded interfaces of concrete composites after 628 days ................................ ................................ ................................ ............. 53 2 15 FTIR spectra of IM7/997 specimens after 500 hours of UV or condensation ..... 62 3 1 Hydrogen bonding between water molecules ................................ ..................... 68 3 2 Chemical structures of Bisphenol A, gylcidyl phenyl ether, and n ethyldiethanolamine ................................ ................................ ............................ 70 3 3 ATR IR spectrum of cured cement paste ................................ ............................ 76 3 4 ATR spectra of the neat small molecule epoxy analogs and cement paste ........ 77
10 3 5 ATR spectra of the hydroxyl region of cement samples after treatment ............. 78 3 6 ATR IR spectra of BPA treated samples ................................ ............................ 79 3 7 Subtracted ATR spectra of BPA treated sample s ................................ ............... 80 3 8 13 C NMR spectra of EDA composites ................................ ................................ 81 3 9 1 H MAS NMR spectra of BPA laponite composites ................................ ......... 82 3 10 1 H NMR spectra of EDA laponite composites ................................ .................. 83 3 11 1 H NMR spectra of GPE laponite composites ................................ .................. 83 3 12 13 C MAS NMR spectra of BPA composites ................................ ........................ 84 3 13 The struct ure of jennite ................................ ................................ ....................... 86 3 14 1 H Solid State MAS NMR spectra of cement paste composites ......................... 88 3 15 1 H Solid State MAS NMR spectra of additional BPA loadings ............................ 89 3 16 1 H Solid State MAS NMR spectra of various BPA loadings on cement paste .... 91 3 17 2D 1 H NMR spin diffusion and double quantum results for cement + GPE ........ 92 3 18 2D 1 H NMR spin diffusion and double quantum results for cement + EDA ......... 92 3 19 2D 1 H NMR spin diffusion and double quantum results for cement + BPA ......... 92 3 20 2D 1 H NMR spin diffusion experiments for 20 and 30% BPA. ............................ 94 3 21 29 Si NMR spectra of cement paste before and after treatment with BPA. ........... 96 4 1 Change in color of epoxy dogbones after exposure to water at various temperatures and UV radiation ................................ ................................ ......... 102 4 2 Stress ........................ 104 4 3 Change in modulus for the model epoxy system with hygrothermal exposure 106 4 4 Change in peak stress for the model epoxy system with hygrothermal exposure ................................ ................................ ................................ ........... 106 4 5 Change in strain for the model epoxy system with hygrothermal exposure ...... 107 4 6 Change in modulus with UV and water exposure for the model epoxy system 108
11 4 7 Chan ge in peak stress with UV and water exposure for the model epoxy system ................................ ................................ ................................ .............. 109 4 8 Change in strain with UV and water exposure for the model epoxy system ..... 109 4 9 Change in modulus for Sikadur 300 with hygrothermal exposure ..................... 110 4 10 Change in peak stress with hygrothermal exposure for Sikadur 300 ................ 111 4 11 Change in strain wi th hygrothermal exposure for Sikadur 300 ......................... 111 4 12 Absorption of water by model epoxy system at various temperatures up to 8 weeks e xposure ................................ ................................ ................................ 113 4 13 Fitting of the linear region of the various samples ................................ ............ 114 4 14 Calculation of Ea/R for the model epoxy system ................................ .............. 116 4 15 DRIFT IR sp ectra of the model epoxy system ................................ .................. 117 4 16 ATR IR spectra of samples after 8 weeks of exposure ................................ ..... 119 4 17 ATR IR spectra of samples after 8 weeks of exposure ................................ ..... 120 5 1 The reaction process of an alkoxy silane with a hydroxyl surface .................... 123 5 2 Variation in contact angle a s a result of surface roughness ............................. 125 5 3 Near mirror finish of a ceme nt paste sample after polishing ............................. 128 5 4 2D and 3D AFM height images of polished cement paste ................................ 131 5 5 Contact angle measurements of silane treated cement pastes ........................ 133 5 6 DRIFT IR spectra of 0.4 and 0.5 w/ c ratio cement pastes ................................ 133 5 7 XPS survey of treated and untreated cement pastes ................................ ....... 135 5 8 C1s electron orbital XPS spectra including curve fitting of untreated cement paste ................................ ................................ ................................ ................. 138 5 9 C1s electron o rbital XPS spectra including curve fitting of ATEPS treated cement paste ................................ ................................ ................................ .... 139 5 10 Ca2p3 electron orbital XPS spectra including curve fitting of untreated cement paste ................................ ................................ ................................ .... 139 5 11 N1s electron orbital XPS spectra including curve fitting of ATEPS treated cement paste ................................ ................................ ................................ .... 140
12 5 12 O1s electron orbital XPS spectra including curve fitting of untreated cement paste ................................ ................................ ................................ ................. 142 5 13 Si2p electron orbital XPS spectra including curve fitting of untreated cement paste ................................ ................................ ................................ ................. 142
13 LIST OF ABBREVIATION S APTES Aminopropyltriethoxy silane ASTM American society for testing and materials AFt Alumina ferric oxide trisulphate AFm Alumina ferric oxide monosulphate AFM Atomic force microscopy ATR Attenuated total reflection BPA Bisphenol A CFRP Carbon fiber reinforced polymer C 2 S Di calcium silicate C 3 S Tri calcium silicate C 3 A Tri calcium aluminate C 4 AF Tetra calcium aluminoferrite CSH Calcium silicate hydrate CH Calcium hydroxide DGEBA Diglycidyl ether of b isphenol A DSC Direct scanning calorimetry DMA Dynamic mechanical analysis DRIFT Diffuse Reflectance Infrared Fourier Transform EDA N ethyldiethanolamine FRP Fiber reinforced polymer FTIR Fourier transform infrared FWHM Full width at half maximum GFRP Glass fiber reinforced polymer GPE Glycidyl phenyl ether
14 GPTMS 3 glycidyloxypropyltrimethoxy silane NMR Nuclear magnetic resonance MAS Magic angle spinning PDMS Methoxy terminated polydimethxyl siloxane POPDA Poly(oxypropylene) diamine RMS Root mean square d Tg Glass transition temperature TMA Thermomechanical analysis w/c Water to cement XPS X ray photoelectron spectroscopy
15 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy STUDY OF CEMENT EPOXY INTERFACES, ACCELERATED TESTING, AND SURFACE MODIFICATION By Andrew Stewart May 2012 Chair: Elliot P. Douglas Major: Materials Science and Engineering Adhesion between concrete and fiber reinforced polymer ( FRP ) repair materials is of great importance for increasing the longevity of damaged surfaces in civil infrastructure in the US. This adhesion is dominated by a relatively weak interfacial bond that is attacked by environmental agents including water and UV radiation. Therefore, a strong bond between the two materials is essential in making a successful, long lasting repair. In this study, the fundamental question of the epoxy concrete interface i s investigated in terms of how it exists, how it is attacked by the environment, and how it may be improved to resist the elements. In this study, interactions between epoxy analogs and cement paste were investigated to get a fundamental understanding o f the elusive interfacial bond. Various FTIR and NMR techniques were utilized to demonstrate the nature of this interaction. Only one of the small molecules (Bisphenol A), suggested a small interaction, however, based on the results it appeared that hydr ogen bonding did not take place. The second part of this work focused on accelerated aging two epoxy systems, including a model DGEBA POPDA system, and a commercial system. Mechanical testing was performed along with IR spectroscopy and diffusion modeling to determine
16 the change in properties and mechanisms of degradation. The absorption of water up to 90 C resulted in a decrease in modulus and stress, along with an increase in strain. IR data confirmed the absorption of water within the epoxy network. With UV exposure, oxidation was demonstrated by the appearance of carbonyl peaks in the FTIR spectra. UV and water exposure gave evidence of hydrolysis through a decrease in ether groups and an increase in hydroxyl groups. While samples exposed to UV sho wed less of a loss in modulus as compared to those only exposed to water at the same temperature, the peak stress of samples exposed to UV was lower than those exposed to water. However, the data from the mechanical testing of the UV exposed samples was n ot statistically significant. The final section of this work used silane coupling agents to modify cement pastes. AFM measurements were performed to demonstrate the effectiveness of the polishing technique and uniformity of the surface, which the literat ure indicated was necessary for comparison purposes. The contact angle was shown to increase for the PDMS based silane, decrease for the amino based silane, and not change for the epoxy based silane. XPS data confirmed successful covalent linkages betwee n the cement paste and silane coupling agents as evidenced by an increase in bridging O1s and Si2p electron orbitals after curve fitting.
17 CHAPTER 1 INTRODUCTION Background In 2003 the American Society of Civil Engineers reported that 27.1% of the bridges were structurally deficient. Therefore, they sug gested significant changes need possible repair solution could be the use of fiber reinforced polymer s (FRP s ). Compared to tearing down and rebuilding a bridge, this method represents substantial time and cost savings. The FRP can be wrapped around columns or applied to the tension side of a beam, using a wet layup approach or by bonding a pre cured laminate. Epoxy resins have been widely used for this application due to their high performance in strength, stiffness 1 and resistance to creep 2 Despite the advantages to FRP repair, que stions still exist concrete panels fell on a car at the Big Dig in Boston. The concrete panels were attached to the ceiling using an adhesive anchor, which appeared to fail at the epoxy concrete interface. Another important concern is the issue of bond degradation due to environmental exposure. Many investigations have been done on how environmental exposure affects metal/epoxy 3 glass/epoxy 4 carbon/epoxy 5 and glass/vinylester 6 FRP, but little is known regarding the chemical nature of the bond between epoxy and concrete. Civil Vs Chemistry Perspective Civil engineering, b y definition, deals with the design, construction, and maintenance of physical and naturally occurring structures. Therefore, the size of the materials and dimensions they must consider can range from millimeters to kilometers.
18 Civil engineers have to ha ve a practical consideration of the variables and components of the systems they deal with as they affect things on the macro scal e. On the other hand, chemists are concerned with atoms and interactions at the nano scale, which determine some of the most important properties of materials. The fact that these two disciplines examine different length scales, results in different theories regarding the nature of adhesion between epoxy and concrete. The two main mechanisms are mechanical interlock and chemic al bonding. The mechanical interlocking theory assumes that good adhesion between two materials occurs when an adhesive penetrates into the pores, holes and irregularities of the surface, and locks mechanically to the substrate Numerous investigations h ave shown enhanced bond strength of concrete systems as a result of rough surface preparation using a variety of test methods 7 On the other hand, since good adhesion has been reported between epoxy and smooth surfaces, such as copper or aluminum 8 another theory must also be considered. It is highly possible that a chemical bond between epoxy and concrete exists. While the exact stoichiometry and chemical makeup of cement is still debated, many scienti sts have acknowledged that hydroxyl and silanol groups exist within the material. The presence of highly polar aliphatic hydroxyl and ether groups in epoxy chains can serve as sites for the formation of strong hydrogen bonds between the epoxy and cured ce ment. A few studies that have examined the chemical nature of the interface suggested hydrogen bonds are formed between epoxy and concrete 9 and water tends to cluster at the interface and form hydrogen bon ds 4a
19 Goals of This Research In this study, the fundamental question of the epoxy concrete interface is investigated in terms of how it exists, how it is attacked by the environment, and how it may be improved to resist the elements. The epoxy amine syst em used in this study is a common DGEBA and polyoxypropylene diamine, and one other commercial system. In the first part of this work, interactions between epoxy analogs and cement paste were investigated to get a fundamental understanding of the elusive interfacial bond. Various spectroscopic techniques were utilized to investigate if a chemical bond exists (Chapter 3). The second part of this work focused on accelerated aging of the two epoxy systems. Mechanical testing was performed along with IR spe ctroscopy and diffusion modeling to determine the change in properties and mechanisms of degradation (Chapter 4). The last section of this work used silane coupling agents to modify cement pastes. By measuring the contact angle and characterizing the fun ctionalized surface, a quantitative measure of the chemical bond was shown (Chapter 5). The specific aims of this work are as follows: (1) to test the hypothesis that hydrogen bonding exists between epoxy and cement paste; (2) to investigate the change in chemical and mechanical properties of epoxy using accelerated techniques; (3) to functionalize cement surfaces by the use of different silane coupling agents. We hope that this study lead s us to advance our fundamental understanding of epoxy concrete systems allowing us to improve the performance and durability of these composite systems in the construction industry
20 CHAPTER 2 ACCELERATED DEGRADAT ION OF EPOXY AND FRP S What is Cement? The most general meaning of cement is a substance that hard ens and can bind a masonry building material that was composed of crushed rock with burnt lime as a binder. While c ement and concrete are commonly used interchangeably, there are key technical differences Cement is the key ingredient in concrete. It is a finely made powder, that when mixed with water creates cement paste that hardens during hydration and curing. During this process it can bind with rocks, called aggre gates, to form concrete. When only sand particles are used as aggregates, the cured material is referred to as mortar, while when bigger stones are used it is referred to as concrete. In 1824, Joseph Aspdin patented a cement that he called Portland due t o the similarity in color with the stone quarried on the Isle of Portland off the British coast. His method of producing cement, which he developed in his kitchen, was based on a precise ratio of finely ground limestone and clay 10 Current day Portland cement manufacturing is a complex process. The first step in the manufacturing process is obtaining raw materials consisting of limestone, chalk, clay, sand or iron oxide mined from a quarry. Two different methods, wet and dry, are used in the heating process. In the dry process, dry raw materials are proportioned, ground to a powder, blended together and fed to the kiln in a dry state. In the wet process, a slurry is formed by adding water to the properly proportioned raw materials The grinding and blending operations are then completed with the materials in slurry form. After blending, the mixture of raw materials is fed into a tilted rotating, cylindrical
21 kiln where the mixture passes through the kiln at a rate controlled by th e slope and rotational speed of the kiln. Inside the kiln, raw materials reach temperatures of up to 1650 C. Heating releases H 2 O and CO 2 and causes reactions between the solids. The dry process uses more energy in grinding, but less in the heating proc ess in the kiln. At 1480 C, a series of chemical reactions cause the materials to fuse and create cement clinker pellets. Clinker is poured from the lower end of the kiln and transferred to various types of coolers to lower the clinker to handling tempera tures 11 Cooled clinker is combined with gypsum and ground into a fine gray powder that passes through a 75 m sieve. The result of this process is a fine gray powder called Portland cement 12 Cement Chemistry and Nomenclature Cement is such a complex mat erial that in practice, all of the elements a re described as oxides. Table 2 1 indicates the shorthand notation for each of the elements. Table 2 1. Cement composition abbreviations Cement notation formula Traditional formula Name Mass % C CaO Calcium oxide 61 67% S SiO 2 Silica 19 23% A Al 2 O 3 Aluminum oxide 2.5 6% F Fe 2 O 3 Iron oxide 0 6% $ SO 3 Sulfur trioxide 1.5 4.5% H H 2 O Water CO 2 Carbon dioxide Using this nomenclature, the main components of Portland cement are C 3 S (tri calcium silicate, also called Alite), C 2 S (di calicum silicate, also called Belite), C 3 A (tri calicum aluminate), and C 4 AF (tetra calcium aluminoferrite, also called ferrite) 10 The composition of these different components varies depending on the type of Portland
22 cement (Type I, II, III, IV, or V). C 3 S is responsible for the early strength of concr ete and composes 45 to 75% of the total mass C 2 S contributes to later age streng th and hydrates and hardens slowly occupying 7 to 32% of the total mass C 3 A releases a large amount of heat during the first few days of cure and slightly contributes to early strength development. C 4 A F hydrates rapidly but contributes little to the st rength and is responsible for the gray color of hydrated cement 12 Hydration Products the reaction of C 3 S and C 2 S with water, two important products are formed: CSH (calcium sili cate hydrate) and CH (calcium hydroxide). CSH provides the main source 2 and CaO. It forms a gel with limited local crystalline domains, but no long range order 13 CH is the only hydration prod uct with a well defined stoichiometry and crystal structure. When C 3 A and C 4 A F react with gypsum, they form two important groups of products: AFt (alumina ferric oxide trisulphate) and AFm (alumina ferric oxide monosulphate). AFt has the general formula [Ca 3 (Al,Fe)(OH) 6 12 H 2 O] 2 X 3 nH 2 O where X represents a doubly charged anion or, sometimes, two singly charged anions 12 Ettringite is the most common and most i mportant member of the AFt phase in which sulfate is the X. AFm has the general formula 3 CaO(Al,Fe) 2 O 3 CaSO 4 nH 2 O and is also known as monosulfate. Mo de ling of CSH The most abundant reaction product, and main binding phase in cement is the CSH gel. It governs fundamental properties such as strength, ductility, fracture behavior, and durabili ty. It has poor crystallinity and is highly variable in chemical
23 composition, nanostructure, and morphology. For this reason, it has been extremely challenging to characterize at the nanoscale and an ongoing debate surrounding its exact nature continues. Therefore, numerous models for CSH paste exist. Powers and Brownyard were the first to systematically investigate cement paste in the late 1940s. Their model did not distinguish between the different solid phases in cement paste, but referred to them co water is divided into three types: chemically bound, physically bound, and free water 14 Chemically bound water is non e the structure of the hydration products, physically bound water is adsorbed on the surfaces of the hydration products, and free water is contained in the capillary and gel pores of the hydrated cement p aste. Brunauer considered the gel particles described in the Powers Brownyard model consisted of two to three layers of CSH that could roll into fibers. He found that the specific surface area was approximately 200m 2 g 1 as measured by water sorption. L ower values measured by nitrogen were attributed to the failure of nitrogen molecules in entering all the pore space 15 Feldman and Sereda described the gel as a three dimensional assembly of CSH layers with pores of dimensions equal or greater to interla yer spaces. This simplified model is shown in Figure 2 1. They disagreed with value 16
24 Figure 2 1. Simplified model for hydrated Portland cement proposed by Feldman and Sereda 16 The Jennings and Tennis model considers two forms of CSH: high density and low density 17 Their model suggests surface area measurements using nitrogen are a result of mostly low density CSH and assumes high density CSH is not accessible by nitrogen, as shown in Figure 2 2. The main feature of their model is that the CSH is made of globules, which pack together to form LD and HD CSH structures, and finally these units pack together to form the microstructure of CSH. Figure 2 2. Jenning s model for LD and HD CSH formed after drying 18
25 The crystalline calcium silicate hydrate that is considered to be most similar to CSH in cement is tobermorite with an interlayer spacing of 1.4 nm 19 assumes CSH gel consists of a mixture of 14 tobermorite and jennite domains on the nanometer scale 20 Results of both Fuji and Kondo 21 and Cong and Kirkpatrick 22 show evidence that CSH gel exists as a solid solution between tobermorite and Ca(OH) 2 These models are good at describing syntheti c CSH phases, while the Taylor model address CSH obtained through hydration processes of cement. More recently, in 2009, Pellenq et al. developed a molecular model of cement paste from the bottom up beginning with tobermorite with an interlayer space of 1 1 23 The model was further optimized with Monte Carlo and molecular dynamics simulations and validated against XRD, FTIR, and nanoindentation data. The chemical composition of their model was (CaO) 1.65 (SiO 2 )(H 2 O) 1.75 and is visualized below in Figure 2 3. Figure 2 3. (A) TEM image of clusters of CSH (B) the modecular model of CSH. Blue and white spheres are oxygen and hydrogen atoms of water molecules, respectively. Green and gray spheres are inter and intra layer calcium ions respectively. Yellow and red sticks are silicon and oxygen atoms in silica tetrahedra. 23
26 Accelerated Testing of Concrete The hydration process of concrete is a series of complex reactions which occur rapidly at first and slow down over time. The ultimate strength of concr ete, therefore, cannot be measured immediately so standards have been developed to estimate the strength, typically after 28 days. Numerous techniques have been developed to decrease this waiting time, however they have the ir limits. The 3 main methods accepted by ASTM C684 are: 1) Warm Water Method; 2) Boiling Water Method; and 3) Autogenous Method. The warm water method consists of curing standard cylinders (200mm long with a diameter of 100mm) in a 35 C water bath for 24h immediat ely after molding. The specimens are then demolded and tested in compression. The main limitation of this method is that the strength gain is less than half as compared to 28 day moist cured concrete at standard conditions, so job site testing may be re quired 24 The boiling water method consists of standard curing of concrete for 24h, followed by a 3.5h cure in boiling water at 100 C, then is tested 1h later. This method may produce products of hydration that are slightly different from normal curing c onditions. In the autogenous method, specimens are placed in insulated molds made of polyurethane foam immediately after casting and are tested 48h later. No external heat is provided. The strength gain of this method is not high, and is the least acc urate method of the three. Modified techniques for both the warm and boiling water methods have been demonstrated in the literature. Udoeyo et al followed a modified boiling method in which they cured samples in moist conditions for 23h, immersed the sam ples into boiling water for 3.5h, then tested the samples at 28.5h for compression 24 In addition they replaced up to 50% of sand with laterite. The compressive strength increased with
27 the addition of laterite with accelerated samples having between 72 to 84% of the 28 day strength. However, their sample size was relatively small. Naik also used a modified boiling water method in which samples were moist cured for 23h then put in 96 C water for 3.5h, beginning at the 23.5h stage. Th ese samples were t ested in compression at 28h after being allowed to cool for 1h. Naik found that there was little to no influence on the strength due to admixtures, aggregates, or type of cement. However, there was some variation in the strength, as 5% of the test result s were in error of 15 to 20% when compared to the actual 28 day strength 25 The standard boiling water technique, ASTM C 684, was validated by Resheidat and Ghanma on blended cements manufactured in Jordan 26 Their results were compared to other findings using linear regression models which indicated good correlation between accelerated and 28 day samples. High values of coefficients of variation were found for both normal and accelerated samples, but this finding was expected since the samples were not prepared in the laboratory. Tantawi and Gharaibeh made cubes that were placed in an oven ramped up to 93 C in 1h then kept at this temperature for a total of 6h. T he samples were demolded allowed to cool for 30 minutes, and then tested in compression. The water to cement ratio was varied from 0.45 to 0.6 and two cement doses, 300 and 350 kg m 3 were used 27 Good accuracy in prediction of the 28 day strength was found using the accelerated technique. Meyer compared 2 data sets from two different locat ions and time periods in Wellington and Auckland, New Zealand. A straight line fit and a dynamic linear model were used to evaluate the data sets which both indicated smaller residual variances for the warm water method than the hot water method. This w ork also suggested that the
28 methods could be improved by the incorporation of cement chemistry into a master equation and that for super high strength concrete a linear model is not appropriate 28 While accelerated testing of concrete usually is performe d to determine if the ceramic has met compressive strength requirements, there are many more considerations for accelerated testing of polymers. The chemistry and structure properties of epoxy are completely different than concrete, and will be discussed below. What is Epoxy? Epoxy resins are a class of thermosetting polymers that have a broad range of applications depending on the chemical makeup and curing condition of the epoxy system. Most epoxy resins are produced by a reaction between epichlorohydri n and bisphenol A, in which two glycidyl groups, called oxirane or epoxy groups, are attached to the ends of bisphenol A. The molecular weight of this resulting molecule depends on the ratio of epicholorhydrin and bisphenol A. In the ring opening reactio n, the active epoxide groups react with a curing agent, or hardener, to form a highly cross linked, three dimensional structure. Amine curing agents are the most commonly used and the structure and number of amino groups determine the rate of crosslinking and final properties. The reaction between the epoxide group of the epoxy resin with a primary amine is shown below in Figure 2 4. Figure 2 4. Reaction between an epoxide and primary amine While epoxy resins have many applications, this work focuses on their use in construction materials, specifically FR P repair. FRP (Fiber reinforced polymer)
29 composites are increasingly becoming the materials of choice for the repair of damaged concrete structures. These high performance materials, using epoxy as t he matrix, have unique properties that make them especially attractive for use in civil applications due to their quick cure time, good mechanical strength, and easy processing. Prediction of the lifetime and performance of the repairs using these materia ls requires accelerated testing which can include variables such as temperature, humidity, aqueo us solutions, or UV exposure which will be further discussed. Effect of Temperature The physical properties of epoxies are highly sensitive to the effect of tem perature. Increasing the temperature typically produces a decrease in elastic modulus, reduction in tensile strength, and an increase in ductility. As the temperature increases past the glass transition temperature (Tg) almost all of the properties rela ted to its processing and performance are drastically affected 29 The Tg is the most important transition and relaxation phenomenon observed for amorphous polymers. It is a reversible structural change between a hard and relatively brittle state and flex ible, rubbery state. At lower temperatures, in the glassy state, conformational changes are severely restricted, but as the temperature increases past the Tg, motion of side groups begins, followed by large segments, until the entire chain can flow. Free volume theory can be used to explain the physical changes that happen during a glass transition process. At the beginning of the glass transition, a part of the solid polymer turns into a liquid, and the free volume, or the sum of the holes in a polymer d ue to atomic packing irregularities, increases. The increase in free volume corresponds to changes in the interatomic and intermolecular spacing within the polymer. In general, as the viscosity decreases, the free volume increases. With a
