A Deep Learning Approach to Design and Discover Sustainable Cementitious Binders: Strategies to Learn From Small Databases and Develop Closed-form Analytical Models

To reduce the energy-intensity and carbon footprint of Portland cement (PC), the prevailing practice embraced by concrete technologists is to partially replace the PC in concrete with supplementary cementitious materials [SCMs: geological materials (e.g., limestone); industrial by-products (e.g., fly ash); and processed materials (e.g., calcined clay)]. Chemistry and content of the SCM profoundly affect PC hydration kinetics; which, in turn, dictates the evolutions of microstructure and properties of the [PC + SCM] binder. Owing to the substantial diversity in SCMs’ compositions–plus the massive combinatorial spaces, and the highly nonlinear and mutually-interacting processes that arise from SCM-PC interactions–state-of-the-art computational models are unable to produce a priori predictions of hydration kinetics or properties of [PC + SCM] binders. In the past 2 decades, the combination of Big data and machine learning (ML)—commonly referred to as the fourth paradigm of science–has emerged as a promising approach to learn composition-property correlations in materials (e.g., concrete), and capitalize on such learnings to produce a priori predictions of properties of materials with new compositions. Notwithstanding these merits, widespread use of ML models is hindered because they: 1) Require Big data to learn composition-property correlations, and, in general, large databases for concrete are not publicly available; and 2) Function as black-boxes, thus providing little-to-no insights into the materials laws like theory-based analytical models do. This study presents a deep learning (DL) model capable of producing a priori, high-fidelity predictions of composition- and time-dependent hydration kinetics and phase assemblage development in [PC + SCM] pastes. The DL is coupled with: 1) A fast Fourier transformation algorithm that reduces the dimensionality of training datasets (e.g., kinetic datasets), thus allowing the model to learn intrinsic composition-property correlations from a small database; and 2) A thermodynamic model that constrains the model, thus ensuring that predictions do not violate fundamental materials laws. The training and outcomes of the DL are ultimately leveraged to develop a simple, easy-to-use, closed-form analytical model capable of predicting hydration kinetics and phase assemblage development in [PC + SCM] pastes, using their initial composition and mixture design as inputs.

Influence of Alkalis on Natural Carbonation of Limestone Calcined Clay Cement Pastes

Vulnerability to atmospheric carbonation is one of the major durability concerns for limestone calcined clay cement (LC3) concrete due to its relatively low overall alkalinity. In this study, the natural carbonation behaviors of ternary ordinary Portland cement-metakaolin-limestone (OPC-MK-LS) blends containing various sulfate salts (i.e., anhydrous CaSO4, Na2SO4, and K2SO4) are studied, with the aim of revealing the influence of alkali cations (Na+, K+). Detailed analyses on the hydrated phase assemblage, composition, microstructure, and pore structure of LC3 pastes prior to and post indoor carbonation are conducted. The results show that the incorporation of sulfate salts accelerates the setting and strength gain of LC3 pastes, likely through enhancement of ettringite formation, but undermines its later age strength achievement due to the deleterious effect of alkali cations (Na+, K+) on late age OPC hydration. The carbonation resistance of LC3 systems is considerably undermined, particularly with the incorporation of Na2SO4 or K2SO4 salts, due to the simultaneous pore coarsening effect and reduced CO2-binding capacity. The carbonation-induced phase and microstructural alterations of LC3 pastes are discussed and compared with those of reference OPC pastes.

Effect of Boron and Water-to-Cement Ratio on the Performances of Laboratory Prepared Belite-Ye’elimite-Ferrite (BYF) Cements

The effect of superplasticiser, borax and the water-to-cement ratio on BYF hydration and mechanical strengths has been studied. Two laboratory-scale BYF cements—st-BYF (with β-C2S and orthorhombic C4A3S¯) and borax-activated B-BYF (with α’H-C2S and pseudo-cubic C4A3S¯)—have been used, and both show similar particle size distribution. The addition of superplasticiser and externally added borax to BYF pastes has been optimised through rheological measurements. Optimised superplasticiser contents (0.3, 0.4 and 0.1 wt % for st-BYF, B-BYF and st-BYF with externally added 0.25 wt % B2O3, respectively) result in low viscosities yielding homogeneous mortars. The calorimetric study revealed that st-BYF is more reactive than B-BYF, as the values of heat released are 300–370 J/g and 190–210 J/g, respectively, after 7 days of hydration; this fact is independent of the water-to-cement ratio. These findings agree with the higher degree of hydration at 28 days of β-C2S in st-BYF (from 45 to 60%) than α’H-C2S in B-BYF (~20 to 30%). The phase assemblage evolution has been determined by LXRPD coupled with the Rietveld method and MAS-NMR. The formation of stratlingite is favoured by increasing the w/c ratio in both systems. Finally, the optimisation of fresh BYF pastes jointly with the reduction of water-to-cement ratio to 0.40 have allowed the achieving of mortars with compressive strengths over 40 MPa at 7 days in all systems. Moreover, the st-BYF mortar, where borax was externally added, achieved more than 70 MPa after 28 days. The main conclusion of this work does not support Lafarge’s approach of adding boron/borax to the raw meal of BYF cements. This procedure stabilises the alpha belite polymorph, but its reactivity, in these systems, is lower and the associated mechanical strengths poorer.

Property Analysis of Slag Composite Concrete Using a Kinetic–Thermodynamic Hydration Model

Slag is increasingly unitized for the production of sustainable concrete. This paper presents a procedure with which to analyze the property development of slag composite concrete. Experimental studies of the hydration heat and compressive strength development and simulation studies using a kinetic hydration model and a thermodynamic model were performed. First, we performed an experimental study of the isothermal hydration heat of cement–slag blends. Based on the results of the experimental study on cumulative hydration heat, the reaction degree of slag was determined. We found that the reaction degree of slag decreased as the slag content increased. Second, the reaction degree of slag and cement were used as the input parameters for the Gibbs energy minimization (GEM) thermodynamic equilibrium model. Moreover, the phase assemblage of hydrating cement–slag was determined. The trends of calcium silicate hydrate (CSH) are similar to those of strength. Based on the CSH content, the strength of hardening cement–slag blends was determined. In addition, the calcium hydroxide (CH) content resulting from the thermodynamic model shows good agreement with the experimental results. In summary, the integrated kinetic–thermodynamic model is useful for analyzing the properties of cement–slag blends.

X-ray Total Scattering Study of Phases Formed from Cement Phases Carbonation

Carbonation in cement binders has to be thoroughly understood because it affects phase assemblage, binder microstructure and durability performance of concretes. This is still not the case as the reaction products can be crystalline, nanocrystalline and amorphous. The characterisation of the last two types of components are quite challenging. Here, carbonation reactions have been studied in alite-, belite- and ye’elimite-containing pastes, in controlled conditions (3% CO2 and RH = 65%). Pair distribution function (PDF) jointly with Rietveld and thermal analyses have been applied to prove that ettringite decomposed to yield crystalline aragonite, bassanite and nano-gibbsite without any formation of amorphous calcium carbonate. The particle size of gibbsite under these conditions was found to be larger (~5 nm) than that coming from the direct hydration of ye’elimite with anhydrite (~3 nm). Moreover, the carbonation of mixtures of C-S-H gel and portlandite, from alite and belite hydration, led to the formation of the three crystalline CaCO3 polymorphs (calcite, aragonite and vaterite), amorphous silica gel and amorphous calcium carbonate. In addition to their PDF profiles, the thermal analyses traces are thoroughly analysed and discussed.