Improved resilience of the U.S. civil infrastructure requires better knowledge of the performance of concrete structural components. This could be enabled by providing the concrete industry with tools to cope with and predict the possible interactions among the many inorganic and organic components that can influence setting time, strength development, and contributors to “infant mortality” of concrete. These tools are not available because of significant gaps in data and knowledge about the fundamental reactivity and mutual interactions of the components in water. This project will address those knowledge gaps by using both new and established characterization methods to obtain complementary measurements of the reaction rates of cementitious components. Kinetic Monte Carlo and cellular automaton models will be tested against these measurements, and computational fluid dynamics models will be used to better understand the compositional dynamics near the liquid-solid interface.
Objective - Generate and publish reaction rate data and enhanced computer modeling tools to better understand and predict the rates of microstructure development and phase interactions in portland cement concrete binders.
What is the new technical idea? Reflection digital holographic microscopy (DHM) has demonstrated that the dissolution rates of cement component minerals are sensitively dependent on the type and density of structural defects at the reacting surfaces. This means that the shape and size distribution of powder particles in cement should have an even greater influence on cement reactivity than has been anticipated previously. This project will continue the fundamental measurements on crystal surfaces by DHM and will study reaction rates in single-component and multi-component suspensions. However, extracting meaningful reaction rate measurements from suspensions is complicated by (1) the more complex surface geometry, (2) the possibility of multiple simultaneous reactions, and (3) the inability to directly visualize the microstructure in situ. A rapid flow-through powder bed vacuum filtration system will be built to cope with these complications, whereby a liquid of known composition can be forced quickly through a thin layer of powder before any secondary components can precipitate.
Analysis of the effluent solution by inductively coupled plasma optical emission spectrometry (ICP-OES), and of the powder bed by quantitative X-ray diffraction, will indicate the extent of reactions as the liquid flows through the capillary porosity. Computational fluid dynamics (CFD) simulations of the experiments can help determine the flow patterns and residence time distribution of liquid as it flows through the powder bed, which will help make sense of the effluent composition changes and the influence of particle size distribution on reaction rates in solution. The measurements must be interpreted in terms of molecular mechanisms to establish, and this will be aided by Kinetic Monte Carlo (KMC) simulations of individual particles in water.
What is the research plan? Powders of each of the four major clinker phases, as well important secondary phases such as gypsum and calcite, have already been obtained. The powders must be thoroughly characterized to be able to draw structure-property relationships from the experiments and simulations. This includes measurement of particle size distribution, specific surface area, and phase purity. In addition, the particle shape distribution will also be measured by X-ray microtomography in collaboration with Ed Garboczi at the NIST Boulder campus. At the start of the project, effort will be devoted to constructing and testing devices for measuring powder dissolution rates and to establishing procedures for using them. As this is happening, CFD simulations of flow past reacting surfaces of varying roughness will be undertaken to help understand and interpret DHM measurements made in FY17 of the dissolution of gypsum, calcite, and tricalcium aluminate (C3A). Kinetic Monte Carlo models of dissolution will be developed and tested on simple minerals such as calcite and gypsum, for which experimental data have already been obtained.
Mineral powders will be size classified, and each size class will be subjected in isolation and in selected combinations to dissolution measurements in pure water and in electrolytes. Measurements will be made using at least three different temperatures to obtain apparent activation enthalpies for the reaction. The powders’ size distribution, specific surface area, and phase composition will be measured after the experiments to correlate solution compositions with particle erosion rates. Selected systems will be analyzed before and after by high-resolution X-ray microtomography to provide additional validation points for 3D microstructure development models such as HydratiCA and THAMES. Individual powder particles will also be observed in situ with polarized light microscopy to characterize changes in shape during reaction. The KMC models will be used to simulate retreat of nominally flat but structurally defective mineral surfaces, from which correlations will be made between reaction rate spectra and defect type and distribution.
Both KMC and cellular automata (HydratiCA) models will be tested against measurements of powder reactivity as a function of particle size. Results of this testing are expected to reveal general relationships between size and reactivity that can be embedded within coarser microstructure models for improved capabilities in predicting cement binder microstructure development. Further experimental data sets will be generated with mixtures that more closely approximate portland cement compositions, monitoring the time-dependent phase development by in situ XRD for at least 24 h, repeated at three different temperatures. These datasets will also be made publicly available and will form excellent, well-characterized systems for direct testing of hydration models.