30 decrease in vi scosity, or increase in the mobility of polymer chains, there is a change in the heat capacity. The heat capacity in the liquid phase above the Tg is greater than in the solid state below the Tg. This property can be measured by numerous techniques inclu ding differential scanning calorimetry (DSC), dynamic mechanical analysis (DMA), and thermomechanical analysis (TMA). In DSC, the instrument monitors heat flow, or the energy release on heating, between a sample and reference As the heat capacity change s during the transition, there is also a change in heat flow. DMA works by applying a sinusoidal deformation (either a controlled stress or strain) to a sample of a known geometry. DMA measures the ratio of the loss to storage modulus, and the Tg can be curve. Lastly, TMA applies a static force and reports dimensional changes. Coefficients of thermal expansion can be easily measured with TMA when the sample carries a zero or negligible load and the material is allowed to fre ely expand or contract. TMA is significantly more sensitive than DSC for measuring the Tg of crosslinked materials. This difference in measurement is manifested by the variables inherent in the instrument. Since DSC measures heat capacity, the heating rate is a critical parameter, while in TMA, the sample thickness is a critical component. Therefore, correlation of the Tg using the two instruments is not possible because both methods have their own considerations. The glass transition temperature of epoxies can vary greatly depending on the curing agent. Table 2 2 shows the change in the glass transition temperature for diglycidyl ether of bisphenol A (DGEBA) epoxy cured with different curing agents. There is a significant dependence of the Tg on the molecular weight of a polymer
31 system. A linear polymer has higher mobility at the chain ends than the center of the molecule because the chain ends are only bonded to one repeat unit, while the inner repeat units are bonded on both sides. Decreasing the molecular weight of this system results in an increase in the concentration of chain ends, and therefore the average mobility of all the repeats units is increased, resulting in a decrease in Tg 30 This relationship has been shown in the Fox Flory equ ation in equation 2 1 where K is an empirical parameter and M is the molecular weight. (2 1) Table 2 2. Change in Tg of DGEBA cur ed with different curing agents 31 Curing Agent Tg ( C ) DETDA 217 DDM 190 DDS 189 TETA 139 Jeffamine 130 65 Jeffamine 230 47 Jeffamine 800 0 Throughout this dissertation, DGEBA, or EPON 826, and Jeffamine D230, or poly(oxypropylene) diamine (POPDA) were used. EPON 826 has an epoxy equivalent weight of 178 to 186g/equiv, and Jeffamine D230 has an average molecular weight of 230 g/mol, as reported by the manufacturer. Their chemical structures are found below in Figure 2 5. Figure 2 5. DGEBA and Jeffamine D230 chemical structures
32 Effects of Water Epoxy resins can easily absorb up to 7% of their weight, due to moisture from humid environments 32 This is due to the creation of polar hydroxyl groups from the epoxide ring opening reaction, and tertiary amines during the reaction, which form its cross linked structure. Water can form hydrogen bonds with other water molecules or polar groups in the polymer. It acts as a plasticizing agent, reducing the intermolecular forces which hold the macromolecule together. Soles and Yee 33 investigated sub Tg moisture transport in epoxy resins. Although direct experimental observation remains to be performed to verify some details, they suggested that water moves through the epoxy through a network of nanopores, which have an average size of 5 to 6 Polar sites, such as amine functional groups regulate transport of water molecules through the nanopores depending on the orientation of the resin. Figure 2 6 demonstrates how polar sites can either block or allow moisture to traverse the epoxy resin. Figure 2 6. Transport of water through nano pore network in epoxy proposed by Soles 33
33 A study by Choi 34 directly examined some anomalous behavior in which the Tg of some epoxy systems increased with exposure to water at certain temperatures. The three potential mechanisms for this phenomena are s ummarized as 1) post curing induced by the elevated temperature of the water; 2) different states of hydrogen bonding water molecules, one of which induced secondary cross linking; and 3) the effect of the biphasic structure of epoxy. Choi quantitatively evaluated changes in Tg using DSC, and measured reaction progress, or conversion of the epoxide group through near infrared IR, and determined that plasticization by water occurred simultaneously as an increase in cross link density. Samples exposed at 30 showed an increase in plasticization as the amount of water increased while samples at Accelerated Testing of Epoxies Change in Properties The inability to generate pr ecise service life for polymer systems exposed in the field has been a challenge for over a century. Typical field exposures can vary tremendously and involve many years at an exposure site such as Arizona, where it is hot and dry, or Florida, where it is hot and humid. However, this data is usually not repeatable or reproducible, since the weather never repeats itself, and the exposure times make it difficult to conduct tests within the timeframe of materials development or qualification cycles. Thus, acce lerated testing is generally performed in attempt to either predict long term performance or at least provide a relative ranking of materials. This accelerated testing usually involves elevated temperatures, corrosive solutions, UV exposure, or other degra dation techniques. 35
34 Laboratory weathering experiments are an attempt to simulate and accelerate real world degradation conditions. Generally these tests involve exposure to temperatures above those experienced in the field and immersion in water. However, if the exposure temperature is close to or exceeds the glass transition temperature of the polymer (Tg), the mechanism of degradation may be different than under the field condition. Immersion in water can also change the Tg, further complicating the test ing. This can make it difficult to compare results obtained from accelerated testing to those from the field. Numerous studies have shown a decrease in the Tg with exposure to elevated temperature and/or water. Ellis and Karasz measured the Tg for a number of epoxy systems, and found that the reduction in the Tg for stoichiometric compositions could be matched with a compositional model for the Tg depression 36 Notably, they found that for epoxy rich samples t C the Tg after drying was higher than the original Tg. This was explained by the formation of a glycol unit by reaction of the water with the epoxide group. However, an elevated Tg was not found with exposure at room temperature. Amine rich samples showed a greater depression in the Tg than was predicted in the model. In a follow up article, they went on to claim that there is no evidence for tightly bound water and that it is unlikely that water disrupts the hydrogen bond network in the epoxy resin 37 In contrast, Zhou and Lucas claimed two types of bound water can exist in epoxy resins 38 Type I bound water acts as a plasticizer, causing a decrease in Tg and Type II bound water promotes secondary cross linking with hydrophilic groups and limits Tg depres sion. In their epoxy system they found an initial depression of the Tg that was
35 fully recoverable upon drying. This same phenomena was also reported by Bockenheimer et al in a series of papers which examined a high and low crosslinked system under thermal and hydrothermal aging 39 Under hydrothermal conditions reversible phase separation occurred for both systems as well as the appearance of a new, second Tg. In a follow up article Fata and Possart found that the primary Tg of the t hermally aged epoxy initially decreases then remains constant and a substantially lower, secondary Tg is formed that increased with time, as can be seen in Figure 2 7 This secondary Tg vanishes upon heating above the primary Tg. Reheating this system does not recover the primary Tg, and is explained by irreversible plasticization due to bonded water 40 Unfortunately DMA experiments were not performed to verify these conclusions, and it is possible that the heating process was not to a high enough tempera ture or a long enough time to remove the water. Figure 2 7. DSC measurement of a highly crosslinked epoxy during hygrothermal 40 Fredj et al. also found the development of two phases with exposure of epoxy to water, as evidenced b y two peaks in the DMA (dynamic mechanical analysis) tan
36 curve in Figure 2 8 The primary Tg slightly increased over time and the secondary Tg decreased. This was explained by high and low crosslinking zones. They also found a ulus which remained constant after the saturation limit was reached 41 Figure 2 8. Evolution of tan ( ) as a function of temperature for an epoxy with and without water 41 Papanicolou et al. characterized the mechanical and viscoelastic properties of a DGEBA / DETA ( diglycidylether of bisphenol / diethylenetriamine) system using 3 point bend DMA in combination with differential scanning calorimetry (DSC). As expected, there was a decrease in flexural strength and modulus with exposure at increased time and temperature 42 and tan peak decreased with increased exposure time and was explained by the secondary network theory proposed by Zhou and Lucas A decrease in the shear strength of epoxies aged at 85 C and 85% relative humidity was found by Lin et al., who explained their results through the common belief that adsorbed moisture attacks the crosslinks in the epoxy
37 network However, as shown in their Fourier transform infrared spectroscopy (FTIR) data th e epoxy was not fully cured before exposure 43 Other studies have performed accelerated testing of epoxies in different solutions. Yang et al. investigated a commercial epoxy system during exposure to water, salt, and alkali environments. As expected, ex posure led to a decrease in the tensile strength, modulus, and tan delta peak height with increased exposure. A slight depression in the Tg was found initially after which there was insignificant change. Alkali environments showed the most reduction in pro perties 44 A different study exposed a blended epoxy system to dichloromethane, aviation fuel, propylene glycol, hydraulic fluid, DI water, urea, and simulated seawater at 65 C 45 The blended epoxy system showed virtually no change in shear strength when exposed to propylene glycol and hydraulic fluid. Epoxy systems exposed to other solutions showed a reduction in properties over time, with the highest degradation due to dicholoromethane. Cyclic exposure is another method of accelerated testing of materia ls and is often coupled with UV exposure. Recently, Singh et al. found that alternating 3 hour cycles of UV radiation and water vapor condensation at 50 C degraded the flexural strength of epoxy up to 81% and induced the removal of surface layers 46 Expos ure to a constant relative humidity of 80% at 50 C resulted in a 47% reduction in the flexural strength, but interestingly, increased the flexural modulus, while the cyclic exposure decreased the modulus. Shi et al. exposed Epon 828 to alternating cycles o f UV at 55 C and water spray at 25 C with varying times 47 Besides a depression of the Tg with increasing relative humidity, higher indentation recovery using AFM was found at higher relative humidities. They explained this result by relaxation of the plas ticized network.
38 Rezig et al. found a loss in the thickness and an increase in surface roughness after exposure to UV at 50 C, with higher degradation occurring at a higher relativ e humidity, as shown in Figure 2 9 This degradation was inhomogenous. The f ormation of nanoscale pits, which deepened and enlarged with time, was observed with confocal microscopy 48 Figure 2 9. 2D and 3D AFM images of samples exposed to different environmental tdoor in Gaithersburg, MD 48 Other changes in properties have been measured on commercial epoxy systems as well. Hu et al. found that after cyclic exposure to radiation and condensation a decrease in the coating resistance was found after 28 days 49 Small blisters and microcracks were formed on the surface which grew over time. In addition the adhesion strength between the epoxy and steel plates decreased over 50% with exposure. Lim et al. found that after exposure to boiling water their epoxy system showed a lower contact
39 angle, surface potential, and surface resistivity. They explained these findings by the formation of carboxyl groups which cause rapid deterioration of electrical properties 50 In summary, most research found a change in the glass transit ion temperature and a decrease in properties with exposure to temperatures up to 100 C, with and without water or other liquids. A summary of the literature in this area can found below in Table 2 3 in which the type of epoxy, exposure conditions, exposure time, and property changes can be found. Table 2 3. Summary of changes in properties for accelerated testing of epoxy systems Materials Exposure conditions Exposure Time Property Changes DGEBA/mPDA, TGDDM/DDS, Fiberite 934:TGDDM/DDS 38a Water at 45, 60, 75, and 90 C 1530 h Initial decrease in Tg, then increase. Tg fully recoverable upon drying DGEBA/DETA 39 40 C and 60 C in air 112 days Decrease of Tg, recoverable on drying DGEBA/DETA 40 Water at 40C and 60C 79 days Increase in Tg for low crosslinked epoxy, the high crosslinked epoxy shows a new second Tg with aging and reversible phase separation. DGEBA/TETA 42 in water at 60 and 80C 1536 h Increase in Tg with exposure, decrease in tan d Amino amide epoxy and polyamide epoxy with 25% filler 41 DI water and Saline at RT 200 days Increase in Tg1 decrease in Tg2, 35 40% decrease in modulus DGEBA/DDA 51 Exposure of epoxy to air and water at 60C 100 days New secondary glass transition and reversible phase separation Epoxy/Ni Au composite 43 85C 85%RH 550 h 50% loss of shear strength Sikadur 30: aromatic hydrocarbon and aliphatic amine 44 Exposure of epoxy to water, alkali, NaCl, 23 C or 60 C 2years E loss of 30 60%,tensile loss of 40 70% Blend of epoxies and amine curing agents 45 Dichloromethane, aviation fuel, propylene glycol, hydraulic fluid, DI water, urea, simulted seawater at 65 C 730 days Aqueous based system showed a 30% decrease in strength, propylene glycol and hydraulic fluid virutally no change Epon 862/Epikure 3274 46 50C 80%RH, cycles of UV and condensation at 50 C 4098 h Decrease in flexural strength up to 47% in moisture, 81% in UV / condensation Epon 828/Epikcure 3115 47 Various wet and dry cycles at 25 C 56h Higher indentation recovery at higher RH UV radiation at 50C with 9% or 75% RH
40 Materials Exposure conditions Exposure Time Property Changes Dow DER 332/bisaminomethyl cyclohexane 48 140 days Up to 38% loss in thickness under chamber, 18% outdoor, RMS increase up to 0.7um Commercial 2 part epoxy varnish 49 UV radiation at 60C for 8 h then condensation at 50C for 4 h 35 days Decrease in coating resistance after 28 days DGEBA/MNA with BDMA accelerator 50 Boiling water 1000h Decrease in contact angle, surface potential, surface resistivity Degradation Mechanisms of Epoxies In the literature many mechanisms are suggested to explain results of accelerated degradation experiments. One of the most common explanation for the decrease in properties is the plasticization of the epoxy network. Water swells the polymer and reduces th e polymer polymer chain secondary bonding, resulting in a decrease in Tg. Doyle and Pethrick found that for epoxy exposed to water based solvents there was a depression in the Tg. This indicates reduced cohesive forces between polar chains and an increase in polymer motion 45 For non polar solvents, plasticization was coupled with swelling of the resin and was the main suggested mechanism for the change in bond strength. Bockenheimer et al. also cite water as the source of plasticization of the epoxy netw ork in which the water molecules interrupt the physical crosslinks by interacting with the polar groups of the epoxy. Another research group examining epoxies determined that a splitting of the tan peak correlated to the formation of a biphasic structur e. The lower temperature tan peak was due to the water plasticized fraction of the specimen, while the higher temperature peak was due to the dried fraction of the specimen which reflects the Table 2 3. Continued
41 amount of crosslinking 44, 52 They also mention the competing effects of cure progression and plasticization in the early periods of exposure, followed by hydrolysis at longer periods of exposure. Frigione et al. also mentions the concept of plasticization competing with post cure crosslinking at early periods of ex posure 53 The lowering of the Tg with the absorption of moisture allows the polymer chains to become mobile which allows a limited displacement of chain segments which promote post curing. This could explain higher values of Tg for longer periods of time and higher exposure temperatures accompanied with an initial increase then decrease in ductility over time. In general, while many authors cite plasticization as a mechanism for the decrease in properties, they do not confirm t hat the effect is reversible. Thus, the claims of plasticization as a mechanism are somewhat tenuous and overall require further investigation. Another common ly cited explanation for a decrease in properties is centered around the idea that multiple types of bound water can exist in the epoxy network. NMR and FTIR have been used on epoxy resins to determine that water can be bound to the epoxy resin with a mobility between solid and free water 38a, 54 In a series of papers, Zhou and Lucas questioned the polymer glass transition temperature of epoxy systems because the depression of the Tg of epoxies is not a simple matter of the amount of water uptake of the material. They suggested that the Tg is influenc ed by a dual mechanism process. Type I bound water causes a depression of the Tg due to interruption of interchain bonds and Type II bound water offsets that drop through secondary crosslinking resulting from a water resin interaction. Type I bound water f orms single hydrogen bonds while Type II forms
42 multiple hydrogen bonds and has a higher activation energy, making it harder to remove from the resin. Papanicolou et al. used the two types of bound water model to explain their findings. Exposure of their ep oxy to DI water at elevated temperatures resulted in exposure 42 Li et al. used FTIR to analyze water absorbed into 6 different epoxies and fit the hydroxyl stretching re gion with four different components. These regions corresponded to water molecules without a hydrogen bond, one hydrogen bond, and double hydrogen bonds (loose or tight) 55 2D correlation analysis suggested that diffusion was accomplished by water molecul es with loose double hydrogen bonds and the impeding step originated from the rearrangement of local chains to open additional polar sites and the energy required by water molecules to dissociate from the epoxy network. Wu and Siesler used 2D ATR FTIR to i nvestigate the diffusion of water in epoxy networks at room temperature 56 The basic concept of 2D IR experiments are somewhat similar to 2D NMR experiments, but since vibrational relaxation rates are many orders of magnitude faster than spin relaxations, the double Fourier transformation technique developed for 2D NMR experiments is not really applicable. In 2D IR experiments an external perturbation, such as a change in concentration, is applied to the sample which selectively induces time dependent reo rientations of electric dipole transition moments associated with the individual modes of vibration in the system 57 Individual dipole transition moments respond differently to the external perturbation and have unique reorientations rates which can be us ed to identify highly overlapped IR bands. Wu et al. fit the hydroxyl stretching region with 3 components:
43 weak water water hydrogen bonds, strong water water hydrogen bonds, and water epoxy hydrogen bonds as shown i n the shaded regions in Figure 2 10 T hey explained a positive shift in the OH stretching band as an indication of interactions between water molecules and the carbonyl oxygen in the epoxy matrix, as a result of water diffusion into the epoxy network. Over time the epoxy structure limits the movement of water molecules and forces the water molecules to form clusters with other water molecules Figure 2 10. Asynchronous 2D correlation IR spectra of water in epoxy in the spectral range 2800 3700 cm 1 56 2D FTIR spectroscopy was also used by M usto on epoxies exposed to thermal degradation at 200 C 58 He found the disappearance of hydroxyl, CH 2 and CH groups with increasing exposure. The oxygen attack led to the formation of two groups, amide and aldehyde/ketone groups. Oxidation is another mechanism suggested for degradation of epoxy resins. Monney et al. exposed an epoxy system to UV radiation at 42 C in air and confirmed the existence of a thin photo oxidation layer that evolved at a constant rate 59 This was monitore d by ATR IR which indicated the loss of CH and CH 2 groups which form
44 radicals which aid in the production of carbonyl groups, particularly ketone and ester groups. Hong performed thermal degradation of epoxy films on various metal substrates and also foun d the formation of carbonyl groups and the degradation of CH 2 groups using IR spectroscopy 3b He also found evidence for cleavage of nitrogen phenyl bonds in the epoxy backbone. XPS fitting of the C1s spectra indicated carboxyl species were produced and t hat the degradation occurred more rapidly on the copper substrate than the steel or aluminum. Lim and Lee also found the presence of carboxyl groups after exposure to boiling water 50 After exposure they found an increase in the O1s peak height and chang es in the C1s peak, along with a decrease in contact angle, indicating a change in the hydrophilic property of the surface. Meiser et al. also found that exposure in air at 60 and 120 C with moisture created alkyl radicals which produced carbonyl groups, amide groups, and chain scissions 60 Chain scission was indicated by a decrease in the stretching band of ether groups at 1034 cm 1 as shown in Figure 2 11. In this image, the solid lines are on the surface and the dashed lines are at 350 m below the surf ace. Maljati et al. found that water exposure resulted in the oxidation of CH 2 groups which lead s to the formation of carboxylic acids, while the combination of water and UV light caused hydrolysis 61 Similar short network fragment products were found i n both works.
45 Figure 2 11. ATR IR spectra of various stoichiometries of epoxy and amine curing agent m 60 Hydrolysis is an irreversible process that results in chain scission in epoxies re ported in the literature as a result of exposure to water. After 6 day of exposure to DI water at 90 C, Xiao et al. found that the intercrosslinking chain segment containing nitrogen was cut or leeched out during the aging process. This XPS result was conf irmed by FTIR results which found the absence of a characteristic peak of the DDA curing agent. They suggested that water interacted with the tertiary amine groups in the resin to form N H and OH bonds in the polymer chain 62 Fata and Possart also examin ed a DDA cured system and discovered evidence of both plasticization and hydrolysis. They found that thermal aging at 60 C did not cause any chemical modifications, but a new phase was formed that was reversible upon heating above the Tg. Hydrothermal ag ing also produced a new phase, but simultaneously hydrolysis cleaved crosslinks specific to DDA cured epoxy. The hydrolysis cleaved only imino ether like crosslinks, but amine like or ether like crosslinks were unaffected 51 This mechani sm is depicted bel ow in Figure 2 12 with the corresponding wavelengths for
46 the functional groups that are identified using FTIR s pectroscopy as shown in Figure 2 13 Figure 2 12 Hydrolysis degradation mechanism of DDA cured epoxy 51 Figure 2 13 FTIR of the DDA cured 51 Jana and Zhong examined the effect of both water and UV radiation and also suggested hydrolysis as the dominant mechanism 63 UV radiation degraded the polymer network more severely than water alone and, as with the ot her studies, an increase in carbonyl groups was observed due to hydrolysis and photooxidation of alkyl, phenyl, and aromatic ester linkage units.
47 A different mechanism was suggested by Hu et al. who exposed their commercial epoxy to UV and water 49 They h ypothesized that the water and oxygen exposure created soluble degradation products from the epoxy which penetrated into the network along with water to form osmotic cells. The cycling of the UV and water caused osmotic pressure eff ects which le d to the fo rmation of blisters which in turn ruptured, which correspond to a deterioration of properties. Table 2 4 gives a brief summary of the degradation mechanisms for epoxy systems. Essentially, at lower exposure temperatures in the presence of water plasticizat ion is proposed as the dominant mechanism however, true plasticization requires reversibility, which is generally not demonstrated in these studies At higher temperatures and with UV exposure oxidation and hydrolysis are cited as the primary degradation mechanism. Table 2 4. Summary of mechanisms of degradation for accelerated testing of epoxy systems Materials Exposure conditions Exposure Time Mechanism Blend of resins and amine cure agents cured on aluminum 45 Dichloromethane, aviation fuel, propylene glycol, hydraulic fluid, DI water, urea, simulted seawater at 65 C 730 days Plasticization/Swelling DGEBA/DETA 39 40 Water at 40C and 60 C 79 days Plasticization 4,40 isopropylidenephenol epichlorohydrine with an aliphatic amine hardener 52 DI water at 23, 40, 60 C 24 months Plasticization competing with postcuring in early stages, followed by hydroylsis Sikadur 30: aromatic hydrocarbon and aliphatic amine 44 Exposure of epoxy to water, alkali, NaCl at 23 or 60 C 2 years Post cure vs plasticization Sika S50, M16, M20 53 DI water at RT 28 days Plasticization and water substrate interaction vs crosslinking DGEBA/mPDA, TGDDM/DDS, Fiberite 934:TGDDM/DDS 38a Water at 45, 60, 75, and 90 C 1530 h Type I bound water forms a single hydrogen bond with epoxy, Type II bound water forms multiple hydrogen bonds.
48 Materials Exposure conditions Exposure Time Mechanism DGEBA/mPDA, TGDDM/DDS, Fiberite 934:TGDDM/DDS 38b Water at 45, 60, 75, and 90 C 1530 h Type I bound water plasticizes the network, Type II bound water forms a secondary crosslink. DGEBA/TETA 42 in water at 60 and 80 C 1536 h Single or double hydrogen bonds, Type I plasticizes, Type II forms secondary crosslink Novolac epoxy resin 56 DI water at RT 16 min Carbonyl and hydroxyl groups form hbonds with water at low concentrations, but at high concentrations water clusters DGEBA, TGAP, TGDDM, DDS, DDM 55 20 C and 75 C in DI water 13 h Loose double bound water molecules diffuse throug the network TGDDM/DDS 58 200 C in air 1,000 min 2 competitive pathways forming amide or aldehyde and/or ketone. Molecular breakdown with dissapearance of hydroxyl, CH2, and CH groups. DGEBA/MTHPA 59 UV and 42 C 250, 500, and 1000 h Oxidation of matrix and formation of carbonyl groups. Magnobond 6388 3:TGDDM/amidoamine 3b Cu, Al, and steel at 150 C 1, 45, and 90 h Nitrogen phenyl groups broken and carbonyl and amides formed DGEBA/MNA with BDMA accelerator (Lim and Lee 2000) Boiling water 1000 h Oxidation of matrix and formation of carbonyl groups. DGEBA/DETA 60 90%RH, 60 C 120C 300 days Oxidation, hyperperoxide decomposition and Cope elimination, resulting in chain scission. Phenoxy resin 61 UV and salt water cycles 100 h Water cause oxidation of CH2 groups with formation of carboxylic acids. UV and water caused hydrolysis of phenyl formates. DGEBA/DDA 62 DI water at 90 C 6 days Chain scission through hydrolysis DGEBA/DDA 51 Exposure of epoxy to air and water at 60 C 100 days Plasticization and Hydrolysis cleaves imino ether crosslinks Epon 828/Epikure and reactive graphite nanofibers 63 water at 60C and UV 30 days Hydrolysis Commercial 2 part epoxy varnish 49 UV radiation at 60C for 8 h then condensation at 50C for 4 h 35 days Soluble degradation products penetrate the matrix with water, which forms osmotic cells. Table 2 4. Continued
49 Accelerated Testing of FRPs Change in Properties The presence of the fiber in a composite can have a strong effect on the aging behavior of the system compared to neat epoxies. Ramirez et al. exposed their carbon fiber epoxy composite t and 60 C and found a decrease in Tg up to 19 C, a 50% decrease in the flexural strength, 25% decrease in the transverse strength, and an 18% decrease in the modulus 6d They claimed that since carbon fibers do not absorb moisture, absorption took place at the interface through wicking and p ossibly through cracks. Since concrete was not part of the study, they only examined fiber irreversible loss in properties with exposure in their CFRP epoxy system 64 However, they found that the vinylester CFRP systems could regain their flexural strength upon re drying. Adams and Singh found all but one of their epoxy composites had a recoverable loss in shear modulus and interlaminar shear stress 5 The one composite that was not recoverable did not exhibit degradation in thermal conditioning, but only with exposure to steam in the form of delamination along the fibers. This set of composites was not post cured, unli ke the other systems, so it was m ost likely more easily plasticized. Unfortunately, chemical analysis was not performed in the study. Another study by Zhang et al. showed reversible property loss of flexural strength, modulus, and interlaminar shear strength after exposure to boiling wat er at 100 C for 2 days, then drying at 105 C for 1 day 65 When exposed to other solutions, such as salts or alkali, FRP composites showed similar reduction in properties. Chen et al. exposed two FRP systems (E glass / vinylester and carbon / epoxy) to var ious salt solutions under cyclic exposure at
50 different temperatures and found reductions in tensile strength and interlaminar shear stress 66 The solution containing 2:20:2 of NaOH, KOH, and Ca(OH )2 showed the most degradation of all the solutions althoug h the epoxy CFRP samples showed better resistance to the environmental conditions. Micelli and Nanni examined FRP rods consisting of carbon fibers and epoxy/vinylester matrices under cyclic environments including alkaline solution, elevated temperatures, and UV irradiation 67 Alkali aging resulted in up to 40% loss in shear strength, but almost no loss after environmental cycles of freeze thaw, high relative humidity, high temperature, or UV radiation. The samples lost minimal tensile strength in alkali a ging and environmental cycles. The authors recognized a major limitation of the work being the absence of stress during the accelerated aging processes. Murthy et al. exposed vinyl ester and epoxy systems to artificial seawater at room temperature for sev eral weeks. The flexural, tensile, and interlaminar shear strength all decreased with increased exposure with the epoxy system exhibiting more degradation than the vinyl ester based composites 68 Some studies showed little to no change in properties after environmental exposure. The epoxy CFRP composites of Lee and Peppas exhibited no change in Tg with exposure to water at various temperatures 69 At 80 C the interfacial shear strength of the DGEBA system was strongly reduced, while the TGDDM system was no t significantly degraded in any of the exposure systems. Ray found the effect of changing temperature and humidity had little impact degrading the ILSS and shear strength of his GFRP epoxy/polyester composites 70 No chemical analysis was performed so it i s difficult to assess if the samples were trul C.
51 Some studies showed that exposure to accelerated aging conditions had no effect on longitudinal properties, but the transverse directions were significantly degraded. Kumar et al. performed cyclic exposure of condensation and UV light on a com mercial CFRP, with continuous oriented fibers, known as IM7/997 for a period of 1,000 hours They found reduction in the transverse ten sile strength 71 No real changes were found in the longitudinal properties, but the authors warned that over longer periods the accelerated conditions would lead to so much matrix erosion that even the longitudinal properties would be affected. Based on the principles of the rules of mixtures for composite materials, the modulus of the composite will be higher than the modulus of the matrix in the longitudinal direction, so no decrease in the transverse tensile strength or modulus is expected. For compo sites tested in the longitudinal direction the total load sustained by the composite, F c is equal to the loads carried by the matrix phase, F m and the fiber phase F f as shown below in equation 2 1 72 F c = F m + F f (2 1) Hulat et al. also found no real changes in longitudinal properties of epoxy GFRP and CFRPs exposed to wet/dry cycles and salt solutions 73 Also UV radiation and salt solution produced no adverse effects on the ultimate fa C exposure produced some degradation of the composites. However, there was a relatively large amount of error reported in this work, with a coefficient of variance above 10% for many samples. Boualem and Sereir performed a predictive study on the mechanica l degradation properties of a graphite epoxy under hygrothermal aging using quadratic failure criterion in stress space 74 They found that hygrothermal stresses
52 reach their maximum values at first times of moisture diffusion and longitudinal direction pro perties are unaffected, while in the transverse direction the properties are significantly reduced. Change in Properties of FRPs with Concrete Similar trends can be found in the literature regarding accelerated degradation of epoxy FRPs bonded to concrete. Benzarti et al. examined CFRP and unidirectional carbon fiber sheets with four different epoxies bonded to concrete under conditions of 40 C and 95% RH, and found a decrease in pull off strength with exposure, seen in Figure 2 14 75 This decrease was no t consistent with data from shear tests, which did not show any change and was explained by the evolution of the interfacial load transfer which counterbalanced the effect of joint weakening. The aged specimens showed greater strain levels in the free edge region, and over time the load was progressively redistributed over a larger length of the lap joint. However, the concrete substrates underwent a surface treatment and were either sandblasted or diamond ground, which could lead to a more complex interfac e. They also found a drop in the Tg for one type of epoxy, which became more severe over time, while another epoxy system did not exhibit any loss. Some samples of concrete also underwent carbonation treatment. These samples showed higher initial strength and less degradation with exposure than untreated samples.
53 Figure 2 14 Change in pull off strength of bonded interfaces of concrete composites after 628 days 75 A different study by Mufti et al. examined GFRPs made of E glass and vinylesters used in repair conditions in Canada for 5 to 8 years They were exposed to freeze thaw conditions, wet / dry cycles, deicing salts, saltwater, and thermal loading 76 The researchers found no reduction of the Tg of the matrix or cracking of the concrete in any of the samples, which is inconsistent with results from simulated or accelerated laboratory studies. However, the thermal range of the field conditions did not exceed 35 C, which is typically well below the glass transition temperature of most commerc ial epoxy systems. Lyons et al. also found minimal degradation on their FRP composite with various exposures up to 40 days 77 Using modified double cantilever beam samples to determine the Mode I strain energy release rate indicated the only statistically significant loss in toughness was observed at 100 C and 95% RH. However, they used masonry bricks and did not perform a chemical analysis of the epoxy. In
54 contrast, Leveque and Schieffer found their composite had a Tg that increased up to 100 hours of exp osure then decreased afterwards after thermal exposure at 210 C 78 Toutanji and Gomez examined FRPs bonded to type II Portland cement under cyclic environmental conditions, but with salt solution 79 They used three different amine cured epoxy systems with two different carbon fibers and two different glass fibers. E ven with the harsh environments of wet/dry cycling, the beams exhibited an increase in load capacity (compared to samples without FRP) when the FRPs were bonded to the tension side of the concr ete beams. However, the ultimate load decreased as compared to the non exposed samples (up to 33%). The change in properties was attributed to the degradation of the epoxy since the fibers did not break at specimen failure and debonded at the fiber concre te interface. Other studies exhibited a similar reduction in properties. Two different studies by Buyukozturk on commercial epoxy CFRP systems found a sharp reduction in fracture toughness with exposure to DI water at 23 C and 50 C 80 Peel fracture experiments exhibited a greater loss than shear fracture experiments but both methods demonstrated a separation at the epoxy/concrete interface as compared to dry samples which failed in a concrete delamination mode. Interestingly, they fo und that after a certain time, the bond strength approached a value after which no significant degradation occurred and that after drying the composite could not regain its original strength after wet dry cycles. A series of studies by El Hawary in Kuwait examined the effect of sea water on epoxy repaired concrete comparing field use to laboratory results 81 No real loss in tensile or bond strength was found over the course of 18 months of exp osure.
55 Interestingly, the build up of sea shells was reported t o increase the strength of samples exposed directly to the sea water. A major limitation of the work was the small population size of only 2 samples per group. Mourad et al. also found no significant changes in tensile strength or modulus in glass/epoxy composites after 1 year of exposure to seawater at 23 C or 65 C while glass/polyurethane composites exhibited degradation at the fiber/matrix interface 82 Silva also found little to no loss in properties for samples exposed to various environments 83 His samples were commercial GFRP epoxy composites that were exposed to salt fog, humidity cycles, and UV degradation. No change in tensile modulus was found for water immersion at room temperature and UV exposure, and a change of less than 10% loss was observ ed for temperature and salt fog cycles. No change was found in the tensile strength under room temperature water, and less than 12% loss for all other conditions after 10,000 hours of exposure. A series of papers by Silva and Biscaia in simulated tide wat er, and salt fogging indicated the bending load capacity of GFRP composites on concrete was improved after exposure 84 There was no degradation for samples immerse d in DI water and samples in moisture or salt fog cycles showed an initial decrease after 6 ,000 hours that increased by 10,000 hours. Slip was affected by artificial aging and decreased for all conditions with the bond visibly more degraded by the salt fog process. Finite element analysis was performed and was in good agreement with the experime ntal data. The 2D and 3D models matched the experiment with the exception of the maximum bond stress and slip at the interface, where the 3D model gave better representation. Karbhari et al. performed peel tests on 2 different epoxy FRPs on cement mortar t o determine interfacial fracture energies 7b In addition to finding that exposure of the
56 composites to ambient water temperatures depressed the Tg, they also found that exposure of the composites to 15.5 C and freeze thaw cycles of 15.5 C to 20 C actual ly increased the G IC and G IIC values for measurements up to 60 days. Exposure to DI water and synthetic sea water at room temperature resulted in a slight decrease which was more pronounced in the G IIC values. G IIC values are primarily dependent on the pro perties of the epoxy while G IC values are primarily dependent on the properties of the interface. It also appeared that the carbon fiber systems exhibited less change in properties than the glass fiber systems. A summary of the aforementioned findings can be found in Table 2 5 The specific materials, exposure conditions, exposure times, and changes in properties are organized for easy reading. The changes in properties are more complicated than neat epoxy. Overall, a decrease or little to no change in prop erties is found with environmental exposure. Table 2 5. Summary of changes in properties for accelerated testing of FRP systems Materials Exposure conditions Exposure Time Property Changes T700 and AS4 carbon fiber, MAS epoxy, VE D411, VE D8084, VE H922L 6d Salt water, DI water 40 and 60 C 1200 h Decrease in Tg, 50% decrease in flexural strenght, 18% decrease in modulus CFRP with DGEBA and amine hardeners 75 40C 95% RH 20 months Changes in Tg dependent on epoxy type, strength of all epoxies decrease by factor of 2 or 3. composite pull off tests result in decrease, nothing for shear GFRP 76 o 35 to 35 C. outdoor conditions in Canada 5 to 8 years No Change in Tg or degradation TGDDM/DDS DGEBA/DDA and CFRP 69 Water at 50,70,90, and 100 C 1200 h Minimal effect on Tg. Large decrease in interfacial shear strength. Commercial Poly amide cured epoxy with glass fiber tows on concrete masonry bricks 77 RT, 60C 95% RH, 100C 95% RH 163 days 35% decrease in fracture toughness only for 100C 95%RH Carbon fiber / epoxy composite 78 180 and 210 C 2 months Tg increased till around 100h of aging then decreased.
57 Materials Exposure conditions Exposure Time Property Changes Eglass vinylester and CFRP epoxy 66 NaOH, KOH, Ca(OH) 2 NaCl, Na 2 SO 4 RT, 40 C, 60 C, cyclic 120 days Reduction in tensile strength varying from 8 50% difference, 8 30% reduction in interlaminear shear strenght CFRP, epoxy vinylester modified resin 67 Alkaline solution at 22 and 6 0C, 200 cycles 18 to 4 C then 600 cycles 16 to 49 C with UV, then 160 cycles RH of 60 100% at 16 C followed by 27 C then 38 C up to 200 cycles Alkali aging resulted in 30 40%loss tensile, t ransverse loss 0 90% dependant on T and material, environment cycles average 0 5% change on tensile and transverse GRFP/CFRP epoxy composite 68 Seawater 450 days Decrease in ILSS, flexural, and tensile strength Type II cement, 3 different amine cured epoxies, 2 carbon and 2 glass FRPs 79 35 g/l salt at 35 C 90% RH 4h wet 2h dry 75 days 5 to 30% reduction in flexural strength Commercial epoxy and concrete 80b 23C and 50 C in DI water and cyclic exposure 8 weeks Up to 70% decrease in fracture toughness C FRP with amine cured epoxy 80a 23 and 50 C with 50% and 100% RH 8 weeks Up to 60% shear fracture toughness loss at 50C in water Epikote 828 and T300 carbon fibers 64 150C in air, 95 C in water, 1 cycle wet then dry. Testing performed at 25, 50, 80, 100, and 120 C 2 or 120 h Irreversible loss in strength upon drying as compared to vinylesters Ciba Geigy 913, 914, 924 epoxies and E glass, TS carbon, XAS carbon fibers 5 Steam and 100 C over silica gel and 100C 756 h Shear loss and ILSS factor up to 2x recoverable upon drying, except for 913 composites Vicotex 1452 with amide curing agent and T300 CFRP 65 Boiling water at 100C and DI water at 70C 100%RH 2 days exposed, 1000 hrs for RH 30 40% reduction in strength, 50 50% reduction in ILSS, reversible aft er drying. DGEBA based resin and polyester resin with E glass fibers 70 50 C 60%RH to 50 C 95%RH and 50 C 60% RH to 70 C 60% RH 38 h Little change in ILLS and shear strength Type I and V cement with 3 different commercial Kuwatii epoxies 81b Open air, room temp, 60 C, 80 C 18 months No apparent difference in strengths with time or temperature Table 2 5. Continued.
58 Materials Exposure conditions Exposure Time Property Changes Type I and V cement with 3 different commercial Kuwatii epoxies 81a Open air, room temp, 60 C, 80 C 18 months No change in tensile strength, 25% decrease in bond strength after 18 months. Unidirectional glass / epoxy 82 Sea water at 23 and 65 C 1 year No change in tensile properties SEH 51/Tyfo GFRP with Tyfo S Epoxy 83 Salt fog cycles with variation from 20% to 90% RH and UV cycles. GFRP jacketed cylinders exposed to 50 g NaCl / l at 18C and salt fog cycles 10,000 h Little degradation under various conditions IM7 carbon fiber 997 epoxy 85 UV at 60C and 50C 100%RH with cyclic exposure 1 000 h Slight increase in modulus for low amplitude fatigue samples Uniaxial and flexural strength decreased ~25% with UV and condensation SEH 51/Tyfo GFRP with Tyfo S Epoxy 84b Tide Simulation (5%NaCl for 3 days then 4 days dry), salt fog cycles (8h salt fog at 35C 98% RH 16h dry), immersion 10,000 h Salt water immersion improved properties, however plasticization of interface. SEH 51/Tyfo GFRP with Tyfo S Epoxy 84a 50g/l salinity at 35C and 10 C for 12 h then 10 C for 12 h 10,000 h Increase in load capacity with salt fog and immersion cycles reduced capacity with freeze thaw cycles GFRP, reduced pullout all GFRP, CFRP reduced capacity and pull out under all conditions C ommercial epoxy and mortar with Eglass or carbon fibers 7b 15.5 C for 24 h, 20 C for 24 h 60 days Energy decrease in sea water and water. Increase in energies at 15C and under freeze thaw conditions IM7/997 carbon fiber epoxy composite 71 UV at 60 C and condensation at 50 C with cyclic exposure 1 000 h Very small variations in elastic modulii a nd poissons ratio. 29% reduction of transverse tensile strength. No effect on longitudonal properties Carbon or glass fiber/epoxy prepreg 73 22, 45, and 60 C in air, water/salt wet/dry cycles, and UV exposure 2,000 h No effect on longitudinal modulus for all conditions. Strength decreased only for 0/90 composites in water. Graphite/epoxy 74 Longitudonal direction strength remains constant, but in transverse direction properties are significantly reduced Table 2 5. Continued.
59 Degradation Mechanisms of FRPs Several degradation mechanisms for FRPs are discussed in the literature, with water absorpton being the most common. A recent paper by Benzarti et al. examined four different epoxy CFRPs with concrete exposed at 40 C and 95% relative humidity and found a decrease in pull off strength and a drop in the Tg over time 75 Pull off tests indicated a ch ange in the failure mode from a substrate failure to cohesive failure within the polymer joint. These changes with accelerated aging supported their claim that water absorption was the mechanism for degradation. Karbhari et al. also found the same change in failure mode and claimed plasticization of the matrix, but gave more insight. After the water has penetrated into the resin through capillary flow through microcracks and voids, debonding stresses across the fiber res in interface can occur due to resin swelling and osmotic pressure 7b While water exposure typically causes a degradation in properties, Silva and Biscacia reported that exposure of their GFRP concrete system to salt water caused improvement in the properti es of the concrete, but a higher plasticization of the interface 84b After 10,000 hours of exposure to environmental conditions, the interface was strongly degraded, with minor slippage between the GFRP and concrete. The same water absorption mechanism i s also cited for FRP systems without concrete. Khan et al. found a reduction in ILSS and considered microcracking as a possible mechanism for degradation, but found no change in moisture content after reaching saturation. As a result, they concluded plasti cization was responsible for the loss in properties 86 Lee and Peppas examined the diffusive and mechanical behavior of TGDDM and DGEBA resin based epoxy / carbon composites and found a weakening of the interfacial bond strength could be caused by hygrothe rmally induced interfacial
60 stresses 69 Ray explained his variation and lack of degradation of the ILSS of his GFRP composites as a result of water absorption and reducing internal stresses 70 In addition to water absorption Zhang et al. reported resin sw elling during exposure of their CFRP to 70 C and 100% relative humidity, along with a change in failure mode from brittle fracture to yielding and ductile failure 65 Specimens immersed in boiling water examined under SEM showed that bare fibers and voids i n the matrix were evident, implying the boiling water leached the resin out and destroyed the interface. They suggested that boiling water was not an appropriate accelerated aging process. Mourad et al. claimed a dual mechanism of stress relaxation swelli ng mechanical adhesion and breakdown of chemical bonds between the fibers and matrix at the interface, due to seawater exposure of their glass/epoxy composites 82 They also found no change in the modulus and an increase in the ductility which they explaine d by plasticization and improved mechanical adhesion between fiber and matrix which was explained by hydrolysis causing swelling of the resin. While SEM images were taken, no chemical characterization of the composite was performed, so the exact degradatio n of certain groups was not evident. Tuakta and Buyukozturk acknowledged that reversible degradation of their composite would undergo plasticization and swelling while irreversible degradation would involve hydrolysis or microcracking 80b They found irreve rsible degradation of their epoxy CFRP concrete composite under cyclic moisture and temperature cycles, and as a result claimed hydrolysis of their system. As with the previous study, they did not include chemical characterization to verify their suggested mechanism, nor did they include SEM images.
61 Park et al. suggest the three dominant deterioration mechanisms at the matrix fiber interface were osmotic cracking, interfacial debonding, and delamination, which caused micro cracks and pore development as a r esult of artificial aging in simulated concrete environments 87 The decrease in ILSS was accompanied by an increase in pore volume and failure typically at the fiber, rather than matrix. However, they used a commercial vinyl ester resin and found the resin did not undergo significant chemical reactions with the hydroxyl ions. Au and Buyukozturk found that after exposure of CFRP epoxy concrete composites to 100% RH the mode of failure changed from delamination to interface separation 80a They suggested an in terfacial toughening and weakening mechanism as a result of water absorption by the epoxy, and admitted the validity of this claim requires further testing. UV studies on epoxy GFRPs by Lee and Lee indicated that degradation developed from the formation of new polar groups on the surface and penetration of the inner layer by reoriented oxygen groups 88 XPS confirmed the oxidation of the polymer and the formation of carbonyl groups. Exposure to UV and moisture on a commercial CFRP by Kumar et al. resulted in continuous weight loss and various morphological and chemical changes 71 Exposure of UV radiation alone caused the formation of microcracks in the epoxy due to increased crosslinking from photo oxidation reactions. No morphological changes were evident fr om moisture alone. However the combined exposure resulted in sever degradation in the form of matrix erosion, void formation, and fiber matrix debonding. The authors suggested a mechanism to explain this observation. Exposure to UV resulted in the creation of a thin layer of chemically
62 modified epoxy, and the following treatment with moisture leached away soluble degradation products which exposed a fresh layer that could once again be attacked by UV radiation. In addition, absorbed moisture in the epoxy ma trix could enhance the photo oxidation reactions due to the increased availability of H+ and OH ions. FTIR data indicated reductions in the peak attributed to C N stretching vibrations, demonstrating the presence of chain scissio n reactions as shown in Fi gure 2 15 The synergistic manner in which UV radiation and moisture attacked the composite led to the most degradation and widest variety of mechanisms in all the literature. Figure 2 15 FTIR spectra of IM7/997 specimens after 500 hours of UV or conde nsation 71 In summary, as with the neat epoxy degradation mechanisms, plasticization is the most commonly cited mechanism although there is no evidence of reversibility and thus the claims of plasticization need further study However, hygrothermally induced stresses and resin swelling are also mentioned as mechanisms for the FRP composites. Table 2 6 gives a brief overview of the mechanisms along with exposure conditions and exposure times for the various FRP systems.
63 Table 2 6. Summary of mechanism s of degradation for accelerated testing of FRP systems Materials Exposure conditions Exposure Time Mechanism CFRP with DGEBA and amine hardeners 75 40C 95% RH 20 months Plasticization Commercial epoxy and mortar with Eglass or carbon fibers 7b o 15.5 C for 24 hours, 20 C for 24 hours 60 days Plasticization of resin and debonding stresses due to resin swelling and osmotic pressure Epoxy prepreg CFRP 86 70C 85%RH 60 days Plasticization TGDDM/DDS DGEBA/DDA and CFRP 69 Water at 50,70,90, and 100 C 1200 h Plasticization and H ygrothermally induced stresses DGEBA based resin and polyester resin with E glass fibers 70 50 C 60%RH to 5 0 C 95%RH and 50 C 60% RH to 70 C 60% RH 38 h Plasticization and reducing internal stresses SEH 51/Tyfo GFRP with Tyfo S Epoxy 84b DI water at 40C with 12h at 20%RH then 12 h at 90% RH, 8 h sa lt fog at 35C 98% RH then 16 h drying, DI water at 22 C 10,000 h Plasticization of interface Vicotex 1452 with amide curing agent and T300 CFRP 65 DI water at 100 C and 100% RH at 70 C 2 days immersion, 1,000 h for RH Resin swelling and plasticization. Boiling water attacked fiber matrix adhesion and produced voids. Unidirectional glass / epoxy 82 Sea water at 23 and 65 C 1 year Stress relaxation, resin swelling, hydrolysis and plasticization Commercial epoxy and concrete 80b 23 C and 50 C in DI water and cyclic exposure 8 weeks Hydrolysis or microcracking Commercial vinyl ester and GFRP 87 Immersion in alkali solution 20 C 120 days Matrix osmotic cracking, interfacial debonding delamination CFRP with amine cured epoxy 80a 23 and 50 C with 50% and 100% RH 8 weeks Interfacial toughening and interface weakening Bisphenol A type epoxy and glass fiber 88 UV 50 h Oxidation and formation of carboxyl groups IM7/997 carbon fiber epoxy composite 71 Cyclic exposure of UV at 60 C and condensation at 50 C 1,000 h UV creates thin layer of soluble degradation products which are leached away by condensation
64 Summary of Accelerated Testing of Epoxy and FRP Systems Under typical accelerated conditions a loss of mechanical properties of epoxy systems is usually observed. From room temperature to 90 o C in water, the most common ly claimed mechanism of degradation is plasticization which causes an increase in polymer chain mobility and a change in the Tg. Exposure to other liquids, such as saltwater or alkali solutions, results in similar changes E xposure to UV to 150 C results in a reduction in properties attributable to oxidation. A combination of water and UV radiation at increased relative humidity and/or temperatures causes degradation by hydrolysis. FRP systems also undergo a loss of properties under accelerated agi ng conditions, typically due to the degradation of the epoxy matrix. As with neat epoxy systems, plasticization is claimed to occur with exposure to water from room temperature to 60 C. There is a loss in shear, tensile, and flexural strength and a decreas e in fracture toughness. However, some studies have found no change in tensile properties with exposure to sea water, DI water, or air at the same temperatures. These studies explain this unexpected result by a stress relaxation swelling mechanical adhesio n mechanism. In addition, under various conditions of UV, water or air, from room temperature to 45 C, no change in longitudinal modulus has been reported. This makes sense because of the orientation of the fibers in the FRP system. Combination of UV and w ater from 50 to 60 C results in a combination of oxidation and hydrolysis of the FRP and a reduction in the tensile strength. Table 2 7 provides an overall summary of the findings from the literature. Since the conditions for accelerated aging testing som etimes overlap in the literature it is difficult to attribute mechanisms precisely to specific testing conditions. However, the
65 overall trends are clear: exposure to water and temperatures higher than room temperature result in a decrease in mechanical pro perties. When exposure temperatures are close to or exceed the glass transition temperature a greater reduction in properties has been reported so note must be taken of the Tg of the polymer, and rigorous control of the testing conditions must be observe d before testing begins. Table 2 7. Summary of property changes and mechanisms for epoxy and FRP systems Material Exposure Conditions Exposure Time Property Change Mechanism Epoxy water/air 40 90C up to 2 years decrease in Tg, strength, and modulus Water absorption 60 100C/UV at 42C up to 300 days decrease in contact angle, surface potential and resistivity Oxidation UV at 60C/UV at 60C and water up to 35 days decrease in flexural strength Hydrolysis FRP water from 40 100C up to 20 months decrease in Tg, strength, toughness Water absorption UV at 60C/Condensation at 50C 50 hours decrease in flexural strength Oxidation UV at 60C/Condensation at 50C 1,000 hours decrease in tensile strength Hydrolysis 23 to 80C with and without sea water up to 18 months no significant change in tensile or shear strength 22 60C in air, water/salt wet/dry cycles, and UV at 60C up to 2,000 hours no significant change in longitudinal properties While the literature has addressed the changes in properties for epoxy and FRP concrete systems with environmental exposure, there is still much work to be performed regarding chemical interactions between epoxy and concrete. There are numerous theories regarding why epoxy performs so well as a repair ma terial, but the fundamental epoxy concrete bond has still not fully been explored. This dissertation will provide insight into the issues of this important interaction considering both civil and engineering standpoints.
66 CHAPTER 3 NATURE OF THE CONCRE TE EP OXY INTERFACE Motivation Epoxies are an important class of structural adhesives. An adhesive is a substance used to join the surfaces of two solid surfaces, referred to as adherends. Intimate molecular contact between the adhesive and adherend is an impo rtant factor, but is not the only criteria for effective bonding. The forces across the interface must also be sufficiently strong to hold them together under an applied load. While epoxy systems are frequently employed as adhesives in the engineering fi eld, little is known regarding their interfacial bonding with concrete at the nanoscale. For the general description of adhesion between various types of materials, several different mechanisms have been proposed: mechanical interlocking, diffusion, elect rostatic, and adsorption theory 89 From the civil engineering standpoint, mechanical interlocking is a sufficient theory to explain adhesion between FRP materials and concrete. Numerous research has shown the effect of surface preparation on the mode of failure 90 fracture energy 7b or bond strength 91 However, from a chemistry standpoint this theory is inadequate in describing intermolecular forces and atomic interactions that explain physical properties and mechanisms of degradation. Theories of Adhesion Mechanical interlocking involves the flow of the adhesive into the pores, holes, and crevices of the surface of the substrate, and after curing, locks mechanically to the surface. Numerous articles by other authors on concrete surfaces have shown that a rough surface preparation leads to higher bond strength 7b, c, 92 However, these increases can also be attributed to other factors such as efficient removal of weak
67 surface layers, improved interfacial contact, and the enhancement of energy dissipative mechanisms in the adhesive 93 Wake et al. suggested the effects of mechanical interlocking and surface force components could be multiplied to give a result for the measured joint strength 94 Diffusion th eory suggests that when an adhesive contains an adherent solvent, the adhesive can diffuse into the substrate with an interchange of molecules. This theory requires that both the adhesive and adherend are polymers and has found limited application when th e polymer is below its Tg. According to the electrostatic theory of adhesion, upon contact of adhesive to substrate, electrons are transferred across the interface to balance their respective Fermi levels. This will result in the formation of a double la yer of electrical charge at the interface. This theory is regarded as a dominant factor in biological and particle adhesion, but for an insulating polymer like epoxy, will not play a large role. Adsorption is the adhesion of atoms or molecules of gas, liq uids, or dissolved solids to a surface. It differs from absorption, which only deals with liquids. The exact nature of the bonding depends on the interacting species, but the process is generally classified as physisorption or chemisorption. Physisorpti on is caused by weak intermolecular forces and does not involve a significant change in the electron orbitals of the species involved. Chemisorption is due to valence forces in which new types of ionic or covalent bonds are formed. Typical binding energi es for physisorption range from 10 100 meV, while chemisorption ranges from 1 10eV. Hydrogen Bonding The hydrogen bond is a type of dipole dipole interaction in which a hydrogen atom of one molecule is attracted to an electronegative atom of the same
68 (int ramolecular) or different (intermolecular) molecule. For the classical view of hydrogen bonding, we can turn to the directional interaction of water molecules. The large difference in electronegativity between the H and O atoms makes the O H bonds of the water molecule characteristically polar, with the H atom having a partial atomic charge of +0.4 and 0.8 for the O atom. Neighboring water molecules orient themselves so that the local dipoles O H + point at O as shown in Figure 3 1. The classical v iew can be extended to analogous interactions of X H A formed by strong polar groups X H + interacting with A where X or A may be O,N, or F 95 Figure 3 1. Hydrogen bonding between water molecules The hydrogen bond really is a complex interaction m ade up electrostatic, van der Waals, and covalent bonding interactions. The electrostatic contribution is directional and the van der Waals is isotropic. The strength of the hydrogen bond lies between that of van der Waals and covalent bonds and may vary depending on the specific interactions, from 0.2 to 40 kcal per mole. IR and NMR spectroscopy have both become standard methods to investigate hydrogen bonds in the solid state, however, if there are many symmetry independent bonded groups, band overlap normally prevents detailed analysis. Although not much is known about the epoxy cement interface, prior work on the metal epoxy interface provides some insight 8b When a metal surface is exposed to
69 water, the oxide surface of the metal forms hydroxyl groups which are available to the epoxy for hydrogen bonding. The strength of the bond depends on the concentration of hydroxyl groups on the surface. Pretreatment of the metal surface can be used to r emove oxide layers and change the hydroxyl concentration, or to create a rougher surface which also improves bonding. Three mechanisms are proposed as to why water has a detrimental effect on this bond: one, water displaces epoxy, which forms hydrogen bo nds with the oxide layer; two, a hydration layer forms which acts as a weak boundary layer, resulting in a decrease in adhesion strength; three, the metal corrodes, leading to delamination of the epoxy. Clearly only the first mechanism is applicable to the ceramic polymer interface when epoxy is bonded to concrete. Thus, a strong hypothesis is that epoxy bonds to cement via hydrogen bonding, and that water can displace these hydrogen bonds leading to a weakening of the adhesion strength. The objective of this chapter is to determine if hydrogen bonding interactions are responsible for the strong adhesion between concrete and epoxy. Small molecule epoxy analogs will be adsorbed to cement paste particles to isolate specific functional groups that may int eract. ATR FTIR will be used to examine the very surface of the molecules and solid state 1 H, 13 C, 29 Si MAS NMR will be used to investigate bulk properties of the composites. Experimental Procedure S mall molecule analogs of the larger crosslinked epoxy sy stem were used in order to isolate specific bonds that may form between cement and different components of the epoxy network. The chose n epoxy analogs are: Bisphenol A (BPA, 99% purity), n ethyldiethanolamine (EDA, 98% purity) a nd glycidyl phenyl ether ( GPE, 99% purity).
70 These small molecules were purchased from Sigma Aldrich and used without further purification. The chemical structures of these molecules are shown in Figure 3 2 Figure 3 2. Chemical structures of Bisphenol A, gylcidyl phenyl eth er, and n ethyldiethanolamine Composite s amples were prepared by curing cement paste (without sand or other aggregates) with a water to cement ratio of 0.3 7 5 for 3 weeks followed by 1 week immersion in lime solution Then the cured cement paste was dried, ground, and f iltered to a particle size of 38 m or smaller. Next various mass loadings of the epoxy analogs were dissolved in 2mL of acetone and stirred for 15 minutes. The solution was then poured into a flask with 0.5 g of the cement pasted particle s, mixed for 15 minutes, and dried in an oven overnight at 60 C to evaporate the solvent. FTIR Theory and Technique Fourier Transform Infrared (FTIR) spectroscopy is a useful tool for polymer characterization. In the transmission technique, IR radiation is guided through an interferometer, that splits the beams in two, where one beam travels a different optical distanc e. The beams are recombined and passes through the sample, which then travels to a detector. The measurement acquired is made in the time domain, which is then Fourier transformed to the frequency domain to give a spectrum. Although the sample is irradi ated with a whole range of IR frequencies, absorption only occurs at specific frequencies that match the vibrational frequencies of the molecule.
71 Another IR sampling technique is Attenuated Total Reflectance (ATR). Instead of the IR beam passing throug h the sample, the beam is passed through an ATR crystal that is in contact with the sample. The beam forms an evanescent wave that extends into the sample, only when the crystal is made of an optical material with a higher refractive index than the sample The intensity of the evanescent wave decays exponentially with distance from the interface. The penetration depth (dp) into the sample is typically 0.5 to 2 m, or the near surface, but depends on many factors as shown in equation 3 1. (3 1) Where is the wavelength, n IRE is the refractive index of the crystal, and n S is the refractive index of the sample. An additional important experimental consideration is the contact pressure between the sample and crystal. Diffuse Reflectance Infrared Fourier Transform (DRIFT) spectroscopy collects the scattered radiation from the bulk of powder samples, typically in a KBr matrix. The limitation of this technique is that there is no linear relation between the band intensit y and concentration, so quantitative analysis is very complicated. In this section of work, the ATR technique was used to quantitatively analyze the bonding between the epoxy analogs and concrete. FTIR samples were analyzed using a hemispherical ZnSe crystal with a multi angle attach ment on a Magna IR E.S.P. System 760 Spectrometer with 128 sample and reference scans at 4cm 1 resolution from 600 to 4000 cm 1 The spectra were ATR corre cted using OMNIC 6.1 software. Solid powder samples were placed i n firm contact with the ZnSe crystal by placing the
72 powder on a metal plate which was then raised by turning a screw until the powder was touching the crystal and further pressure could not be applied by hand. Liquid samples were pipetted onto the crystal and no further pressure was required. NMR Theory and Technique Nuclear Magnetic Resonance (NMR) spectroscopy is a nondestructive technique and was first successfully used on cementitious materials in the early 1980s 96 It is based on analysis of the interaction between an oscillating radio frequency with a collection of atomic nuclei in the presence of a strong magnetic field. In NMR, experiments are not performed on the electrons of atoms, but on the nuc lei. Nuclei with an odd numbered mass, such as 1 H and 13 C, have a spin. The number of orientations for the specific nucleus is determined from quantum mechanics where a nucleus of spin l has 2 l + 1 orientations. In the case of 1 H and 13 C, when an extern al magnetic field is applied, the two energy levels split. The initial populations of the energy levels are described by the Boltzman distribution, where the fractional number of particles Ni/N, occupying a set of states i, possessing energy E i is (3 2) K B is the Boltzman constant, T is temperature, g i is the number of levels having energy E i and Z(T) is the partition function. The lower energy levels will contain slightly more nuclei than the higher level as a result. T he frequency of radiation required to excite these nuclei into the higher level is determined by the difference in energy between the energy levels. This precise resonant frequency is dependent on the magnetic field experienced by the nucleus, and can be affected by electron shielding, which is dependent on the chemical environment. Typically the proton reference frequency is
73 measured relative to the frequency of tetramethylsilane (TMS) for ease in comparing data. The chemical shift is defined as nuclear shielding / applied magnetic field and is expressed as 0 0 0 The location of the chemical shift reveals fundamental information regarding the local structure of materials. The numerical value of the chemical shift is so s mall that it is common to multiply the value by 10 6 and then express its value in parts per million (ppm). While NMR can yield useful data for liquid samples, it is more challenging to perform on solids. A broadening of the resonances of NMR spectra for solid state samples occurs because most atoms can only undergo restricted vibrations and rotations at room temperature, each in a slightly different electronic environment, which can make interpretation almost impossible. Magic Angle Spinning (MAS) was f irst used successfully in 1958 by Professor Edward Andrew, which allowed him to obtain high resolution spectra of solid state samples. In MAS, the sample is spun at a high frequency (between 5 and 60 kHz for commercial probes) at the magic angle of 54.74 with respect to the direction of the magnetic field. Dipolar and quadrupolar interactions along with chemical shift anisotropy can be averaged out using MAS to yield spectra with much sharper signals 97 In one dimensional NMR, the signal is recorded a s a function of one time variable, and then Fourier transformed to produce a spectrum that is a function of one frequency variable. In two dimensional NMR, the signal is recorded as a function of two time variables, t 1 and t 2 and then Fourier transformed twice to give a spectrum that is a function of two frequency variables.
74 For a 1D experiment, right after the pulse sequence, data is acquired. A 2D experiment involves a few more steps. In the first part of this experiment, called the preparation time, the sample is excited by 1 or more pulses. The group of pulses may be purely radio frequency, or may include magnetic gradient pulses. When a pulse is applied, the magnetic moment of the nuclei rotates away from alignment of the external magnetic field, along the +z direction, down towards the x y plane. This resulting magnetization is allowed to evolve for the first time period, t 1 Then another pulse or series of pulses is applied, called the mixing time, in which magnetization is transferred from th e first nucleus to second one. Finally, the signal is recorded as a function of the second time variable, t 2 1D NMR experiments were conducted at two locations : College de France in Paris, France and Sandia National Labs in Albuquerque, New Mexico. The instrument in Paris recorded 1 H spectra using an Advance III Bruker s pectrometer operating at at Magic angle spinning (MAS) was applied with a spin speed of 60 khz with a 5s repetition delay 13 C NMR experiments were performed with an Advance III 300 Bruker spectrometer operating at MHz. Proton to carbon cross polarization magic angle spinning (CP MAS) sequence was applied with pulse lengt hs of 5 s, 1ms contact time and a recycle delay of 10 to 20s. For 29 Si experiments samples were packed into a 7mm rotor and relaxation delay of 60s was used The other instrument recorded 1 H NMR spectra using an Advance III 600 Bruker spectrometer opera 1 H) using a 2.5 mm rotor Magic angle spinning (MAS) was applied with pulse lengths of 2.5 s and a recycle delay of 5s with a spin speed of 30 k H z. Spectral deconvolutions were performed using DMFIT software.
75 2D homonuclear 1 H NMR experiments were conducted using the instrument in Paris. Spin diffusion 98 (2D NOESY) and double quantum 99 (BABA or back to back) experiments were performed to determine the specific interaction/proximities between functional groups. The Nuclear Overhaus er Effect (NOE) is a phenomena in which transfer of nuclear spin polarization from one nuclear spin population to another occurs through cross polarization. NOESY experiments make use of NOE to establish correlations of nuclei that are spatially close. D ouble quantum experiments make use of strongly dipolar coupled networks instead of fighting the dipolar interactions 100 Interpretation of data for both types of experiments is similar. For diffusion experiments, both the x and y axis correspond to the chemical shift and the intensity of the peaks are presented in a third dimension, such as contour lines or different colors. In this plot, diagonal peaks have the same coordinate, while cross peaks have a different value for each coordinate and are found on the off diagonal. Diagonal peaks correspond to the peaks in the 1D experiment, while cross peaks indicate coupling between pairs of nuclei. In the case of double quantum experiments, the chemical shift is doubled in the second dimension. Auto correla tion appears at 2* i while correlations between signals at 1 and 2 ppm respectively show a cross peak at 1+ 2 Results and Discussion FTIR Figure 3 3 shows an FTIR ATR spectrum of cured cement paste at an incident angle of 42 This spectra has been run through the ATR correction algorithm to compensate for variation in sampling depth. The bands at 874 and 1420 cm 1 corresponds to C O from CaCO 3 the peak at 96 0 cm 1 corresponds to SiO 4 in calcium
76 silicate hydrate, the peak at 1654 cm 1 corresponds t o OH in water molecules, and the broad peak from 3000 to 3700 cm 1 is the hydroxyl region. For the neat, cured cement paste the hydroxyl band was deconvoluted into 2 peaks, one at 3 642 cm 1 corresponding to Ca(OH) 2 and one at 3 410 cm 1 corresponding to ca licum silicate hydrate. These assignments are in good agreement with the literature 101 A summary of these characteristic IR absorption bands is shown in Table 3 1. Figure 3 3. ATR IR spectrum of cured cement paste Table 3 1. Assignment of observed peaks for cured cement paste 101a, 102 Peak Position IR Assignment (Wavenumber, cm 1 ) 874 Asymmetric stretching vibration of CO from CaCO 3 960 SiO 4 from CSH 1110 Stretching vibration of S O in [SO 4 ] 2 1420 Asymmetric stretching vibration of CO from CaCO 3 1654 Bending mode of OH from H 2 O 3413 Symmetric and asymmetric stretching vibration of OH from CSH 3643 OH in Ca(OH) 2
77 ATR spectra of the small molecules at room temperature are shown in Figure 3 4 where GPE and EDA are liquids, and BPA is a solid powder. Both the BPA and EDA show a broad, asymmetric band attributed to the v O H of self hydrogen bonded hydroxyl groups. Th e width of this band is related to the different lengths of formed hydrogen bonds which adsorb at slightly different frequencies. For BPA, the peak of this band is located at 3315 cm 1 and for EDA it is located at 3320 cm 1 Figure 3 4. ATR spectra of the neat small molecule epoxy analogs and cement paste Figure 3 5 shows the hydroxyl region of ATR spectra taken at 45 after applying the small molecules to the cement paste particles. All the samples have been corrected, using OMNIC software, for CO 2 a nd H 2 O, and then ATR corrected to account for the variation in penetration depth with wavelength and incident angle. While there is much noise due to residual water vapor in the chamber, major shifts in the OH band may still be observed. Cement paste bef ore treatment had its broad band located at 3398 cm 1 and after treatment with GPE the position did not change. After treatment with BPA and EDA a shift to lower wavenumbers for the broad peak took place. The BPA sample shifted to 3321 cm 1 and the EDA sample shifted to 3359 cm 1 Most likely
78 this is simply spectra overlaying of the hydroxyl groups of the small molecules with the OH region of the cement paste. IR work on BP A by other authors has demonstrated an increase in the v O H band after hydrogen bonding with another molecule 103 These results are counterintuitive since the hypothesis envisioned the cement paste as being able to hydrogen bond with the various small molecules containing hydroxyls. Figure 3 5. ATR spectra of the hydroxyl region of cement samples after treatment; a) cement paste, b) cement paste + GPE, c) cement paste + BPA, d) cement paste + EDA Additional loadings of BPA on the cement paste were performed to determine if a gradual shift in the v O H band could be observed. Howev er, increasing the amount of BPA did not yield any increase in the wavenumber of the overall peak shown in Figure 3 6. In addition, no overall change in the shape of the band was observed. Using the DRIFT IR technique also gave the same results.
79 Figure 3 6. ATR IR spectra of BPA treated samples. a) neat cement paste, b) cement paste + 30% BPA, c) cement paste + 20% BPA, d) cement paste + 10% BPA, and e) cement paste + 7% BPA Since additional hydroxyl groups were contributed by the small molecules, simple observation of the IR spectra does not yield enough information about the nature of the interaction because of the overlapping bands. Peak subtraction was performed to investigate if the data contained more information than initially assumed. This process has been proven in other systems to show interactions with water and different states of hydrogen bonding 55, 104 First, samples were ATR corrected using 1.5 as the index of refraction of the sample and 2.4 for the ZnSe crystal. Next, the composite sample spectrum had the small molecule contribution subtracted in OMNIC software with the formula I=A k*B where A is the composite sample, B is the neat small molecule, and k is a constant. The constant was determined by nor malizing to the V CH 3 stretch located from 2,975 to 2,965 cm 1 Unfortunately, peak subtraction yielded no new results, as shown in Figure 3 7. The result of the peak subtraction were spectra similar to the neat cement paste with no significant change in the location of the V O H band.
80 Figure 3 7. Subtracted ATR spectra of BPA treated sample s. a) neat cement paste, b) cement paste + 30% BPA, c) cement paste + 20% BPA, and d) cement paste + 10% BPA NMR Before probing epoxy cement interactions using solid state NMR, epoxy silica nanoparticles and epoxy laponite systems were investigated. These two systems were examined because they have a more well defined structure than ceme nt. Silica nanoparticles ( Aerosil 380) were 7nm in diameter. Laponite exists as platelets made up of three sheets containing a middle sheet of magnesium ions in an octahederal coordination with oxygen atoms and hydroxyl groups and two outer sheets composed of tetra hedral silica sh eets 105 The platelets have an average diameter of 25 nm and a thickness of 0.92 nm, with these finite dimensions being responsible for the occurrence of silanol groups on the edges. During the manufacturing process, some magnesium ions are substituted w ith lithium ions resulting in a negative charge which is balanced by interlayer sodium cations. This synthetic clay has a negative face charge and a rim charge that may be negative or positive depending on the pH 106
81 structure has be en reported by the manufacture r as Na +0.7 [Si 8 Mg 5.5 Li 0.3 ]O 20 (OH) 4 ] 0.7 Compared to other clay s laponite has a relatively high edge to surface r atio of 0.07 107 Liquid NMR on the small molecules was performed to identify the resonances of the functional groups before interaction with the two nanoparticle systems. For these experiments the small molecules were dissolved in 13 C enriched acetone. The same procedure for the adsorption of the small molecules to the cement powder was followed for the two types of composites with a 10mM/g concentration. Initial solid state 1 H and 13 C experiments indicated no apparent interactions between the silica nanoparticles and sm all molecules, so the more complex laponite system was investigated more thoroughly. Figure 3 8 shows the 13 C NMR spectra of the EDA composites and Figure 3 12 shows the 13 C NMR spectra of BPA and silica and laponite nanoparticles. In all the 13 C NMR spe ctra the resonance at 30 ppm is due to unevaporated acetone. Figure 3 8. 13 C NMR spectra of EDA composites
82 Solid state 1 H NMR spectra of the pure laponite samples showed 2 resonances at 0.5 and 4 ppm, assigned to silanol groups and interlayer water, re spectively. Figure 3 9 shows the 1 H spectra of laponite, liquid BPA, and laponite BPA composite. For the liquid BPA the resonance at 1.5 ppm (peak 1) was assigned to the methyl group, the 6.6 and 7 ppm signals (peak 3 and 4) to aromatic hydrogens, and th e resonance at 5 ppm to the hydroxyl groups. The solid BPA resonances were located in slightly different chemical shift positions. Signals at 0.7ppm, from 6.6 to 7.1 ppm and at 8.5 ppm were assigned to the methyl groups, the aromatic hydrogens and the hy droxyl groups re s pectively. The small shoulder at 4.8 ppm may be due to non hydrogen bonded hydroxyl groups, as BPA has been shown to have, but has only been demonstrated using FTIR 108 While there may be some slight changes in the chemical shift peaks b etween BPA and the laponite composite, no strong change in the hydroxyl group chemical shift was observed. Figure 3 9. 1 H MAS NMR spectra of BPA laponite composites
83 Although the EDA small molecule also has hydroxyl groups, the composite did not exhi bit a downfield shift in its 1 H NMR spectra as seen in Figure 3 10. The GPE composite did not show any significant changes in the 1 H NMR spectra either, but that was expected since the small molecule did not have any groups that could contribute to hydrog en bonding with the cement. Figure 3 11 shows the 1 H NMR spectra of the GPE composites. Figure 3 10. 1 H NMR spectra of EDA laponite composites Figure 3 11. 1 H NMR spectra of GPE laponite composites
84 Since laponite has no carbon atoms in its chemical makeup, it gave no peaks in 13 C NMR spectra, so interpretation was straightforward. Figure 3 12 shows the 13 C spectra of the various BPA composites. For identification, the various carbon peaks for the solid BPA have been indicated on the spectra corresponding to the molecule. Figure 3 12. 13 C MAS NMR spectra of BPA composites Wu et al investigated blends of phenolic resins and poly (hydroxyl ether) of bisphenol A and found as the quantity of phenoxy increased inter molecular hydrogen bonds increased, as indicated by a downfield shift of the OH substituted carbon in the phenoxy at 66 pm 109 For the bisphenol A composites investigated in this work, the
85 carbon resonance that would correspond to a group that could hydro gen bond would be peak 1 at 154 ppm, corresponding to the carbon of the phenol group. This downfield shift was not observed for either the aerosol or laponite composites. In fact, no significant changes in the spectra for any of the laponite small molecu le systems was observed in 13 C spectra. Although concrete does not have as well defined a chemical structure as laponite, it yielded more interesting NMR results. Previous work on cement has focused mainly on calcium silicate hydrate, CSH, since it is th e main hydration product of cement and is the major phase responsible for the mechanical properties of concrete. The CSH structure of hydrated cement is still not fully resolved because it is nearly an amorphous material. However a large body of work has been performed on synthetic CSH systems where the CaO to SiO 2 ratio has been varied, usually between 0.7 and 2.0. These model systems closely resemble the minerals tobermorite and jennite. The CSH structure is made up of distorted Ca O sheets with silic ate tetrahedral chains on each side. These layers stack along the  direction and calcium and water molecules are contained in the interlayer space. In jennite, half of the oxygens from the Ca O sheet are shared with OH groups, where in tobermorite all of the oxygens are shared with the silicate chains. Figure 3 13 shows an image depicting the structure of jennite according to Churakov 110
86 Figure 3 13. The structure of jennite. Ca and Si sites are shown as polyhedra, interlayer Ca sites are shown as large gray spheres, and water molecules are represented by spheres with ellipsoids. 110 The role of water in hydrated cement paste is very complex and controversial. Besides water vapor held in the pores, it has been suggested that water can exist in c apillary water, adsorbed water, and interlayer water 111 Capillary water is water molecules held in voids larger than 5 nm, while adsorbed water is held by hydrogen bonds and is physically bound. Chemically bound water is located in between the calcium s ilicate lamellae and is directly incorporated into the structure of the cement hydration products 112 More specifically, work performed by Cong et al. on crystalline calcium silicate hydrates confirmed the presence of hydrogen bonding groups Si OH and Ca OH using 29 Si and 17 O MAS NMR 113 A more recent study has demonstrated a
87 covalently bonded polymer CSH meso composite using alkyoxysilanes that interact with the silicate chains 19 as evidenced by a new T 1 resonance in the 29 Si spectra. This is an importan t distinction, because previous theories suggested polymers became bound in the interlayer space. To confirm the formation of covalent bonds between the polymer and CSH, 29Si CP MAS NMR experiments were performed on composites prepared in D 2 O. For the ne at CSH sample, no resonances were able to be distinguished, while for the composites, a resonance in the Q 1 region was found, indicating polarization transfer from protons to inorganic Q 1 silicates. Since all of the protons are located in the polymer, it was concluded that the polymer chains are spatially close to silicate tetrahedral and that the silanes are incorporated into CSH silicate chains. The current study examines the interaction of small molecules with hydrated cement paste. As with the lapon ite studies, the neat cement paste gave no carbon signals without direct excitation, so only 1 H MAS solid state experiments were performed. The 1 H spectrum of hydrated cement paste gave 2 broad resonances assigned to the hydroxyls of Ca(OH) 2 at 1.2 ppm and the interlayer water of calcium silicate hydrate at 5.4 ppm. These resonance assignments are in good agreement with work by other authors 114 Figure 3 14 shows the 1 H MAS NMR spectra for the cement paste and its various small molecule composites. N o narrow signal from any of the organic components is found, indicating the small molecules have been immobilized in the inorganic matrix. Results by other authors on white cement and phenol suggests that d 5 phenol exists in Portland cement in two forms, unbound and bound to the cement matrix 115 The unbound environment is assumed to be liquid like and exist in cement pore water in ionized form. The bound environment demonstrates reduced
88 motion of the organic inside the cement, and implies binding of the phenol through its hydroxyl groups. Figure 3 14. 1 H Solid State MAS NMR spectra of cement paste composites As with the FTIR data, no significant differences in the spectra, compared to the neat molecules, could be observed. The only exception was th e BPA treated sample which shows a small shoulder from 8 to 10 ppm. For the sample treated with BPA the methyl resonances were located at 2.4 ppm, the phenyl ring resonances were located at
89 7.1 ppm, and the hydroxyl resonances were located at 9.1 ppm. Al though the hydroxyl resonance is small, it is a downfield shift from the solid bisphenol A OH resonance at 8.7 ppm. To determine if this small shoulder was meaningful, further BPA loadings were prepared. Figure 3 15 shows the 1 H spectra of the BPA loadin gs from 7 to 30%. Figure 3 15. 1 H Solid State MAS NMR spectra of additional BPA loadings
90 For the 20% BPA loading, 2 new small peaks are found at 9.7 and 10.9 ppm. The presence of 2 peaks is puzzling, however this strong downfield shift is typically in dicative of a strong hydrogen bond. Since the CSH gel may have both silanols and interlayer water, this may provide two different chemical environments for the BPA hydroxyls to interact with. At the 30% BPA loading there is the emergence of 2 additional small peaks at 14.8 and 17.7 ppm (not shown). These two peaks were also found on a different instrument with the same BPA loading. However, these peaks may be a result of contamination or impurities, as proton resonances this high are usually only due to P OH groups. 1 H MAS NMR experiments were conducted at a different location on a different instrument, on similar samples. The spectra of these BPA cement paste composites is shown in Figure 3 16. The 5% and 15% BPA loadings show a very small peak at 10 .4 ppm. However, comparison to the solid BPA is difficult due to the lack of detection of the hydroxyl resonance. This may be due to a slower spinning speed used in the experiment (30kHz). In any case, the detection of the small peak suggests some kind of interaction between the hydroxyls of the BPA exists. Unfortunately, further experimentation with other loadings of BPA on the cement paste did not produce these resonances above 10 ppm or new resonances that were visible at all.
91 Figure 3 16. 1 H So lid State MAS NMR spectra of various BPA loadings on cement paste 2D NMR Both spin diffusion experiments and double quantum experiments were performed to further characterize the specific interactions taking place between the small molecules and cement pas te. For the spin diffusion experiments, typically a mixing time of 40ms was used. Figures 3 17, 3 18, and 3 19 show the 2D spin diffusion and double quantum results for GPE, EDA, and BPA cement composites respectively. The loading of small molecules on the cement paste was 7% by weight.
92 Figure 3 17. 2D 1 H NMR spin diffusion and double quantum results for cement + GPE Figure 3 18. 2D 1 H NMR spin diffusion and double quantum results for cement + EDA Figure 3 19. 2D 1 H NMR spin diffusion and dou ble quantum results for cement + BPA
93 Figure 3 17 indicates the absence of interaction between specific GPE protons and cement paste. Since off diagonal peaks indicate interactions between specific groups and only diagonal peaks are present, we can assume that there are no interactions between GPE hydrogen groups and the cement paste. The resonances for the various groups have been approximated. The phenyl ring was assigned to the signal 7 ppm, the ether linkage at 4 ppm, and the epoxide hydrogens at 2 p pm. The diffusion experiment did not indicate any interaction between the organic and inorganic, only the organic molecule groups in spatial proximity with itself. The double quantum experiment gave an autocorrelation resonance around 1 ppm, indicating t he cement paste Ca(OH) 2 group interacting with itself. In Figure 3 18 the resonances for ED A at 1, 3 and 4 ppm were assigned to the CH 2 methyl and hydroxyl groups, respectively The diffusion experiment did not indicate any interaction between the organic and inorganic components, only the organic molecule groups exhibits a spatial proximity with itself. As with the previous double quantum experiment, the same intramolecular i nteraction was found. In Figure 3 19 the BPA resonances at 1 and 7 ppm were assigned to methyl groups and phenyl rings respectively For both the diffusion and double quantum experiment two sets of peaks were found, most likely corresponding to dipolar coupling between the methyl group and phenyl ring. However, since the cement paste also has a peak located around 1ppm due to Ca(OH) 2 there may be a possibility of an organic inorganic interaction. Therefore, further experiments on different loadings of BPA were performed.
94 Figure 3 20. 2D 1 H NMR spin diffusion experiments for 20 and 30% BPA. For the 2D spin diffusion experiments with 20 and 30% BPA, shown in Figure 3 20, the same correlation between the 1 ppm and 7 ppm appeared, as evidenced in the red circles within the larger blue region. However, for the Cement + 20% BPA, there is an additional interaction at 10 ppm (seen as a blue dot in the top left corner, and as a region of ellipsoidal shapes in the bottom right corner) with the broad peak fr om 6 to 8 ppm and 1 ppm. This may indicate the hydroxyl group interacting with the phenyl ring, methyl group, or cement paste. While this interaction is not as strong as the BPA methyl phenyl ring interaction, this is to be expected due to the relatively weak strength of hydrogen bonds. In addition, spin diffusion experiments only give spatial information, and not necessarily a chemical interaction. Unfortunately, the resolution is not sufficient to determine if the interactions are intramolecular or in termolecular. 29 Si experiments were performed to see if any interactions between cement silanols and bisphenol A took place. Id eally, HETCOR 1 H > 29 Si experiments should allow determination of the presence of possible cross peaks between Si OH signals and 1 H resonances of BPA. Unfortunately, no 1 H > 29 Si CP MAS spectra could be recorded
95 with a reasonable S/N ratio, probably due to an averaging of the 1 H 29 Si dipolar coupling induced by the mobility of the BPA molecules. Figure 3 21 shows the cement paste before and after treatment. Original Portland cement contains Q 0 sites in the calcium silicate phase, that are unreacted material without a network of Si O Si bonds in the silicates 114a Q 1 sites are present in the form of linked silicate tetrahedr al endgroups and Q 2 sites are present in the middle of silicate chains 116 The 29 Si spectra of the cement paste contains the Q 0 site located at 72 ppm, the Q 1 site at 80 ppm, and the Q 2 site at 85 ppm. No significant change in the chemical shift of these peaks was found after treatment with BPA. A slight increase in the Q 2 /Q 1 ratio from 0.4 to 0.8 was found, indicating a change in the degree in the silicate polymerization. The broadness of the Q 2 peak demonstrates the presence of a variety of diffe rent silicate chain lengths, and the peak at 86.7 on the shoulder of the Q 2 peak is attributed to the different local environment, such as bridging and paired groups 117 The small peaks between 100 and 110 for the cement + BPA curve are most likely no ise.
96 Figure 3 21. 29 Si NMR spectra of cement paste before and after treatment with BPA. Summary Various FTIR techniques were used to investigate EDA, GPE, and BPA cement composites. No change in the FTIR hydroxyl region from 3,000 to 3,600 cm 1 was obse rved for any of the organic inorganic composites. Peak subtraction and peak deconvolution also showed that there is no hydrogen bonding interaction between epoxy and cement. No strong interactions were found between small molecules and laponite or silica nanoparticles with 13 C and 1 H NMR experiments. 1D and 2D 1 H MAS NMR experiments indicated an interaction for the BPA cement paste composites. Specifically, the shift of the OH group of the BPA to higher chemical shift values was found in some experiment s. The downfield chemical shift of these OH groups suggest deshielding of the proton, which is typically an indication of hydrogen bonding. However, since the amount of hydrogen bonding is susceptible to factors such as acidity, concentration, and tempera ture it can be difficult to predict, so the nature of these
97 organic inorganic interactions could not be determined. 29 Si experiments on the BPA cement composites indicated a change in the degree of silicate polymerization, but did not suggest direct evide nce of a silanol hydroxyl interaction. While these sets of experiments do not show a compelling set of data demonstrating hydrogen bonding between the organic inorganic composite, it is possible that with a higher water to cement ratio (w/c=0.4 0.5) and a different type of Portland cement (white cement) the nature of the bonding may be described more fully.
98 CHAPTER 4 EPOXY DEGRADATION BY ENVIRONMENTAL EXPOSU RE Background While use of FRPs for repair and strengthening of bridges and other infrastructure is routinely performed, accurate prediction of the long term mechanical properties of these materials in the field is quite difficult. While accelerated testing on FRP mat erials typically involves elevated temperatures, exposure to aqueous solutions, and increased stress levels, these methods typically do not take into account the degradation mechanisms experienced by the materials. A number of investigations have determin ed that environmental exposure of FRP s leads to a significant loss of properties as discussed in Chapter 2 In all cases, a significant loss in bond strength was observed, or the failure mode changed from cohesive in the concrete to adhesive at the interf ace with exposure 63 In some cases, a loss in properties is explained by chain cleavage due to water hydrolyzing crosslinks 51 Aiello et al examined the mechanical properties of epoxy and concrete beams after environmental exposure to air from 40 C an d water immersion at 23 C 118 They determined the mechanical properties of the epoxy were almost completely recovered after the service conditions were returned to room temperature They concluded that the effect of an increase of temperature on bond st rength is only due to the viscoelastic properties of the adhesive, and the presence of water induces a reduction in bond strength and a high er weakness at the interface Ellis et al discussed the importance of sample handling and preparation during DSC to determine the Tg of epoxy under various conditions. From their data they implied that the ice melting endotherms observed were due to the presence of free
99 water produced by the diluents diffusion out of the sample during thermal analysis. This is in agreement with their previous research, which suggested that the free volume approach can be used to explain the effect of water on Tg and that water acts as a plasticizer i n epoxy systems 36 Two different approaches have been used for explaining property loss by plasticization in epoxy systems. The first is the disruption of interchain hydrogen bonds by water molecules, which plasticizes the polymer 38a, 119 Alternativel y, under wet conditions, water can reach the interfacial region and break the hydrogen bonds between the epoxy and substrate since water is a strong hydrogen bonding agent. This will result in displacing the epoxy from the substrate and create a weak wate r layer at the interface 3c As discussed in Chapter 2, water molecules can also be found clustered together into sub micro scale cavities within the epoxy matrix 120 In this study, accelerated aging conditions are used to investigate a loss in properties and change in degradation mechanism. Experimental Procedure Two different epoxy systems were investigated in this work. The model epoxy system consisted of a diglycidyl ether of bisphenol A (DGEBA, EPON 826) and poly(oxypropylene) diamine (POPDA, Jeffami ne D 230) purchased from Momentive and Huntsman, respectively. The other commercial system investigated (Sikadur 300) had an unknown chemical composition, due to the proprietary nature of the chemicals. The mass ratio of DGEBA to POPDA for the model epox y system was 100 to 32.9, to reach stoichiometric equivalence between functional groups while the commercial system components were mixed by weight The two liquid components were mixed vigorously for 5 minutes to ensure even mixing. The
100 mixed material was then degassed for 30 minutes under vacuum to remove air bubbles, and transferred into dogbone molds made from Silastic T 2, a commercial PDMS mold system, using a pipette. The dimensions of the dogbones are corresponding to ASTM D638 Type V dogbones. After 1 week of curing at room temperature, samples were demolded and placed in controlled water baths with constant temperatures from 30 90 C. Samples immersed in wa ter from 30 60 C were kept at the UF Coastal Engineering Labs, and samples from 70 90 C were kept at the State Materials Office Labs. Samples were also exposed to UV aging with and without water at 60 C using a Q Sun Xenon Model XE 3 weatherometer maintai ning 0.68 W/m 2 and a relative humidity of 45%. Every 2 weeks samples were removed from the environmental chambers and tested for mechanical properties and weighed for water absorption Five dogbones per experimental condition were tested for mechanical properties to get an accurate population size. Samples were tested in tension using an Instron according to ASTM D3039 using a strain rate of 5 mm/min and a 1,000 lb load cell until failure. All mechanical data was analyzed using IBM SPSS using an multiv ariate analysis of variance (MANOVA) test. A general linear model was used, with repeated measures, with a 95% confidence level. Within subjects variables were temperature, or UV exposure and the between subjects factor was time; the covariate was the une xposed set of epoxy samples. In order to evaluate the mechanisms of accelerated degradation experienced by the epoxy dogbones Fourier transform infrared spectrometry was used (FTIR; Nicole t Magna 760, Thermo Electron Cor poration), in the diffuse reflectan ce infrared Fourier
101 transform (DRIFT) mode and ATR mode The DRIFT and ATR spectra were recorded over the range of 650 4000 cm 1 with a KBr beamsplitter and an MCTA detector using 64 scans at a resolution of 4 cm 1 For the DRIFT sample preparation, slices of the aged epoxy were taken after mechanical testing then combined with KBr powder in a proportion of 1:100. After the epoxy and KBr powder were milled for 30 seconds, spectra were acquired. Sample preparation for ATR samples involved cutting a middle section, close to the fracture surface of the sample, using a razor blade and clamping the part of the sample that was directly exposed to the UV radiation to the ATR crystal. The ATR spectra were taken using a hemispherical ZnSe crystal at an incident ang le of Results and Discussion Change in Mechanical Properties of the Model Epoxy System During Hygrothermal Exposure Over the course of eight weeks of exposure to water the model epoxy system exhibited a gradual color change from colorless to orange which can be seen in Figure 4 1. With increasing temperatures, this effect was more evident. This is most likely due to oxidation of the ether and nitrogen groups 59, 121 The samples exposed to UV radia tion also underwent a color change, which was accompanied by a loss in mass and an increase in surface roughness. The tensile test results after 8 weeks are summarized in Table 4 1 and the full list of data is shown in Appendix A along with examples of lo ad displacement curves in Appendix B. After 8 weeks of exposure, up to 33% of the modulus and 38% of the peak tensile strength was lost. The strain at failure increased up to 5.1% from 3.5%. The loss in physical properties for epoxy systems experiencing hygrothermal exposure has
102 been reported by other authors and is typically attributed to the formation of hydrogen bonds between the epoxy hydroxyl groups and adsorbed water 122 Figure 4 1. Change in color of epoxy dogbones after exposure to water at various temperatures and UV radiation Table 4 1. Summary of changes in properties for the model epoxy system after hygrothermal exposure Exposure t (weeks) E (GPa) E s MPa) s (%) s Control 8 2.45 0.08 74.5 4.8 3.5 0.4 0 30 8 2.13 0.16 64.1 2.1 4.6 0.8 1.14 40 8 1.93 0.07 65.5 4.8 4.7 0.7 1.34 50 8 1.91 0.15 67.6 6.2 5.1 0.9 1.35 60 8 1.77 0.14 56.5 6.2 4.7 0.4 0.82 70 8 1.65 0.07 49 3.5 4.5 0.7 4.36 80 8 1.67 0.19 49 3.5 4.3 0.2 7.47 90 8 1.62 0.17 46.2 4.1 4.2 0.5 6.22 Note : t = time; E = modulus; E s = standard deviation of E; = peak stress; s = standard deviation of ; = strain at failure; s = standard deviation of A = % change in cross sectional ar e a after exposure, but prior to mechanical testing.
103 It is possible to predict the mechanical properties of crosslinked polymer systems by using rubber elasticity theory. A required assumption is that there is no volume change on deformation and that affine deformation occurs 123 The average length of a ch ain in the strained state can be determined by (4 1) where represents the fractional change in dimension of each of the three directions. For most polymers, is equal to approximately 1. The stress may be predicted by the equation of state for rubber elasticity. (4 2) In this equation represents the concentration of active network chain segments, R is the gas co nstant, and T is temperature. When a polymer becomes swollen this results in a decrease in the network chain segment concentration, and therefore a decrease in stress. A swollen polymer is also predicted to exhibit a decrease in modulus, due to a decrea se in crosslink density, v e (4 3) The engineering stress and strain were determined from the measured load and displacement, using the original cross sectional area of samples. In the elastic region, modulus being the constant of proportionality. (4 4) Once the polymer passes through the elastic region, plastic deformation occurs and the polymer starts to yield. Beyond the yield point, necking begins and a decrease
104 in the cross sectional area occurs. Until the neck forms, the deformation is essentially uniform throughout the sample, but after necking all further deformation takes place in the neck. The neck becomes sm aller and smaller, with local stress increasing, until the sample fails. Figure 4 2 shows an example of a stress strain graph for a control sample Figure 4 2. Stress It is evident from the graph that the unexposed, control system has a greater slope, corresponding to a higher modulus, than the exposed sample. In addition, there is a higher peak stress, and a lower strain at failure. The control sample does not underg o any plastic deformation and fails in a brittle manner, while the exposed sample undergoes necking and much more elongation as a result of absorption of water in the epoxy network.
105 Figure 4 3 shows the changes in the modulus for samples exposed to water at temperatures from 30 to 90 C over the period of 8 weeks; note that only one error bar for a set of samples is present for ease of visibility. From the graph it seems that after 4 to 6 weeks, the value of the modulus remains constant. The increase in sample mass confirmed saturation of the sample around 4 weeks and will be discussed later in the chapter. While some additional decrease or increase may be found after the initial 4 weeks, these changes are not statistically significant. Results of the s tatistical analysis are shown in Appendix E. As predicted by rubber elasticity theory, with higher exposure temperatures, lower values for the modulus were found. Based off of rubber elasticity, a 10% increase in volume, due to swelling, and assuming a c onstant number of crosslinks, the new density would be 1/(1.1 x area), or 90.9% of the original density so the modulus would only drop about 9%. Thus, other factors besides swelling may have impacted the modulus, such as plasticization and chain scission. However, since the Tg of these systems was not measured, plasticization cannot be verified. Sample modulus over time was not. Samples exposed to 30 C showed the le ast loss in properties, while those at 90 C showed the most loss in properties. The same basic trend was found for peak stress shown in Figure 4 4 The Tg of this system is 45 C after 1 week of cure in ambient conditions, as measured by DSC 34 Once ag ain, after about 4 weeks the loss in properties seems constant, however, there was no statistically significant difference in the stress values with any of the variables.
106 Figure 4 3. Change in modulus for the model epoxy system with hygrothermal exposu re Figure 4 4. Change in peak stress for the model epoxy system with hygrothermal exposure
107 Samples exposed to water demonstrated a greater strain at failure than the control sample as shown in Figure 4 5 While the difference in strain as a function of temperature was statistically significant, the change in strain over time was not. For all the exposed samples the strain throughout the exposure time was higher than the control group, as a result of chain swelling due to water absorption The increase in the cross sectional area was found to increase up to 7% at 8 weeks of exposure, which indicates an interruption of inter chain hydrogen bonding rather than simply occupying the free volume 124 Figure 4 5. Change in strain for the model epoxy system with hygrothermal exposure Change in Mechanical Properties of the Model Epoxy System During UV and Water Exposure UV exposure was conducted on the model epoxy system with and without the presence of water at 60 C The change in the tensile modulus after exposure is shown in Figure 4 6 and the control and water exposure at 60 C are also provided for
108 comparison. While some loss in properties occurred, it seems the UV exposure was not as detrimental to the m odulus as water exposure alone. Initial degradation up to 2 effect on the tensile modulus is more dominant than UV exposure. However, the only statistically significan t relationship was the change in modulus as a function of the time. There was no statistically significant difference between the UV and UV with water exposure conditions. A table of the details of the mechanical properties may be found in Appendix C. Figure 4 6. Change in modulus with UV and water exposure for the model epoxy system The peak stress of samples exposed to UV exposure are shown in Figure 4 7. Unfortunately, none of the variables showed a statistically significant difference.
109 Fig ure 4 7. Change in peak stress with UV and water exposure for the model epoxy system Figure 4 8 shows the change in strain during UV exposure. Note error bars have been omitted. Over time, the strain decreased with UV exposure. The combined effects of UV degradation and absorption of water makes a clear explanation difficult. Once again, none of the variables showed a statistically significant difference. Figure 4 8. Change in strain with UV and water exposure for the model epoxy system
110 Change in Mechanical Properties of the Commercial Epoxy System The commercial epoxy system (Sikadur 300) showed a similar trend as the model epoxy system in regards to change in modulus and peak strength with exposure to water at various temperatures. In addition the same gradual color change from colorless to orange was seen. The full list of data for the Sikadur 300 epoxy system is shown in Appendix D. Figure 4 9 shows the change in modulus with water exposure for the Skidaur 300 epoxy system. Figure 4 9. Change in modulus for Sikadur 300 with hygrothermal exposure As with the model epoxy system, the tensile modulus decreased with higher temperatures; note error bars have been removed for ease in visibility. A lower modulus was seen for samples exposed ab modulus showed a statistically significant difference, while time did not.
111 Figure 4 10. Change in peak stress with hygrothermal exposure for Sikadur 300 The same pattern follows for the peak strength. With h igher exposure temperature, a lower peak strength was found. As with the modulus, the effect of temperature on the peak stress showed a statistically significant difference, while time did not. Figure 4 11. Change in strain with hygrothermal exposure for Sikadur 300
112 Unlike the model epoxy system, there was no significant change in the strain to failure with exposure. The chemical explanation for the lack of change in strain with exposure is not known, due to the proprietary chemical nature of the ep oxy systems. There was no statistically significant difference for any of the variables for the change in strain for the commercial system. Diffusion of Water Into the Model Epoxy System Water diffusion in epoxy resins has frequently been described by Fi law 120 This mathematical expression is described in equation 4 5, where C represents concentration of the diffusing substance as a function of time t and position x, and D is the diffusion coefficient. (4 5) The total amount of substance diffusing into the epoxy resin, M t as a function of time across the thickness, h, is given by the integral of the solution of equation 4 5. This equation assumes the diffusion occurs only in one dimension. M max is the maximum amount of the diffusing substance at infinitie time. (4 6) This curve can be divided into two parts, when Dt/h 2 is greater than 0.05 and less than 0.05. If Dt/h 2 > 0.05 equation 4 6 can be rewritten as (4 7) If Dt/h 2 < 0.05 equation 4 6 can be rewritten as (4 8)
113 Figure 4 12 shows the diffusion behavior of the model epoxy system at various temperatures with exposure up to 8 weeks, where % weight gain is plotted against t 1/2 Water uptake in the epoxy ap the initial stage the sorption curve is linear, and above the linear region the sorption approaches saturation. The diffusion coefficient can be calculated from the slope of the linear region of a plot of water uptake (M t ) vs. time (t 1/2 ), shown in Figure 4 13. Samples were typically saturated within 3 weeks. Water uptake was obtained from the average weight gain of two samples during exposure, where M 0i is the initial mass and M ti is the mass of the sample at time i (4 9) Figure 4 12. Absorption of water by model epoxy system at various temperatures up to 8 weeks exposure
114 Figure 4 13. Fitting of the linear region of the various samples From these line ar fits, the diffusion coefficients were calculated where m is the slope of the fit. M sat corresponds to the point at which the epoxy samples were saturated with water, or no significant change was shown in the weight gain. (4 10) The diffusion coefficients for the different water temperatures are summarized in Table 4 2. Since the sample dimensions used in this study are not flat sheets, but dogbone shaped with a more complex geometry, the calculated diffusion coefficients are hig her than the true diffusion coefficients It is important to note that this study only investigates a short time period of 8 weeks, so it is assumed that primarily stage 1 diffusion occurs.
115 Table 4 2. Diffusion coefficients at different water temperature s for the model epoxy system D (10 9 m 2 /s) 30 20.36 40 34.76 50 39.71 60 41.45 70 68.92 80 97.85 90 126.36 Assuming Fickian behavior of water diffusing into the epoxy, an Arrhenious equation can be used to calculate the diffusion activation energy, Ea, where D o is a pre exponential factor, R is the gas constant, and T is absolute temperature. (4 11) Fitting ln D against 1,000/T gives a slope E a /R, by which the activation energy can be determined. This plot is shown in Figure 4 14. The calculated activation energy for the model epoxy system was 27.29 2.89kJ/mol. This activation energy is in agreement to those calculated by other authors for similar epoxy systems, where Nunez et al. calculated 26.01 kJ/mol for their DGEBA diaminecyclohexane system 122c and Barral et al. calculated 25.8 kJ/mol for their DGEBA bisaminomethylcyclohexane system 125
116 Figure 4 14. Calculation of Ea/R for the model epoxy system IR Characterization of Degradation DRIFT and ATR IR experiments were performed on the model epoxy system to determine the mechanism of property loss for the cured epoxy after various environmental exposure conditions. Table 4 3 indicates IR assignment of the various absorption peaks. These assignments are in good agreement with those found by other authors 63, 126 Table 4 3. IR peak assignment of the absorption bands in the model epoxy system 43, 127 Peak Position (Wavenumber, cm 1 ) IR Assignment 829 In plane deformation of phenyl H 917 Epoxy bending 1,033 C O C stretching 1,097 O C C stretching 1249, 1182 C C O C stretching 1297 Twisting mode of CH 2 units 1456, 1362 C H in plane deformation in
117 aliphatic units 1609 1582 1510 Aromatic ring stretching 1650 O H bending vibration 2960 2930 2870 C H and CH 2 stretching 3050 Phenyl H stretching 3300 N H stretching 3440 Hydroxyl stretching Figure 4 15. DRIFT IR spectra of the model epoxy system. a) unexposed epoxy sample after 8 weeks and b) exposed samples after 8 weeks. Figure 4 15 shows the IR spectra of the samples exposed to water for 8 weeks from 30 to 90 C. While there is much overlap for the e xposed samples, it appears that there is an increase in the hydroxyl region from 3,100 to 3,700 cm 1 relative to the aromatic ring stretch at 1510 cm 1 typically used as an internal standard. In addition there is the appearance of the OH bending vibratio n of water at 1625 cm 1 To estimate the amount of absorbed water for each exposure condition, the area under these peaks was obtained and normalized to the area under the phenyl ring. While these normalized Peak Position ( Wavenumber, cm 1 ) IR Assignment Table 4 3. Continued
118 values cannot be used to determine absolute concentrations, they can be used for relative comparisons. This process has been demonstrated by other authors on epoxy system 32a, 34 These values are shown in Table 4 4. While the normalized areas indicate that the epoxy network has adsorbed a large am ount of water, unfortunately, there is no trend as a function of exposure temperature. Table 4 4 Area comparisons of water absorption peaks relative to the phenyl ring for exposure at various water temperatures Temperature 1510 cm 1 Area 3434 cm 1 Area 1625 cm 1 Area H 2 O / Phenyl OH / Phenyl Total Water / Phenyl control 1.994 28.146 4.794 2.404 14.115 16.519 30 0.809 29.728 3.79 5 4.68 9 36.728 41.416 40 0.50 9 25.674 3.347 6.58 3 50.489 57.072 50 0.575 21.261 2.997 5.209 36.956 42.165 60 0.721 45.929 5.364 7.4 40 63.70 2 71.141 70 0.38 8 24.437 2.882 7.432 63.013 70.445 80 0.53 5 23.89 9 3.042 5.6 90 44.695 50.384 90 0.496 28.722 3.5 60 7.17 4 57.884 65.057 The bending mode of water in epoxy occurs at a lower wavenumber than the band of liquid water at 1,648 cm 1 Typically, the formation of hydrogen bonds affects the vibrational spectra of the groups involved by decreasing the frequency of stretching modes and increasing the frequency of bending modes 126b For water a bsorption in epoxies, it has been proposed by other authors that since the wavenumber of the water held in the epoxy resin is between the wavenumber of free water and liquid water, it is hydrogen bonded to the epoxy, and this interaction is reversible 40, 1 04 In the current study, the presence of the absorption band at 1,625 cm 1 suggests that the water absorbed by the epoxy has hydrogen bonded to the hydroxyl groups of the epoxy. Using rubber elastic ity theory, with swelling of the polymer network by hydrogen
119 bonding of water with the epoxy, a decrease in strength is predicted. While the epoxy demonstrated swelling up to 7%, the loss in strength was much greater. Figure 4 16 ATR IR spectra of samples after 8 weeks of exposure to a) UV and water b) UV only and c) unexposed Figure 4 16 shows the IR spectra of the samples exposed to UV and water for 8 weeks at 60 C. Spectra have been corrected for residual CO 2 and H 2 O in the chamber using calibra ted spectra in the OMNIC software package. In addition the spectra were ATR and baseline corrected for ease of comparison. It seems that after UV exposure there is an increase in OH groups around 3,300 cm 1 and C=O groups at 1,650 cm 1 and 1,725 cm 1 These changes have been reported by other authors as indicative of the oxidation mechanism 35, 63, 127b, 128 In addition, for the samples exposed to both UV radiation and water immersion at 60C, a decrease in ether groups was found at 1,085 for the trans ether link and 1,232 cm 1 for the aromatic ether ink and can be seen in Figure 4 17. A comparison of the relative areas of the OH, CO, and C=O to the phenyl group is shown in Table 4 5. The results of the table show a large increase in C=O groups and a small decrease in C O groups with exposure to UV only. With UV and
120 water exposure, there is a large increase in OH groups, and a large decrease in C O groups. The formation of carbonyl groups may result from oxygen attack of ether groups or methylene gro ups linking two phenyl groups in the DGEBA epoxy resin 3b Since the sample exposed to UV only did not show a large increase in the OH area, it can be assumed that solely oxidation took place. In the presence of both UV radiation and water, it is generall y believed that hydrolysis of the ester linkages occurs, with the formation of a hydroxyl and carbonyl 43 This process decreases the effective average crosslink molecular weight, and absorbed moisture can attack the cross linked chains and cause chain sci ssion. As suggested by rubber elasticity theory, a decrease in the average crosslink molecular weight will result in a proportional drop in modulus which is larger than expected by swelling alone. From Table 4 5, only half the number of C O groups are de tected for samples exposed to UV and water after 8 weeks, as compared to the control, yet only a 15% reduction of modulus was found. Figure 4 17. ATR IR spectra of samples after 8 weeks of exposure to a) UV only b) unexposed and c) UV and water
121 Table 4 5. Area comparisons of various absorption peaks relative to the phenyl ring for exposure to UV radiation with and without water Exposure 1510 cm 1 Area 3434 cm 1 Area 1085 cm 1 Area 1232 cm 1 Area 1725 cm 1 Area OH / Phenyl C=O / Phenyl Total C O /Phenyl control 1.948 9.426 2.914 4.585 0.048 4.839 0.000 3.850 UV 1.01 8 3.55 0 0.63 6 2.708 0.463 3.489 0.455 3.286 UV+ water 1.10 2 9.60 3 0.287 2.015 0.081 8.715 0.074 2.089 Summary Both the model epoxy system and commercial epoxy system exhibited a decrease in mechanical properties with exposure to water at various temperatures. The model epoxy system increased in strain to failure with exposure, while the commercial epoxy system did not show much change in strain to failure with exposure. DRIFT IR spectra of the model epoxy system after exposure to water indicated an increase in hydroxyls and an increase in water at 3,400 cm 1 and 1625 cm 1 respectively. UV samples exhibited a loss in properties, however, the data was not statistically significant. The IR data indicated an increase in carbonyl groups, suggesting that with UV exposure epoxy undergoes oxidation. The samples exposed to both UV radiation and water showed a decrease in ether groups and an increase in hydroxyl groups, providing evidence for hyd rolysis. Although IR data suggested hydrolysis took place for samples exposed to UV and water, the mechanical data did not show as large of a reduction as reported by other authors 71 and as implied by the theory of rubber elasticity.
122 CHAPTER 5 MODIFICATION OF CEME NT PASTE SURFACES WI TH SILANE COUPLING AGENTS Background While repair and rehabilitation of civil infrastructure in the US is of great concern, there is much evidence which show s that the current methods and practices are not as resista nt to environmental conditions as expected This evidence is discussed in detail in Chapter 2 The mechanism proposed by numerous authors suggests that when epoxy is exposed to water, the water molecules replace epoxy concrete hydrogen bonds with water e poxy hydrogen bonds 4a, 38a Previous work on this topic by Choi et al demonstrated that the application of an epoxy functional silane coupling agent could ect on the concrete epoxy bond 129 This was attributed to the formation of covalent bonds, and a decrease in the diffusivity of water due to the hydrophobic silane 130 Slant shear measurements indicated that the application of the epoxy functional silane coupling agent lead to signific ant improvement in the durability of concrete cylinders exposed to water from 30 to 60 C over a 12 week period. This work further investigates the chemical modification of the cement surface at a chemical level. The reaction mechanism of a silane coupling agent with an inorganic material has been studied in depth by many researchers 131 Silicone is the center of the silane molecul e which contains an organic functional group (Y) and alkoxy functional groups (OR). The alkoxy groups become hydroxyl groups ( OH) with hydrolysis, and then form an alcohol (R OH). Over the course of the reaction, hydroxyl groups condense with each other as a result of dehydration. The end product is a chemical bond between an inorganic and organic material. This process is dep icted in Figure 5 1. Since hydrolysis
123 and condensation are dependent on pH and catalysts, typically a weak acid such as acetic acid is used in the procedure to ensure that the rate of hydrolysis is much greater than the rate of condensation. Figure 5 1 The reaction process of an alkoxy silane with a hydroxyl surface. 132 The determination of the contact angle between a liquid and a solid is the most commonly used technique to characterize the surface properties of a material. The relation between the contact angle and free energy of a liquid and solid is explained with shown in the following equation 0 = SG SL cos c (5 1) where t he solid vapor interfacial energy is SG the solid liquid interfacial energy is SL the liquid vapor energy is C is the equilibrium contact angle. As can be seen from the above equation, wetting of a solid surface with a high surface energy, such as a ceramic, can be achieved more readily than wetting a low energy solid surface, such as a polymer. On a smooth flat surface, contact angle m easurements are easy to perform. H owever, due to the nature of cementitious surfaces, surface roughness and
124 cavities and pores make these measurements difficult 133 Wenzel introduced the rough ness ratio r to quantify this surface roughness effect in the Young and Dupre equation under the form cos ( A ) = r cos ( ) (5 2) w A is the apparent contact angle and r is the ratio of the real rough surface area to the ideal smooth surface Note that r is greater than 1 for rough surfaces and equal to 1 difference between rough surfaces from smooth surfaces, it does not describe contact angle hysteresis, or t he difference between the advancing and receding contact angles. In rough surfaces, a wetting liquid will not be completely absorbed by the surface cavities, while a non wetting liquid may not penetrate into pores This will result in the formation of ai r pockets. Cassie and Baxter extended the Wenzel equation to accommodate non homogenous s urfaces with cavities and pores cos ( A ) = r f f 0 + f 1 (5 3) w here r f is the roughness ratio of the wet surface and f is the fraction of solid surface area wet by the liquid. While plenty of research has been dedicated to the impact of surface roughness on concrete strength, fundamental issues regarding the interfacial bond ha ve been overlooked 7a, 7c, 134 For the most part, past research only investi gated the micro scale and assume mechanical interlock, while ignoring issues at the nano scale. Garbacz et al. directly examined mechanical interlock, however their treatments to vary the roughness resulted in the smoothest untreated samples having the h ighest pull off strength 7c This was explained by the appearance of voids at the interface for the rougher surfaces.
125 A different study on a variety of materials including ceramics, polymers, and metals have found a direct correlation between surface rough ness (from 0.2 to 8 m) and contact angle 135 Kubiak et al. combined Wenzel and Cassie Baxter theories with 2D surface morphology analysis and found good agreement between their model and experimentally determined contact angle. (5 4) w here is the contact angle for the ideal surface, A is the measured contact angle, R Lo is the length of the roughness profile expressed in % expansion from the smooth profile, and R mr is the relative material ratio of the roughness profile measured in the vertical position. The most variation was found in the ceramic and copper alloy samples with minimums in the contact angle vs log average roughness graphs between 0.3 and 0.9 m as s hown in Figure 5 2 Figure 5 2. Variation in contact angle as a result of surface roughness. 135
126 M omber directly examined the contact angle of cement paste and concrete and found a decrease in the contact angle with grit blasting and sawing. He acknowled ged the fact that if the water drop covered grooves it would be lower in angle and that more investigations are necessary to determine a relationship between contact angle and the location of measurement 136 This study avoids these issues by polishing sam ples to a uniform surface roughness and by avoiding measurements on areas with grooves or areas with irregular surfaces. Besides physical property measurements, a specific spectroscopic technique was used to determine chemical changes after silane treatme nt. X Ray Photoelectron Spectroscopy (XPS) is a useful tool to examine the surface region of nearly any solid material. In this technique low energy X rays strike a sample, and emitted photoelectrons are analyzed. The binding energy of each of the emitted electrons can be deter mined by the following equation E binding = E photon (E kinetic + ) (5 5) w here E binding is the binding energy of the electron, E photon is the energy of the x ray photons, E kinetic is the kinetic energy of the electron a s measured by the instrument, and is the work function of the spectrometer. Each element produces a characteristic set of peaks at a range of binding energy values and the number of detected electrons in these peaks is directly related to the relative amount s of each element within the irradiated area. Atomic percentage values are determined by dividing the signal intensity by a sensitivity factor and normalizing over the elements detected. The objectives of this chapter are to determine the contact a ngles of cement paste with water to cement ratios of 0.4 and 0.5 before and after treatment with various
127 silane coupling agents. By using an automated polisher, a uniform surface roughness will be achieved, so contact angle variations will be minimal. Ch emical modification of the cement pastes will be analyzed using XPS. Experimental Procedure Commercial Florida Portland Type I/II cement was used to prepare cement pastes with water to cement ratios of 0.4 and 0.5 to evaluate the effect of different chemi cal compositions on silane treatment. After vigorous hand mixing for 5 minutes, the pastes were transferred to 4cm by 4cm teflon molds and were cured at room temperature for 1 week. After 1 week, samples were demolded and transferred to a lime water solu tion to cure for 3 weeks to complete the hydration process. Next, the samples were polished by hand using increasing grits of sandpaper from 300 to 1200. In order to further reduce surface roughness and improve uniformity, a 120 South Bay Technology lapp ing and polishing machine was used with 300, followed by 800 grit sandpaper until the surface developed a smooth, near mirror finish. In this work, the description of a near mirror finish refers to a visual light reflection on the surface of the cement pa ste. The near mirror finish on the cement paste sample is shown in Figure 5 3. All samples were then cleaned with hexane in a sonicator for approximately five minutes after polishing then rinsed with deionized water Miller et al. have demonstrated the ability of automated polishing to attain a mirror finish on concrete mortar 137
128 Figure 5 3. Near mirror finish of a cement paste sample after polishing Photo courtesy of Andrew Stewart. Polished cement pastes were prepared with three silane coupling a gents with specific functionalities. Aminopropyltriethoxy silane (APTES), 3 glycidyloxypropyltrimethoxy silane (GPTMS), and methoxy terminated polydimethxyl siloxane (PDMS) were chosen, for evaluating their compatibility with cement pastes. APTES was sup plied by Alfa aesar, GPTMS was supplied by Acros, and PDMS was supplied by Gelest All the silanes were used as received without further purification. Silane coupling agents were applied to the surface using two different procedures. Horizontal vapor de position was performed at 90 C for 1 hour to evaporate the silane and condense the silane on the cement paste surface This was performed by placing 2mL of the silane liquid in the bottom of rectangular glass holders with a hemisphere removed. Glass sli des held the cement paste above the liquid with the polished side down. Two samples for each water to cement ratio and silane coupling agent combination were prepared. The other procedure was performed using an aqueous solution where 1wt% of the selected coupling agent was added to a 90:10 by weight ethanol:deionized water mixture. After the pH was adjusted to approximately 5 by the addition of a few drops of acetic acid, the solutions were stirred for 60 minutes to allow
129 complete hydrolysis of the coupli ng agent. The silane solution was then pipetted onto the polished cement paste surface and was placed into an oven at 60 C for 1 hour 20 minutes. Tapping mode AFM measurements were performed to evaluate the surface roughness of the cement pastes Both topological and phase images were recorded with a Dimension 3100 using Nanoscope V. The scan rate was 1Hz and scan size was 50 m by 50 m using a probe with a cantilever spring constant of 40N/m and a resonance frequency around 270 kHz The free a mplitude A 0 of the cantilever tip in air was kept around 20nm and the setpoint A sp /A 0 was automatically determined by the instrument during engaging. FTIR samples were analyzed in DRIFT mode using KBr powder on a Magna IR E.S.P. System 760 Spectrometer wi th 128 sample and reference scans at 4cm 1 resolution from 600 to 4000 cm 1 Static contact angle measureme nts were obtained using a Ram H art Inc. auto pipetting goniometer, with a Schott Fostec Inc. lighting system. A 5 L droplet of deionized water wa s deposited onto the substrate and allowed to stabilize for 1 to 3 seconds before capturing the image used to analyze the experimental contact angle. Measurements were taken on 5 different locations for each sample. Survey and multiplex XPS s pectra were r ecorded with a Perkin Elmer 5100 XPS system with an Al K x ray source operating at 15kV and 300 mA. Prior to acquiring XPS spectra, samples were dried in a vacuum oven for 1 day at 10 0 o C to evaporate residual vapors, so an ultra high vacuum could be main tained in the instrument. The samples were mounted to the holder with double sided adhesive tape and placed in a
130 vacuum in the range of 10 8 to 10 7 torr. All samples were run in survey mode from 1200 to 0 eV with a pass energy of 89.45 eV and high resol ution multiplexes were taken using a pass energy of 35.75 eV and a step size of 0.05 eV. The take off angle was 45 degrees relative to the detector. Binding energy positions were calibrated against the adventitious carbon C1s peak position located at 285 eV. Spectra were analyzed and deconvoluted with AugerScan software. Results and Discussion AFM Measurements Previous research on imaging and modulus measurements on cementitious materials using AFM has indicated a variation in mechanical properties betwe en hydrated and unhydrated particles and highlighted the importance of polishing techniques 138 A specific polishing procedure was outlined by Miller et al that takes into consideration the multi phase composite nature of cement paste 137 Their technique involves grinding with polishing pads, with the sample mounted on a disc with adhesive with a constant weight on top. A slow lapping speed was used up to 8 hours to minimize sample disturbance. Roughness measurements verified the effectiveness of their procedure, showing a RMS roughness of less than 100 nm after 2 hours of polishing that decreased to less than 30 nm after 4 hours. Figure 5 4 shows the 2D and corr esponding 3D height images of polished cement paste. A 50 m by 50 m scan size was chosen so a relatively large area could be analyzed. If a smaller area was used it would not be representative of a typical sample, since most samples were approximately 3cm by 3cm. Based on a 50 m by 50 m scan size the RMS roughness was measured to b e 84.6nm. The small bright yellow
131 spots in the 2D height image are most likely noise. Using a Detak II A profilometer to measure a 5mm scan line gave an average roughness of less than 1 m, as compared to greater than 3 m for the unpolished cement. Sinc e the two techniques measure different length scales, this difference in roughness measurement was expected. Based off the work of Momber et al., the RMS roughness data suggests that the roughness of the cement paste lies in a range where small variations will not substantially modify contact angle measurements. Therefore, it can be assumed that the roughness after the silane treatment will have a negligible effect on the contact angle measurement. Figure 5 4. 2D and 3D AFM height images of polished c ement paste Contact Angle Measurements Initial measurements of the contact angle of unpolished cement pastes ranged from 70 to over 100 degrees. This variation correlates well with the results from the aforementioned study on various engineering surfaces by Kubiak et al 135 In this case the roughness of the unpolished surface was greater than 3 m. The cement pastes polished by hand had a roughness less than 1 m and showed a contact angle of
132 approximately 60 This value is similar to values found in t he literature, shown in Table 5 1, however, the high standard deviation of the data made it necessary to use a more controlled method, i.e. automated polishing. While this other research did examine the surface energy of cementitious materials using a var iety of methods, typical water to cement ratios of 0.4 to 0.5 were not investigated. Table 5 1. Measurement of contact angles using different techniques on various concrete sample types Measurement Method W/C Ratio Sample Type Contact Angle Wilhelmy plate and t ensiometer 133 N/A Solution 60 Captive drop t echnique 136 0.60 Bulk 50 66 Washburn m ethod 139 0.32 Powder 74 Figure 5 5 shows the contact angle measurements before and after treatment of the silane coupling agent s The initial contact angle is represented by the darker color on the left and the final contact angle by the lighter color. For all the samples, the initial contact angle was approximately 60 The average hysteresis was approximately 1.5 degrees for th e neat cement paste and 3 degrees after silane treatment. Overall, APTES decreased the contact angle, while PDMS increased the contact angle. This change in contact angle was expected since the amino terminated silane could hydrogen bond to deionized wat er during contact angle experiments, while the PDMS silanes would make the surface more hydrophobic. The epoxy terminated silane, GPTMS, had a negligible effect on the contact angle. The lack of change in the contact angle is most likely due to the epoxy group having neither a hydrophobic or hydrophilic preference. Furthermore, the 0.5 water to cement ratio samples demonstrated a greater change in contact angle. As the water to cement ratio increases, the spacing between cement particles increases and i s accompanied by a greater degree of hydration 17b, 140 Therefore, the 0.5 w/c ratio samples have a greater
133 number of hydroxyl groups or hydrated species as compared to 0.4 w/c ratio samples. This was conf irmed by DRIFT (Figure 5 6) and ATR (not shown) IR experiments. Figure 5 5. Contact angle measurements of silane treated cement pastes The initial contact angle is represented by the darker color on the left and the contact angle after silane treatment is indicated by the lighter color on the right. Figure 5 6. DRIFT IR spectra of 0.4 and 0.5 w/c ratio cement pastes
134 IR data confirmed the difference in OH and H 2 O present in samples with different water to cement ratios. Spectra were linearl y baseline corrected then normalized to the asymmetric stretching vibration of CO in CaCO 3 centered at 1460 cm 1 Visually, it is clear that there is a greater absorbance in OH groups from 3,000 to 3,600 cm 1 and H 2 O at 1,650 cm 1 for the 0.5 w/c ratio sa mple. Taking the areas under the in OH and H 2 O absorbance peak and dividing by the CO absorbance gave a semi quantitative value to the difference between the two types of cement pastes. Using this approach, the 0.5 w/c samples had 28% more hydroxyl group s than the 0.4 w/c samples. The higher number of hydroxyl groups correlates well with the contact angle measurements which indicated a greater change in contact angle with the greater water to cement ratio samples. While pull off bond strengths were not p erformed on the samples, research by Mansur et al. on polymer modified cement mortar glass tile composites indicated an increase in bond strength with a variety of different silanes 141 The suggested mechanism is the formation of a hybrid interface base d on hydrogen bonds, where hydroxyl side groups can act as a chemical cross linker agent between surface silanols and water found in the interlayer space of hydrated cement. In their system they used cement mortar modified with poly(ethylene co vinyl acet ate) with a water to cement ratio of 0.6. The system used in this study uses no sand and a lower water to cement ratio, however, their suggested interfacial bonding may be appropriate for comparison. The two proposed mechanisms by Choi et al. suggest tha t GPTMS interacts with hydrogen bonding species of concrete to form a covalent linkage after condensation takes place, or the GPTMS directly interacts with the epoxy to prevent moisture uptake. Although
135 Mansur et al. used EVA to promote hydrogen bonding o f surface silanols and water in the interlayer space with the polymer, the idea that these two groups are available to come into contact and interact with the silane coupling agent may be a key to the description of the bonding. XPS To explore the surface chemistry of the silane cement paste interaction, XPS experiments were performed. The XPS survey spectra of the cement paste before and after treatment with the silane coupling agents are shown below, in Figure 5 7. The elemental compositions and binding energies are summarized in Table 5 2. Figure 5 7. XPS survey of treated and untreated cement pastes While Portland cement is made up other elements besides C, O, Si, and Ca, these elements were chosen because they are the most dominant and easiest to
136 resolve in the XPS spectra. For the samples coated with the silane coupling agents, an increase in the atomic concentration of carbon, along with a decrease in oxygen and calcium was found. This trend has been found by other auth ors investigating cement paste before and after coating 9 In addition, this change in atomic concentration indicates the silanes were successfully deposited on the cement paste surface using the horizontal evaporation technique. The increase in the silic on composition is most likely due to the balancing of the elements to 100%. The small (less than 1%) amount of nitrogen found on the silane deposited samples, except APTES, is most likely due to oven cross contamination.
137 Table 5 2. Apparent surface composition and binding energies of treated and untreated cement pastes Sample at % C C1s B.E. (eV) at % O O1s B.E. (eV) at % Ca Ca2p3 B.E. (eV) at % Si Si2p B.E. (eV) at% N N1s B.E. (eV) C.P 37.9 285 / 289.5 46.2 531.7 11.4 347.4 / 351.1 4.3 102.3 0.2 400.4 C.P+PDMS 50.8 285 / 286.3 / 289.2 37.0 532.5 2.5 348.0 / 351.5 8.9 102.7 0.8 400.2 C.P+GPTMS 47.3 285 / 286.2 / 288.8 38.2 532.3 4.5 347.6 / 351.2 8.8 102.8 1.1 400.2 C.P+ATEPS 49.3 285 / 286.0 / 288.9 32.0 532.0 6.6 347.6 / 351.2 7.6 102.6 4.4 399.7 C.P+PDMS (aq) 46.1 285 / 286.5 / 288.9 39.4 532.0 9.6 347.5 / 351.2 3.8 102.6 1.1 400.2 C.P+GPTMS (aq) 44.8 285 / 286.7 / 289.3 41.2 532.0 8.5 347.6 / 351.1 4.7 102.4 0.8 400.0 C.P+APTES (aq) 59.9 285 / 285.9 /288.6 25.4 532.1 2.5 347.6 / 351.1 8 102.8 4.2 399.9 Table 5 3. Binding energy positions and relative amounts of bridging and non bridging atoms Sample Bridging O 1s B.E. (eV) % Non Bridging O 1s B.E. (eV) % Bridging Si 2p B.E. (eV) % Non Bridging Si 2p B.E. (eV) % C.P 532.58 24.5 531.5 75.5 103.22 21.1 102.16 78.9 C.P + PDMS 533.05 53.3 531.9 46.7 103.22 49.4 102.16 50.6 C.P + GPTMS 532.77 55.5 531.6 44.5 103.22 57.9 102.16 42.1 C.P + APTES 532.58 54.6 531.5 45.4 103.22 41.8 102.16 58.2 C.P + PDMS (aq) 532.58 39.9 531.5 60.1 103.22 33.0 102.16 67.0 C.P + GPTMS (aq) 532.77 39.9 531.5 60.1 103.22 22.5 102.16 77.5 C.P + APTES (aq) 532.58 54.1 531.5 45.9 103.22 58.5 102.16 41.5
138 In order to examine specific interactions between the silane coupling agent and cement paste, high resolution XPS data was taken at 0.05 eV intervals. A Shirley background and 80:20 Gaussian/Lorentzian peak shape was assumed in all cases. C1s peaks for t he neat cement paste was fit to 2 peaks, adventitious carbon at 285 eV and carbonates at 289.5 eV, as also reported by other authors 9b After silane depositions the carbonate peak underwent significant attenuation, along with the emergence of a new peak r anging from 285.9 to 286.7 eV, due to the aliphatic carbon chain. No significant changes in the Ca2p3 or N1s binding energies was found. The peak deconvolutions, shown as the dotted lines, for the various electron orbitals are shown in Figures 5 8 to 5 1 1. Note that the data have not been corrected for charging effects. Figure 5 8. C1s electron orbital XPS spectra including curve fitting of untreated cement paste
139 Figure 5 9. C1s electron orbital XPS spectra including curve fitting of ATEPS treate d cement paste Figure 5 10. Ca2p3 electron orbital XPS spectra including curve fitting of untreated cement paste
140 Figure 5 11. N1s electron orbital XPS spectra including curve fitting of ATEPS treated cement paste While other authors have shown a change in peak shape and position in XPS spectra after a change in the chemical environment of a material 142 the findings in the current study are more subtle. In order to determine if the silane treatment was successful, O1s and Si2p binding energies w ere fit with 2 peaks, shown in Table 5 3. The greater binding energy corresponds to bridging atoms and the lesser binding energy corresponds to non bridging atoms. Bridging oxygen atoms are bonded to silicon atoms, in Si tetrahedral chains, while non bri dging oxygen atoms are typically bound to calcium atoms or hydroxides 143 Curve fitting of the O1s line was performed in a method similar to Black et al, where an 80:20 Gaussian/Lorentzian peak shape was as sumed and a 1.9eV FWHM was used to fit the two types of oxygens. A typical O1s electron orbital using these parameters, including curve fitting is shown in Figure 5 12. In this study the bridging
141 and non bridging oxygen contributions of the neat cement p aste were determined by fitting the non bridging oxygen at 531.5 eV and the bridging oxygen at 532.58 eV. This technique was verified by another control sample with good accuracy. The silane treated samples were also fit to these same values so compariso ns could be made between samples. However, for a few of the samples an accurate fit using these values could not be made so the binding energy position was allowed to change up to 0.5 eV. Overall, both types of treatments (evaporation and aqueous) showed an increase in the number of bridging oxygens relative to non bridging oxygens. In the condensation step of the silane coupling agent reaction Si O bonds are formed between the silane and substrate and silane and itself, as shown in Figure 5 1. If the s ilanes had not undergone the condensation step, there would be an increase in non bridging oxygen atoms. Therefore, the increase in bridging oxygen atoms suggests that the various silanes successfully bonded to the cement paste surface. The Si 2p line w as deconvoluted in the same method as the O 1s line, but with a 1.8 eV FWHM. A typically Si 2p electron orbital, including fitting, is shown in Figure 5 13. Larger FWHM values were tried but the fittings were of lower accuracy. As with the O 1s fittings the binding energy positions were kept consistent so contributions of bridging and non bridging silicon atoms could be determined. While the bridging silicon atoms increased with silane treatment, the increase was greater for silanes using the evaporati on technique with the exception of the APTES silane. It is highly probable that the evaporation technique resulted in a monolayer of silane being deposited on the surface, while the aqueous technique deposited a thicker layer.
142 Figure 5 12. O1s electron orbital XPS spectra including curve fitting of untreated cement paste Figure 5 13. Si2p electron orbital XPS spectra including curve fitting of untreated cement paste The motivation behind examining 2 types of silane deposition lies in a previous work by Choi et al. that investigated the durability of epoxy concrete samples with
143 mechanism for improved durability in their system was a change from hydrogen to covalent bonding at the interface, which now seems unlikely due to the minimal change in the bridging silane atoms using that technique. Nguyen et al. found that using aminoethylaminopropypltrimethoxysilane (0.1 wt%) decreased epoxy and E glass moisture absorption and improved interlaminar shear strength 144 Ji et al. found that using GPTMS (5 wt%) as an additive to epoxy prevented moisture absorption 145 This improve the strength of the composite system, rather than a change in bonding. In addition, the relatively high (1 wt%) amount of silane seems likely to interact with the epoxy as much as with the cement paste. Mansur et al also cited a hydrogen bonding mechanism to explain his composite system, but since changes in the binding energies due to hydrogen bonding are quite difficult to detect in XPS this mechanism cannot be verified. Summary In this study, the effect of various silane treatments on cement paste chemistry was investigated. This investigation examined the hypothesis that silanols on cement paste surface may interact with silane coupling agents to create a new functionalized surface. From the AFM data, the care in surfa ce preparation was essential for interpretation of the contact angle changes. Surface roughness has a strong effect of the surface energy in this system and due to the inherent irregular nature of cement, a uniform surface is essential for acquiring quant itative results. The contact angle data demonstrated successful treatment of the surface with the silane coupling agents. APTES made the surface more hydrophillic, PDMS improved the hydrophobicity of the surface, and GPTES had no significant change. In addition, higher water to cement
144 ratio samples demonstrated a higher impact on the change in contact angle. Deconvolution of the O 1s and Si 2p electron orbitals were performed to determine contributions from bridging and non bridging atoms. An increase in bridging silicon and oxygen atoms indicated successful silane condensation and a covalent bond was formed between the cement paste and silanes. Furthermore, in using this peak fitting strategy, the aqueous silane deposition was shown to be less success ful than the evaporation technique.
145 CHAPTER 7 GENERAL CONCLUSIONS AND FUTURE WORK The first part of this work investigated the nature of the epoxy concrete bond. The initial hypothesis was that hydrogen bonding takes places at the interface, so small mo lecule analogs of the epoxy molecule were adsorbed onto cement paste particles to create a more simple system to analyze. ATR FTIR experiments on the surface of these composites did not show any change in the hydroxyl area (3,000 cm 1 to 3,700 cm 1 ) or ot her significant changes giving evidence of a hydrogen bond. Solid state NMR experiments showed a weak interaction between BPA and cement paste and small change in the chemical shift of the OH group. A few select samples with 20 and 30% BPA showed new pea ks with chemical shift values above 10ppm, however these results were not reproducible. 2D NMR experiments also suggested a small interaction, but the functional groups undergoing the interaction could not be directly determined. Most likely, for these s mall molecules and this specific type of Portland cement with a 0.38 water to cement ratio, strong hydrogen bonding does not exist. One of the main limitations of the investigation of hydrogen bonding in this research was only using a water to cement ratio of 0.38. Perhaps using at 0.4 or 0.5 water to cement ratio would increase the number of cement hydrates so there would be a greater number of sites available for interaction on the cement paste surface. Also, the cured cement particle was ground down to 38 m or less, so perhaps even smaller particle sizes down to 1 m or less would give a stronger signal due to the larger surface area. In addition, Type I/II Portland cement was used while other authors have used white cement, due to the lower iron cont ent which reduces the interference of paramagnetic relaxation 115 Another consideration for NMR experiments could be to
146 use D 2 O rather than H 2 O. Due to the smaller magnetic moment of deuterons, the broadening of the spectra would be much less, so that mu ch better resolution may be acquired. The second part of this work fo cused on accelerated aging of the two epoxy systems, including a model DGEBA POPDA system, and a commercial system. Mechanical testing was performed along with IR spectroscopy and diffus ion modeling to determine the change in properties and mechanisms of degradation. A decrease in tensile strength and modulus and an increase in strain to failure was found with higher temperatures, which was consistent with statistical analysis. An incre ase in hydroxyl and H 2 O regions in the IR spectra was found after 8 weeks exposure. With UV and UV with water exposure, the modulus did not decrease as much as with water alone. The strain after UV exposure decreased less than the control after 8 weeks, reducing the ductility of the samples which is typical of UV degradation. However, the data from UV degraded samples was not statistically significant. IR data provided evidence of the oxidative degradation mechanism for samples exposed to UV radiation a lone, while evidence of hydrolysis was found for samples exposed to UV and water. While the goal of this part of the study was to examine how the mechanical properties changed with exposure to accelerated aging conditions, the epoxy loss in properties did not change after 2 to 4 weeks. However, the lack of further change in mechanical properties may be due to the relatively short testing duration (8 weeks). Other authors have reported the mechanism of hydrolysis with water exposure for epoxy in experiment s up to 12 months 52 RH 146 which this study did not find. In the event of hydrolysis, due to the breaking of
147 polymer chains, a greater loss in properties is expected than chain swelling by water absorption. Therefo re, more extensive testing periods covering up to 6 months are essential, along with a greater variation in temperatures to more fully investigate the change in properties and mechanisms of degradation. Also, varying the type of curing agent may be intere sting to examine other evidence for degradation mechanisms. The final section of this work used silane coupling agents to modify cement pastes. The contact angle was shown to increase for the PDMS based silane, decrease for the amino based silane and not change for the epoxy based silane. The 0.5 w/c ratio samples demonstrated a greater change in contact angles, due to a greater amount of silanols and OH groups, as compared to the 0.4 w/c ratio cement paste. Fitting of the O1s and Si2p peaks, i ndicated an increase in bridging oxygen and silicon atoms, giving evidence for covalent bonds being formed between the cement paste and silane coupling agents. Aqueous silane treatments of the cement paste surfaces did not indicate successful bond formati ons for the PDMS and GPTES silanes on the cement paste surfaces. For use in field conditions on civil engineering sites, the aqueous technique would be more appropriate, but large variations and no trends regarding the contact angle limited the findings u sing that technique. Future work on a greater variation of silane coupling agents and application techniques would supplement this part of the work and be quite useful in field conditions.
148 APPENDIX A MECHANICAL PROPERTIE S OF THE MODEL EPOXY SYSTEM DURING HYGROTHERMAL EXPOSUR E Temp Exposure Time (weeks) Peak Stress (MP a) Stdev Modulus (GP a) Stdev Strain (%) Stdev Control 0 64.12 12.41 2.34 0.26 3.48 1.10 2 71.71 8.96 2.47 0.23 3.38 0.30 4 6 6.18 8.27 2.45 0.08 3.43 1.00 6 6 8.26 9.65 2. 48 0.20 2.95 0.60 8 74.46 4.83 2.45 0.08 3.50 0.40 30 2 67.57 10.34 2.24 0.30 3.90 0.36 4 66.88 4.14 2.19 0.10 3.75 0.40 6 66.19 4.14 2.10 0.16 4.40 0.60 8 64.12 2.07 2.13 0.16 4.56 0.76 40 2 64.81 6.21 2.12 0.26 4.47 0.80 4 63.43 4.14 2.02 0.19 4.03 0.40 6 64.81 2.07 2.04 0.12 4.20 0.40 8 65.50 4.83 1.94 0.07 4.70 0.70 50 2 65.50 4.83 2.07 0.10 4.80 0.40 4 67.57 4.83 2.17 0.17 4.50 0.40 6 59.29 8.96 1.97 0.13 3.97 0.90 8 67.57 6.21 1.91 0.15 5.10 0.90 60 2 57.23 5.52 2.01 0.08 6.05 3.80 4 54.47 5.52 1.81 0.20 5.23 1.80 6 55.16 0.69 1.81 0.19 4.49 0.45 8 56.54 6.21 1.78 0.14 4.73 0.37 70 2 59.98 4.83 1.94 0.21 4.95 0.90 4 54.47 4.83 1.92 0.20 4.70 1.30 6 49.64 3.45 1.80 0.09 3.95 0.34 8 48.95 3.45 1.65 0.07 4.52 0.73 80 2 47.57 2.76 1.77 0.15 4.30 0.70 4 48.26 5.52 1.70 0.14 4.10 0.25 6 51.02 4.83 1.74 0.14 4.20 0.32 8 48.95 3.45 1.67 0.19 4.30 0.20 90 2 53.78 0.69 1.95 0.15 3.90 0.30 4 46.19 2.76 1.72 0.15 3.70 0.50 6 46.19 6.21 1.82 0.17 3.40 0.80 8 46.19 4.14 1.62 0.17 4.20 0.50
149 APPENDIX B TYPICAL LOAD DISPLACEMENT CURVES FOR HYGROTHERMALLY E XPOSED SAMPLES AFTER 8 WEEK S Unexposed 30 C
150 40 C 50 C
151 60 C 70 C
152 80 C 90 C
153 APPENDIX C MECHANICAL PROPERTIE S OF THE MODEL EPOXY SYSTEM DURING UV EXPOSURE Exposure Exposure Time (weeks) Peak Stress (MP a) Stdev Modulus (GP a) Stdev Strain (%) Stdev UV 1 73.77 0.69 2.22 0.09 5.22 0.53 2 75.15 3.45 2.28 0.18 4.96 0.45 4 65.50 13.10 1.97 0.11 4.61 1.47 6 59.29 17.24 1.93 0.19 4.10 1.64 8 42.75 11.72 2.27 0.21 2.02 0.49 UV + Water 1 66.19 4.14 2.20 0.09 4.77 0.30 2 64.12 1.38 2.07 0.11 4.26 0.08 4 61.36 6.21 2.04 0.28 4.29 1.09 6 68.26 3.45 2.01 0.18 4.77 0.20 8 50.33 9.65 2.03 0.04 2.82 0.93
154 APPENDIX D MECHANICAL PROPERTIE S OF SIKADUR 300 DUR ING HYGROTHERMAL EXPOSURE Temperature Exposure Time (weeks) Peak stress (MPa) St dev Modulus (GPa) St dev Strain Std ev control 0 62.05 6.21 2.09 0.16 3.663 0.50 2 63.43 11.03 2.12 0.18 4.106 0.41 4 68.26 2.07 2.35 0.13 4.629 1.34 6 65.50 4.14 2.29 0.20 3.611 0.17 8 68.26 3.45 2.29 0.08 4.267 0.81 30 2 62.05 3.45 1.97 0.18 4.187 0.26 4 56.54 4.14 2.02 0.07 5.616 0.38 6 51.71 2.07 2.04 0.14 3.297 0.36 8 56.54 2.76 2.00 0.22 4.153 0.82 40 2 56.54 6.89 1.90 0.13 4.240 0.48 4 50.33 6.21 1.82 0.17 4.429 0.79 6 48.26 4.83 1.80 0.22 3.688 0.62 8 50.33 3.45 1.79 0.08 3.785 0.21 50 2 51.02 7.58 1.76 0.18 3.893 0.20 4 44.13 3.45 1.63 0.16 3.885 0.49 6 44.82 4.83 1.81 0.12 3.717 0.61 8 46.88 2.76 1.74 0.11 3.874 0.24 60 2 46.19 4.14 1.71 0.08 3.923 0.38 4 44.13 3.45 1.53 0.15 4.294 0.72 6 42.75 4.14 1.70 0.13 6.924 4.41 8 45.51 2.07 1.67 0.17 3.995 0.67 70 2 48.95 4.14 1.78 0.11 4.415 1.00 4 44.13 2.07 1.61 0.08 4.128 0.31 6 39.30 2.07 1.68 0.04 9.836 11.72 8 42.06 2.76 1.61 0.11 3.642 0.27 80 2 44.82 2.76 1.61 0.15 3.785 0.25 4 41.37 2.76 1.54 0.08 4.246 0.98 6 39.99 3.45 1.68 0.09 5.212 3.85 8 44.82 3.45 1.58 0.24 3.955 0.49
155 APPENDIX E MANOVA RESULTS FOR M ODULUS VALUES FOR THE MODEL EPOXY SYST EM WITH HYGROTHERMAL EX POSURE Multivariate Tests c Effect Value F Hypothesis df Error df Sig. temperature Pillai's Trace .653 4.084 a 6.000 13.000 .016 Wilks' Lambda .347 4.084 a 6.000 13.000 .016 Hotelling's Trace 1.885 4.084 a 6.000 13.000 .016 Roy's Largest Root 1.885 4.084 a 6.000 13.000 .016 temperature Control Pillai's Trace .613 3.434 a 6.000 13.000 .030 Wilks' Lambda .387 3.434 a 6.000 13.000 .030 Hotelling's Trace 1.585 3.434 a 6.000 13.000 .030 Roy's Largest Root 1.585 3.434 a 6.000 13.000 .030 temperature Week Pillai's Trace .834 .962 18.000 45.000 .516 Wilks' Lambda .367 .879 18.000 37.255 .604 Hotelling's Trace 1.224 .793 18.000 35.000 .694 Roy's Largest Root .665 1.662 b 6.000 15.000 .198 a. Exact statistic b. The statistic is an upper bound on F that yields a lower bound on the significance level. Within Subjects Design: temperature For this analysis a 95% confidence interval was used. If Sig. is 0.05 or less, then the difference between the means is stat istically significant. Hypothesis df and Error df are the v1 and v2 values that are used to identify the tabled value in the f distribution table where = 0.05. If the test statistic, F, is greater than that tabled value, then the null hypothesis (the m eans are equal), is rejected. This is to say that there is a statistically significant difference. When there are more than two levels, there are multiple ways in which the data can be combined to separate the groups, so there are four test criteria to d etermine the Largest Root only uses the variance from the dimension that separa tes the groups most. In this specific example, a MANOVA was run to determine if changes in temperature had a significant effect on the modulus values of my samples. The values in the Sig column are less than 0.05 for temperature and temperature compare d to the control, but not for temperature compared to the week. Therefore, the change in the modulus as a function of week was not statistically significant, but the change in the modulus as a function of temperature, and as compared to the control was s tatistically significant. All of the four criteria yielded the same result for all experiments.
156 LIST OF REFERENCES 1. Chen, J. S.; Ober, C. K.; Poliks, M. D.; Zhang, Y. M.; Wiesner, U.; Cohen, C., Controlled degradation of epoxy networks: analysis of cro sslink density and glass transition temperature changes in thermally reworkable thermosets. Polymer 2004, 45 (6), 1939 1950. 2. Hu, Y. F.; Ellyin, F.; Xia, Z. H., An experimental investigation of normal and shear stress interaction of an epoxy resin and model predictions. Polym. Eng. Sci. 2001, 41 (11), 2047 2060. 3. (a) Jones, R.; Chiu, W. K.; Hanna, S., Potential Faiure mechanism s of bonded composite repairs for metal and concrete Theoretical and Applied Fracture Mechanics 1994, 21 (2), 107 119; (b) Hong, S. G., The Thermal Oxidative Degradation of an Epoxy Adhesive on Metal Substrates. XPS and RAIR Analyses. Polym. Degrad. St abil. 1995, 48 (2), 211 218; (c) Schmidt, R.; Bell, J., Epoxy adhesion to metals Springer Berlin / Heidelberg: 1986; Vol. 75, pp 33 71. 4. (a) Lefebvre, D. R.; Elliker, P. R.; Takahashi, K. M.; Raju, V. R.; Kaplan, M. L., The critical humidity effect in the adhesion of epoxy to glass: role of hydrogen bonding. J. Adhes. Sci. Technol. 2000, 14 (7), 925 937; (b) Dewimille, B.; Bunsell, A. R., Accelerated aging of a glass fiber reinforced epoxy resin in water C omposites 1983, 14 (1), 35 40. 5. Adams, R. D.; Singh, M. M., The dynamic properties of fibre reinforced polymers exposed to hot, wet conditions. Compos. Sci. Technol. 1996, 56 (8), 977 997. 6. (a) Chu, W.; Wu, L. X.; Karbhari, V. M., Durability evaluati on of moderate temperature cured E glass/vinylester systems. Compos. Struct. 2004, 66 (1 4), 367 376; (b) Karbhari, V. M., E glass/vinylester composites in aqueous environments: Effects on short beam shear strength. J. Compos. Constr. 2004, 8 (2), 148 156; (c) Chu, W.; Karbhari, V. M., Effect of water sorption on performance of pultruded E glass/vinylester composites. J. Mater. Civ. Eng. 2005, 17 (1), 63 71; (d) Ramirez, F. A.; Carlsson, L. A.; Acha, B. A., Evaluation of water degradation of vinylester and epoxy matrix composites by single fiber and composite tests. J. Mater. Sci. 2008, 43 (15), 5230 5242. 7. (a) Czarneck, L., Adhesion A challenge for concrete repair. In Concrete Repair, Rehabilitation and Retrofitting II Alexander, M. G., Ed. Taylor & F rancis Group: London, 2008; pp 935 940; (b) Karbhari, V. M.; Engineer, M.; Eckel, D. A., On the durability of composite rehabilitation schemes for concrete: Use of a peel test. J. Mater. Sci. 1997, 32 (1), 147 156; (c) Garbacz, A.; Gorka, M.; Courard, L., Effect of concrete surface treatment on adhesion in repair systems. Mag. Concr. Res. 2005, 57 (1), 49 60; (d) Momayez, A.; Ehsani, M. R.; Ramezanianpour, A. A.; Rajaie, H., Comparison of methods for evaluating bond strength between
157 concrete substrate and repair materials. Cem. Concr. Res. 2005, 35 (4), 748 757. 8. (a) deBruyne, N. A., The adhesive properties of epoxy resins. J. Appl. Chem. 1956, 6 303 310; (b) Schmidt, R. G.; Bell, J. P., Epoxy adhesion to metals. Adv. Polym. Sci. 1986, 75 33 71. 9. (a) Neves, M. I. B.; Oliva, V.; Mrabet, B.; Connan, C.; Chehimi, M. M.; Delamar, M.; Hutton, S.; Roberts, A.; Benzarti, K., Surface chemistry of cement pastes: a study by x ray photoelectron spectroscopy. Surface and I nterface Analysis 2002, 33 (10 11), 834 841; (b) Djouani, F.; Connan, C.; Delamar, M.; Chehimi, M. M.; Benzarti, K., Cement paste epoxy adhesive interactions. Constr. Build. Mater. 2011, 25 (2), 411 423. 10. MacLaren, D. C.; White, M. A., Cement: Its chem istry and properties. Journal of Chemical Education 2003, 80 (6), 623 635. 11. Bye, G. C., Portland cement : composition, production and properties Thomas Telford: London, 1999. 12. Taylor, H. F., Cement Chemistry 2 ed.; Thomas Telford Limited: Heron Q uay, London, 1997. 13. Grutzeck, M. W., A new model for the formation of calcium silicate hydrate (C S H). Materials Research Innovations 1999, 3 (3), 160 170. 14. (a) H.J.H, B., The work of Powers and Brownyard revisited: Part 1. Cem. Concr. Res. 2004, 34 (9), 1697 1716; (b) H.J.H, B., The work of Powers and Brownyard revisited: Part 2. Cem. Concr. Res. 2005, 35 (10), 1922 1936. 15. Brunauer, S.; Kantro, D. L.; Copeland, L. E., The Stoichiometry of the Hydration Dicalcium Silicate and Tricalcium Si licate at Room Temperature. Journal of the American Chemical Society 1958, 80 (4), 761 767. 16. Feldman, R.; Sereda, P., A model for hydrated Portland cement paste as deduced from sorption length change and mechanical properties. Materials and Structures 1968, 1 (6), 509 520. 17. (a) Hamlin M, J., A model for the microstructure of calcium silicate hydrate in cement paste. Cem. Concr. Res. 2000, 30 (1), 101 116; (b) Tennis, P. D.; Jennings, H. M., A model for two types of calcium silicate hydrate in the mi crostructure of Portland cement pastes. Cem. Concr. Res. 2000, 30 (6), 855 863. 18. Jennings, H. M., A model for the microstructure of calcium silicate hydrate in cement paste. Cem. Concr. Res. 2000, 30 (1), 101 116.
158 19. Franceschini, A.; Abramson, S.; M ancini, V.; Bresson, B.; Chassenieux, C.; Lequeux, N., New covalent bonded polymer calcium silicate hydrate composites. Journal of Materials Chemistry 2007, 17 (9), 913 922. 20. H.F.W, T., Nanostructure of CSH: Current status. Adv. Cem. Based Mater. 1993, 1 (1), 38 46. 21. Kondo, K. F. a. W., Estimation of Thermochemical Data for Calcium Silicate Hydrate (C S H). J. Am. Ceram. Soc. 1983, 66 (12), 220 221. 22. Cong, X.; Kirkpatrick, R. J., 29Si MAS NMR study of the structure of calcium silicate hydrate. A dv. Cem. Based Mater. 1996, 3 (3 4), 144 156. 23. Pellenq, R. J. M.; Kushima, A.; Shahsavari, R.; Van Vliet, K. J.; Buehler, M. J.; Yip, S.; Ulm, F. J., A realistic molecular model of cement hydrates. Proc. Natl. Acad. Sci. U. S. A. 2009, 106 (38), 16102 16107. 24. Udoeyo, F. F.; Brooks, R.; Udo Inyang, P.; Nsan, R. O., Early Prediction of Laterized Concrete Strength by Accelerated Testing. IJRRAS 2010, 5 (1), 1 5. 25. Naik, T. R., Utilization of Accelerated Strength Testing Methods. Cem. Concr. Res. 1979, 9 (1), 7 17. 26. Resheidat, M. R.; Ghanma, M. S., Accelerated strength and testing of concrete using blended cement. Adv. Cem. Based Mater. 1997, 5 (2), 49 56. 27. Gharaibeh, H. M. T. a. E. S., Early Estimation of Hardened Concrete Strength. Journa l of Applied Sciences 2006, 6 (3), 543 547. 28. Meyer, D., A Statistical Comparison of Accelerated Concrete Testing Methods. In Journal of Applied Mathematics and Decision Sciences 1997; Vol. 1, pp 89 100. 29. Bicerano, J., Prediction of Polymer Properties CRC Press: Midland, Michigan, USA, 2002. 30. Mark, J. E., Physical Properties of Polymers Cambridge University Press: 2004. 31. Ratna, D., Handbook of Thermoset Resins Smithers Rapra Technology: 2009. 32. (a) Musto, P.; Mascia, L.; Ragosta G.; Scarinzi, G.; Villano, P., The transport of water in a tetrafunctional epoxy resin by near infrared Fourier transform spectroscopy. Polymer 2000, 41 (2), 565 F., The effect of hygrothermal fatigue on physical/mechanical properties and morphology of neat epoxy resin and graphite/epoxy composite. J. Appl. Polym. Sci. 1985, 30 (6), 2527 2549.
159 33. Soles, C. L.; Yee, A. F., A discussion of the molecular mechanisms of moisture transport in epoxy resins. J. Polym. Sci. Pt. B Pol ym. Phys. 2000, 38 (5), 792 802. 34. Choi, S.; Douglas, E. P., Complex Hygrothermal Effects on the Glass Transition of an Epoxy Amine Thermoset. ACS Appl. Mater. Interfaces 2010, 2 (3), 934 941. 35. Gu, X.; Stanley, D.; Byrd, W. E.; Dickens, B.; Vaca Tri go, I.; Meeker, W. Q.; Nguyen, T.; Chin, J. W.; Martin, J. W., Linking Accelerated Laboratory Test with Outdoor Performance Results for a Model Epoxy Coating System. In Service Life Prediction of Polymeric Materials Martin, J. W.; Ryntz, R. A.; Chin, J.; Dickie, R. A., Eds. Springer US: 2009; pp 3 28. 36. Ellis, T. S.; Karasz, F. E., Intereaction of Epoxy Resins with Water The Depression of Glass Transition Temperature. Polymer 1984, 25 (5), 664 669. 37. Jelinski, L. W.; Dumais, J. J.; Cholli, A. L.; Ellis, T. S.; Karasz, F. E., Nature of the Water Epoxy Interaction. Macromolecules 1985, 18 (6), 1091 1095. 38. (a) Zhou, J. M.; Lucas, J. P., Hygrothermal effects of epoxy resin. Part I: The nature of water in epoxy. Polymer 1999, 40 (20), 5505 5512; (b) Zhou, J. M.; Lucas, J. P., Hygrothermal effects of epoxy resin. Part II: variations of glass transition temperature. Polymer 1999, 40 (20), 5513 5522. 39. Bockenheimer, C.; Fata, D.; Possart, W., New aspects of aging in epoxy networks. I. Thermal aging. J. Appl. Polym. Sci. 2004, 91 (1), 361 368. 40. Bockenheimer, C.; Fata, D.; Possart, W., New aspects of aging in epoxy networks. II. Hydrothermal aging. J. Appl. Polym. Sci. 2004, 91 (1), 369 377. 41. Fredj, N.; Cohendoz, S.; Feaugas, X.; Touzain, S., Some consequences of saline solution immersion on mechanical behavior of two marine epoxy based coatings. Prog. Org. Coat. 2010, 69 (1), 82 91. 42. Papanicolaou, G. C.; Kosmidou, T. V.; Vatalis, A. S.; Delides, C. G., Water absorption mechanism and some anomalous effects on the mechanical and viscoelastic behavior of an epoxy system. J. Appl. Polym. Sci. 2006, 99 (4), 1328 1339. 43. Lin, Y. C.; Chen, X.; Wang, Z. P., Effects of hygrothermal aging on anisotropic conductive adhesive joints: experiments and theoretical analysis. J. Adhes. Sci. Technol. 2006, 20 (12), 1383 1399.
160 44. Yang, Q. A.; Xian, G. J.; Karbhari, V. M., Hygrothermal ageing of an epoxy adhesive used in FRP strengthening of concrete. J. Appl. Polym. Sci. 2008, 107 (4), 2607 2617. 45. Doyle, G.; Pethrick, R. A., Environmental effects on the ageing of epoxy adhesive joints. Int. J. Adhes. Adhes. 2009, 29 (1), 77 90. 46. Singh, R. P.; Khait, M.; Zunjarrao, S. C.; Korach, C. S.; Pandey, G., Environmental Degradation and Durability of Epoxy Clay Nanocomposites. Journal of Nanomaterials 2010, 2010 47. Shi, X. D.; Hinderliter, B. R.; Croll, S. G., Environmental and time dependence of moisture transportation in an epoxy coating and its significance for accelerated weathering. J. Coat. Technol. Res. 2010, 7 (4), 419 430. 48. Rezig, A.; Nguyen, T.; Martin, D.; Sung, L.; Gu, X.; Jasmin, J.; Martin, J. W., Relationship betwee n chemical degradation and thickness loss of an amine cured epoxy coating exposed to different UV environments. JCT Res. 2006, 3 (3), 173 184. 49. Hu, J. W.; Li, X. G.; Gao, J.; Zhao, Q. L., UV aging characterization of epoxy varnish coated steel upon exp osure to artificial weathering environment. Mater. Des. 2009, 30 (5), 1542 1547. 50. Lim, K. B.; Lee, D. C., Study of the surface degradation mechanism of an epoxy insulator exposed to water. J. Korean Phys. Soc. 2000, 37 (1), 49 54. 51. Fata, D.; Possar t, W., Aging behavior of a hot cured epoxy system. J. Appl. Polym. Sci. 2006, 99 (5), 2726 2736. 52. Xian, G. J.; Karbhari, V. M., DMTA based investigation of hygrothermal ageing of an epoxy system used in rehabilitation. J. Appl. Polym. Sci. 2007, 104 (2 ), 1084 1094. 53. Frigione, M.; Aiello, M. A.; Naddeo, C., Water effects on the bond strength of concrete/concrete adhesive joints. Constr. Build. Mater. 2006, 20 (10), 957 970. 54. Antoon, M. K.; Koenig, J. L., Irreversible Effects of Moisture on the Ep oxy Matrix in Glass Reinforced Composites J. Polym. Sci. Pt. B Polym. Phys. 1981, 19 (2), 197 212. 55. Li, L.; Yu, Y. F.; Su, H. H.; Zhan, G. Z.; Li, S. J.; Wu, P. Y., The Diffusion Mechanism of Water Transport in Amine Cured Epoxy Networks. Appl. Spectro sc. 2010, 64 (4), 458 465.
161 56. Wu, P.; Siesler, H. W., Water diffusion into epoxy resin: a 2D correlation ATR FTIR investigation. Chemical Physics Letters 2003, 374 (1 2), 74 78. 57. Noda, I., Two dimensional infrared spectroscopy. Journal of the American Chemical Society 1989, 111 (21), 8116 8118. 58. Musto, P., Two Dimensional FTIR Spectroscopy Studies on the Thermal Macromolecules 2003, 36 (9), 3210 3221. 59. Monn ey, L.; Belali, R.; Vebrel, J.; Dubois, C.; Chambaudet, A., Photochemical degradation study of an epoxy material by IR ATR spectroscopy. Polym. Degrad. Stabil. 1998, 62 (2), 353 359. 60. Meiser, A.; Willstrand, K.; Possart, W., Influence of Composition, H umidity, and Temperature on Chemical Aging in Epoxies: A Local Study of the Interphase with Air. J. Adhes. 2010, 86 (2), 222 243. 61. Malajati, Y.; Therias, S.; Gardette, J. L., Influence of water on the photooxidation of KHJ phenoxy resins, 1. Mechanism s. Polym. Degrad. Stabil. 2011, 96 (1), 144 150. 62. Xiao, G. Z.; Delamar, M.; Shanahan, M. E. R., Irreversible interactions between water and DGEBA/DDA epoxy resin during hydrothermal aging. J. Appl. Polym. Sci. 1997, 65 (3), 449 458. 63. Jana, S.; Zhon g, W. H., FTIR study of ageing epoxy resin reinforced by reactive graphitic nanofibers. J. Appl. Polym. Sci. 2007, 106 (5), 3555 3563. 64. Miyano, Y.; Nakada, M.; Sekine, N., Accelerated testing for long term durability of FRP laminates for marine use. J. Compos Mater. 2005, 39 (1), 5 20. 65. Yanzhong Zhang, Z. L. a. P. L., Study on the Behavior of Accelerated Ageing Fabric Reinforced Epoxy Composites. 29th International SAMPE Technical Conference 1997 197 206. 66. Chen, Y.; Davalos, J. F.; Ray, I.; Kim, H. Y., Accelerated aging tests for evaluations of durability performance of FRP reinforcing bars for concrete structures. Compos. Struct. 2007, 78 (1), 101 111. 67. Micelli, F.; Nanni, A., Durability of FRP rods for concrete structures. Constr. Build. Mater. 2004, 18 (7), 491 503. 68. Murthy, H. N.; Sreejith, M.; Krishna, M.; Sharma, S. C.; Sheshadri, T. S., Seawater Durability of Epoxy / Vinyl Ester Reinforced With Glass / Carbon Composites. J. Reinf. Plast. Compos 2010, 29 (10), 1491 1499.
162 69. Lee, M. C.; Peppas, N. A., Water Transport in Graphite Epoxy Composites. J. Appl. Polym. Sci. 1993, 47 (8), 1349 1359. 70. Ray, B. C., Effects of changing environment and loading speed on mechanical behavior of FRP composit es. J. Reinf. Plast. Compos. 2006, 25 (12), 1227 1240. 71. Kumar, H. G.; Singh, R. P.; Nakamura, T., Degradation of carbon fiber reinforced epoxy composites by ultraviolet radiation and condensation. J. Compos Mater. 2002, 36 (24), 2713 2733. 72. Callister, W. D.; Rethwisch, D. G., Fundamentals of Materials Science and Engineering: An Integrated Approach John Wiley & Sons: 2011. 73. Hulatt, J.; Hollaway, L.; Thorne, A., Preliminary investigations on the environmental effects on new heavyweight fa brics for use in civil engineering. Composites Part B: Engineering 2002, 33 (6), 407 414. 74. Boualem, N.; Sereir, Z., Accelerated aging of unidirectional hybrid composites under the long term elevated temperature and moisture concentration. Theoretical a nd Applied Fracture Mechanics 2011, 55 (1), 68 75. 75. Benzarti, K.; Chataigner, S.; Quiertant, M.; Marty, C.; Aubagnac, C., Accelerated ageing behaviour of the adhesive bond between concrete specimens and CFRP overlays. Constr. Build. Mater. 2011, 25 (2) 523 538. 76. Mufti, A. A.; Onofrei, M.; Benmokrane, B.; Banthia, N.; Boulfiza, M.; Newhook, J. P.; Bakht, B.; Tadros, G. S.; Brett, P., Field study of glass fibre reinforced polymer durability in concrete. Can. J. Civ. Eng. 2007, 34 (3), 355 366. 77. Lyons, J.; Laub, D.; Giurgiutiu, V.; Petrou, M.; Salem, H., Effect of hygrothermal aging on the fracture of composite overlays on concrete. J. Reinf. Plast. Compos. 2002, 21 (4), 293 309. 78. Leveque, D.; Schieffer, A.; Mavel, A.; Maire, J. F., Analysis o f how thermal aging affects the long term mechanical behavior and strength of polymer matrix composites. Compos. Sci. Technol. 2005, 65 (3 4), 395 401. 79. Toutanji, H. A.; Gomez, W., Durability characteristics of concrete beams externally bonded with FRP composite sheets. Cem. Concr. Compos. 1997, 19 (4), 351 358. 80. (a) Au, C.; Buyukozturk, O., Peel and shear fracture characterization of debonding in FRP plated concrete affected by moisture. J. Compos. Constr. 2006, 10 (1), 35 47; (b) Tuakta, C.; Byk ztrk, O., Deterioration of FRP/concrete bond system under variable moisture conditions quantified by fracture mechanics. Composites Part B: Engineering 2011, 42 (2), 145 154.
163 81. (a) El Hawary, M.; Al Khaiat, H.; Fereig, S., Performance of epoxy repaired concrete in a marine environment. Cem. Concr. Res. 2000, 30 (2), 259 266; (b) El Hawary, M.; Al Khaiat, H.; Fereig, S., Effect of sea water on epoxy repaired concrete. Cem. Concr. Compos. 1998, 20 (1), 41 52. 82. Mourad, A. H.; Abdel Magid, B.; El Maadda wy, T.; Grami, M., Effect of Seawater and Warm Environment on Glass/Epoxy and Glass/Polyurethane Composites. Applied Composite Materials 2010, 17 (5), 557 573. 83. Silva, M. A. G., Aging of GFRP laminates and confinement of concrete columns. Compos. Struc t. 2007, 79 (1), 97 106. 84. (a) Silva, M.; Biscaia, H., Degredation of Bond between FRP and RC Beams. Compos. Struct. 2008, 85 164 174; (b) Silva, M. A. G.; Biscaia, H. C., Effects of exposure to saline humidity on bond between GFRP and concrete. Compos Struct. 2010, 93 (1), 216 224. 85. Nakamura, T.; Singh, R. P.; Vaddadi, P., Effects of environmental degradation on flexural failure strength of fiber reinforced composites. Exp. Mech. 2006, 46 (2), 257 268. 86. Khan, L. A.; Nesbitt, A.; Day, R. J., Hygrothermal degradation of 977 2A carbon/epoxy composite laminates cured in autoclave and Quickstep. Compos. Pt. A Appl. Sci. Manuf. 2010, 41 (8), 942 953. 87. Park, C. G.; Jang, C. I.; Lee, S. W.; Won, J. P., Micros tructural investigation of long term degradation mechanisms in GFRP dowel bars for jointed concrete pavement. J. Appl. Polym. Sci. 2008, 108 (5), 3128 3137. 88. Lee, B. S.; Lee, D. C., Surface degradation properties of ultraviolet treated epoxy/glass fibe r. IEEE Trns. Dielectr. Electr. Insul. 1999, 6 (6), 907 912. 89. Kinloch, A. J., The science of adhesion. J. Mater. Sci. 1980, 15 (9), 2141 2166. 90. Kim, W. S.; Yun, I. H.; Lee, J. J.; Jung, H. T., Evaluation of mechanical interlock effect on adhesion s trength of polymer metal interfaces using micro patterned surface topography. Int. J. Adhes. Adhes. 2010, 30 (6), 408 417. 91. (a) Beushausen, H., The influence of concrete substrate preparation on overlay bond strength. Mag. Concr. Res. 2010, 62 (11), 84 5 852; (b) Ghafari, E.; Naderi, M., Effect of Different Curing Regime and Cementitious Materials on the Bond Strength of Self Compacting Mortars. Australian Journal of Basic and Applied Sciences 2011, 5 (12), 421 428.
164 92. Ye, L.; Friedrich, K.; Weimer, C. ; Mai, Y. W., Surface treatments and adhesion bonding between concrete and a CFRP composite. Adv. Compos. Mater. 1998, 7 (1), 47 61. 93. Kinloch, A. J., Adhesion and Adhesives Science and Technology. Chapman and Hall: New York, 1987. 94. Wake, W. C., Adhesion and the Formulation of Adhesives Applied Science Publishers: London, 1982. 95. Steiner, T., The hydrogen bond in the solid state. Angewandte Chemie (International ed. in English) 2002, 41 (1), 49 76. 96. (a) Lippmaa, E.; Magi, M.; Tarmak, M.; W ieker, W.; Grimmer, A. R., A high resolution Si 29 NMR study of the hydration of a tricalcium silicate Cem. Concr. Res. 1982, 12 (5), 597 602; (b) Wieker, W.; Grimmer, A. R.; Winkler, A.; Mgi, M.; Tarmak, M.; Lippmaa, E., Solid state high resolution 29Si N MR spectroscopy of synthetic 14 11 and 9 tobermorites. Cem. Concr. Res. 1982, 12 (3), 333 339. 97. Hennel, J.; Klinowski, J., Magic Angle Spinning: a Historical Perspective New Techniques in Solid State NMR. Klinowski, J., Ed. Springer Berlin / H eidelberg: 2005; Vol. 246, pp 1 14. 98. Massiot, D.; Fayon, F.; Capron, M.; King, I.; Le Calve, S.; Alonso, B.; Durand, J. O.; Bujoli, B.; Gan, Z. H.; Hoatson, G., Modelling one and two dimensional solid state NMR spectra. Magn. Reson. Chem. 2002, 40 (1), 70 76. 99. Feike, M.; Demco, D. E.; Graf, R.; Gottwald, J.; Hafner, S.; Spiess, H. W., Broadband multiple quantum NMR spectroscopy. Journal of Magnetic Resonance Series A 1996, 122 (2), 214 221. 100. Schnell, I.; Spiess, H. W., High Resolution 1H N MR Spectroscopy in the Solid State: Very Fast Sample Rotation and Multiple Quantum Coherences. Journal of Magnetic Resonance 2001, 151 (2), 153 227. 101. (a) Ghosh, S. N.; Handoo, S. K., Infrared and Raman spectral studies in cement and concrete (review). Cem. Concr. Res. 1980, 10 (6), 771 782; (b) Chollet, M.; Horgnies, M., Analyses of the surfaces of concrete by Raman and FT IR spectroscopies: comparative study of hardened samples after demoulding and after organic post treatment. Surface and Interface A nalysis 2011, 43 (3), 714 725; (c) Delgado, A. H.; Paroli, R. M.; Beaudoin, J. J., Comparison of IR techniques for the characterization of construction cement minerals and hydrated products. Appl. Spectrosc. 1996, 50 (8), 970 976.
165 102. (a) Mojumdar, S. C.; Raki, L. In Characterization and properties of calcium silicate hydrate polymer nanocomposites Baltimore, Maryland, 2005 04 10; Baltimore, Maryland, pp 1 10; (b) Gao, X. F.; Lo, Y.; Tam, C. M.; Chung, C. Y., Analysis of the infrared s pectrum and microstructure of hardened cement paste. Cem. Concr. Res. 1999, 29 (6), 805 812. 103. Slootmaekers, B.; Desseyn, H. O., Characterization of intermolecular hydrogen bonding in the solid state using variable temperature IR spectroscopy Appl. Spectrosc. 1991, 45 (1), 118 120. 104. Ngono, Y.; Marechal, Y.; Mermilliod, N., Epoxy amine reticulates observed by infrared spectrometry. I: Hydration process and interaction configurations of embedded H2O molecules. Journal of Physical Chemistry B 1999, 103 (24), 4979 4985. 105. Cummins, H. Z., Liquid, glass, gel: The phases of colloidal Laponite. Journal of Non Crystalline Solids 2007, 353 (41 43), 3891 3905. 106. Tawari, S. L.; Koch, D. L.; Cohen, C., Electrical Double Layer Effects on the Brownian Diffusivity and Aggregation Rate of Laponite Clay Particles. J. Colloid Interface Sci. 2001, 240 (1), 54 66. 107. Jaber, M.; Lambert, J. F., A New Nanocomposite: L DOPA/Laponite. J. Phys. Chem. Lett. 2010, 1 (1), 85 88. 108. Fei, B.; Chen C.; Peng, S. W.; Zhao, X. J.; Wang, X. H.; Dong, L. S., FTIR study of poly(propylene carbonate)/bisphenol A blends. Polymer International 2004, 53 (12), 2092 2098. 109. Wu, H. D.; Ma, C. C. M.; Chang, F. C., The solid state C 13 NMR studies of intermole cular hydrogen bonding formation in a blend of phenolic resin and poly(hydroxyl ether) of bisphenol A. Macromolecular Chemistry and Physics 2000, 201 (11), 1121 1127. 110. Churakov, S. V., Hydrogen bond connectivity in jennite from ab initio simulations. Cem. Concr. Res. 2008, 38 (12), 1359 1364. 111. Pellenq, R. J. M.; Lequeux, N.; van Damme, H., Engineering the bonding scheme in C S H: The iono covalent framework. Cem. Concr. Res. 2008, 38 (2), 159 174. 112. Wang, P. S.; Ferguson, M. M.; Eng, G.; Bentz, D. P.; Ferraris, C. F.; Clifton, J. R., H 1 nuclear magnetic resonance characterization of Portland cement: molecular diffusion of water studied by spin relaxation and relaxation time weighted imaging. J. Mater Sci. 1998, 33 (12), 3065 3071.
166 113. Cong, X.; Kirkpatrick, R. J., 29Si and 17O NMR investigation of the structure of some crystalline calcium silicate hydrates. Adv. Cem. Based Mater. 1996, 3 (3 4), 133 143. 114. (a) Rottstegge, J.; Wilhelm, M.; Spiess, H. W., Solid state NMR investigations on the role of organic admixtures on the hydration of cement pastes. Cem. Concr. Compos. 2006, 28 (5), 417 426; (b) Korb, J. P., NMR and nuclear spin relaxation of cement and concrete materials. Current Opinion in Col loid & Interface Science 2009, 14 (3), 192 202. 115. Janusa, M. A.; Wu, X.; Cartledge, F. K.; Butler, L. G., Sol id state deuterium NMR spectroscopy of D5 phenol in white Portland cement A new method for assessing solidification stabilization Environmen tal Science & Technology 1993, 27 (7), 1426 1433. 116. Hanna, R. A.; Barrie, P. J.; Cheeseman, C. R.; Hills, C. D.; Buchler, P. M.; Perry, R., Solid state Si 29 and Al 27 NMR and FTIR study of cement pastes containing industrial wastes and organics Cem. Concr. Res. 1995, 25 (7), 1435 1444. 117. Beaudoin, J. J.; Raki, L.; Alizadeh, R., A 29Si MAS NMR study of modified C S H nanostructures. Cement and Concrete Composites 2009, 31 (8), 585 590. 118. Aiello, M. A.; Frigione, M.; Acierno, D., Effects of envi ronmental conditions on performance of polymeric adhesives for restoration of concrete structures. J. Mater. Civ. Eng. 2002, 14 (2), 185 189. 119. Adamson, M. J., Thermal expansion and swelling of cured epoxy resin used in graphite epoxy composite materia ls. J. Mater. Sci. 1980, 15 (7), 1736 1745. 120. Pethrick, R. A.; Hollins, E. A.; McEwan, L.; Pollock, A.; Hayward, D.; Johncock, P., Effect of cure temperature on the structure and water absorption of epoxy/amine thermosets. Polymer International 1996, 39 (4), 275 288. 121. Noskov, A. M., Photochemistry of epoxy films. Journal of Applied Spectroscopy 1975, 23 (6), 1638 1642. 122. (a) Apicella, A.; Tessieri, R.; Decataldis, C., Sorption modes of water in glassy epoxies. J. Membr. Sci. 1984, 18 (MAR), 21 1 225; (b) Lu, M. G.; Shim, M. J.; Kim, S. W., Effects of moisture on properties of epoxy molding compounds. J. Appl. Polym. Sci. 2001, 81 (9), 2253 2259; (c) Nunez, L.; Villanueva, M.; Fraga, F.; Nunez, M. R., Influence of water absorption on the mechanic al properties of a DGEBA (n=0)/1, 2 DCH epoxy system. J. Appl. Polym. Sci. 1999, 74 (2), 353 358. 123. Sperling, L. H., Introduction to physical polymer science Wiley Interscience: 2006.
167 124. Adamson, M. J., Thermal expansion and swelling of cured epoxy resin used in graphite epoxy composite materials J. Mater. Sci. 1980, 15 (7), 1736 1745. 125. Barral, L.; Cano, J.; Lopez, A. J.; Lopez, J.; Nogueira, P.; Ramirez, C., Isothermal cure kinetics of a diglycidyl ether of bisphenol A/1,3 bisaminomethylcyclo hexane (DGEBA/1,3 BAC) epoxy resin system. J. Appl. Polym. Sci. 1995, 56 (9), 1029 1037. 126. (a) Nikolic, G.; Zlatkovic, S.; Cakic, M.; Cakic, S.; Lacnjevac, C.; Rajic, Z., Fast Fourier Transform IR Characterization of Epoxy GY Systems Crosslinked with Aliphatic and Cycloaliphatic EH Polyamine Adducts. Sensors 2010, 10 (1), 684 696; (b) Mertzel, E .; Koenig, J. L., Application of FT IR and NMR to epoxy resins Adv. Polym. Sci. 1986, 75 73 112. 127. (a) Mertzel, E.; Koenig, J., Application of FT IR and NMR to epoxy resins Epoxy : 1986; Vol. 75, pp 73 112; (b) Musto, P.; Ragosta, G.; Russo, P.; Mascia, L., Thermal Oxidative Degradation of Epoxy and Epoxy Bismaleimide Networks: Kinetics and Mechanism. Macromolecular Chemistry and Physics 2001, 202 (18), 3445 3458. 128. Delor Jesti n, F.; Drouin, D.; Cheval, P. Y.; Lacoste, J., Thermal and photochemical ageing of epoxy resin Influence of curing agents. Polym. Degrad. Stabil. 2006, 91 (6), 1247 1255. 129. Choi, S. M., S; Stewart, A; Hamilton, HR; Douglas, E. The Role of Chemical Bonding on the Durability of Epoxy Adhesion for Structural Strengthening Application. Polym. Eng. Sci. submitted. 130. Ramrus, D. A.; Berg, J. C., Using heterogeneous silane pa tterns to maintain adhesion and decrease water penetration into epoxy/aluminum interfaces. J. Adhes. Sci. Technol. 2006, 20 (14), 1615 1623. 131. (a) Matinlinna, J. P.; Lassila, L. V. J.; Vallittu, P. K., The effect of five silane coupling agents on the b ond strength of a luting cement to a silica coated titanium. Dental Materials 2007, 23 (9), 1173 1180; (b) Meng, X.; Yoshida, K.; Taira, Y.; Kamada, K.; Luo, X., Effect of Siloxane Quantity and pH of Silane Coupling Agents and Contact Angle of Resin Bondin g Agent on Bond Durability of Resin Cements to Machinable Ceramic. Journal of Adhesive Dentistry 2011, 13 (1), 71 78; (c) Kent, M. S.; Smith, G. S.; Baker, S. M.; Nyitray, A.; Browning, J.; Moore, G.; Hua, D. W., The effect of a silane coupling agent on wa ter adsorption at a metal/polymer interface studied by neutron reflectivity and angle resolved X ray photoelectron spectroscopy. J. Mater. Sci. 1996, 31 (4), 927 937. 132. Witucki, G. L., A si lane primer chemistry and applications of alkoxy silanes Jour nal of Coatings Technology 1993, 65 (822), 57 60.
168 133. Courard, L.; Michel, F.; Martin, M., The evaluation of the surface free energy of liquids and solids in concrete technology. Constr. Build. Mater. 2011, 25 (1), 260 266. 134. Jlio, E. N. B. S.; Bran co, F. A. B.; Silva, V. t. D., Concrete to concrete bond strength. Influence of the roughness of the substrate surface. Constr. Build. Mater. 2004, 18 (9), 675 681. 135. Kubiak, K. J.; Wilson, M. C. T.; Mathia, T. G.; Carval, P., Wettability versus roughn ess of engineering surfaces. Wear s2011, 271 (3 4), 523 528. 136. Momber, A. W., Surface issues of profiled cementitious composites. J. Adhes. 2002, 78 (3), 203 221. 137. Miller, M.; Bobko, C.; Vandamme, M.; Ulm, F. J., Surface roughness criteria for cem ent paste nanoindentation. Cem. Concr. Res. 2008, 38 (4), 467 476. 138. Mondal, P.; Shah, S. R.; Marks, L. D., Nanoscale characterization of cementitious materials. ACI Mater. J. 2008, 105 (2), 174 179. 139. Momber, A. W., The wettability of some concrete powders. Part. Sci. Technol. 2002, 20 (3), 243 246. 140. Bentz, D. P.; Aitcin, P. C., The Hidden Meaning of Water Cement Ratio. Concrete International 2008, 30 (5), 51 54. 141. Piscitelli Mansur, A. A.; Do Nascimento, O. L.; Mansur, H. S., Enhancing polymer modified mortar adhesion to ceramic tile surface by chemical functionalization with organosilanes Surf. Rev. Lett. 2009, 16 (1), 127 139. 142. (a) Yousuf, M.; Mollah, A.; He ss, T. R.; Tsai, Y. N.; Cocke, D. L., An FTIR and XPS investigations of the effects of carbonation on the solidification/stabilization of cement based systems Portland type V with zinc. Cem. Concr. Res. 1993, 23 (4), 773 784; (b) Cocke, D. L.; Mollah, M. Y A.; Parga, J. R.; Hess, T. R.; Ortego, J. D., An XPS and SEM/EDS characterization of leaching effects on lead and zinc doped portland cement. Journal of Hazardous Materials 1992, 30 (1), 83 95; (c) Kallury, K. M. R.; Brennan, J. D.; Krull, U. J., Depth profiling of functionalized silane films on quartz and silicon substrates and of urease immobilized on such films by angle resolved X ray photoelectron spectroscopy. Anal. Chem. 1995, 67 (15). 143. (a) Black, L.; Garbev, K.; Stemmermann, P.; Hallam, K. R. ; Allen, G. C., Characterisation of crystalline C S H phases by X ray photoelectron spectroscopy. Cem. Concr. Res. 2003, 33 (6), 899 911; (b) Black, L.; Garbev, K.; Beuchle, G.; Stemmermann, P.; Schild, D., X ray photoelectron spectroscopic
169 investigation o f nanocrystalline calcium silicate hydrates synthesised by reactive milling. Cem. Concr. Res. 2006, 36 (6), 1023 1031. 144. Nguyen, T. B., W. E.; Alsheh, D.; Aouadi, K.; and Chin, J. W. In Water at the Polymer/Substrate Interface and Its Role in the Durab ility of Polymer/Glass Fiber Composites Durability of Fibre Reinforced Polymer (FRP) Composites for Construction (CDCC' 98), Canada, Benmokrane, B. R., H., Ed. Canada, 1998; pp 451 462. 145. Ji, W. G.; Hu, J. M.; Zhang, J. Q.; Cao, C. N., Reducing the wa ter absorption in epoxy coatings by silane monomer incorporation. Corrosion Science 2006, 48 (11), 3731 3739. 146. Brun, E.; Rain, P.; Teissdre, G.; Guillermin, C.; Rowe, S. In Hygrothermal aging of a filled epoxy resin 2007 International Conference on Solid Dielectrics, Winchester,UK, Winchester,UK, 2007; pp 239 242.
170 BIOGRAPHICAL SKETCH Andrew David Duncan Stewart was born to Martin Stewart and Dorothy Nygren in Chicago, Illinois in 1985. In August of 2003, he began attending the University of Illi nois at Champaign Urbana. He worked as an undergraduate researcher under Dr. Paul science and engineering in 2007. In August of 2007, he traveled to the University of Flor ida to pursue a graduate degree in materials science and engineering under the direction of Dr. Elliot Douglas.