ICASCM25 - 20 years of atomistic simulation in cement

Hegoi Manzano Department of Physics Faculty of Science and Engineering University of the Basque Country UPV/EHU 20 years of atomistic simulation in cement 1 20 years of research 2 Started my phD 10 Oct. 2005 Poster presentation at the Trends in Nanotechnology conference in Grenoble 2006 Before 2005: early studies by the clay community 3 Urbana Champaine Kirkpatrik & Kalinichev France Faucon & Nonat, ! R.Pellenq, Labbez First MD studies on tobermorite, ettingite, protlandite… Primitive model to investigate cohesive forces 2005 to 2012: specialised studies 4 Vandervilt Florence Sanchez PSI (Switzerlad) Churakov Basque Country Dolado, Ayuela, Manzano MIT CSHub Ulm, Pellenq, et al. (also me) Melbourne White and Provis First “realistic” C-S-H model Disseminate simulations First determination of elastic properties First calculations on clinker phases First phD thesis First calculations on SCM First calculations on hibrid systems Atomistic Simulation Atomistic Simulation studies of the studies of the Cement Paste Cement Paste Components Components Hegoi Manzano PhD Thesis, 2009 Atomistic Simulation studies of the Cement Paste Components Atomistic Simulation studies of the Cement Paste Components Hegoi Manzano PhD Thesis , 2009 Hegoi Manzano PhD Thesis, 2009 France Labbez 2012 to 2017: generalization of the technique 5 Basque Country Duque-Redondo, Manzano Rice Shasavari Switzerlad PSI, EPFL, ETH Churakov, Flatt Mishra ! Bowen, Galmarini, Kunhi UCLA, UC Irvine Berkley Princeton White Darmstadt Ukrainczyk and Koenders Qingdao Hou Bauchy ! MJA Qomi! Monteiro et al.! MIT CSHub Ulm, Pellenq France Labbez Nanjing Wang, South East U 2018 to present: the BOOM! 6 Switzerlad UC Irvine Berkley Qingdao Princeton Darmstadt Nanjing Basque Country Shangai Tokio Macao Hong Kong Singapour Chenai Harbin Wuhan Melbourne Hokkaido Dehli Rice France 20 years of research 7 20 years of research 8 Kinetic Monte Carlo Lattice Boltzmann Cellular automata Simulated time System size 1Å 1nm 1μm fs ps ns μs Classic (MD &MC) 100nm s… 10nm Coarse Grained Potentials BO force fields ReaxFF DFT TD-DFT Classical force fields ab-initio Meso-scale DFTB Machine Learning Advanced sampling Evolutionary Algorithms The present of atomistic simulations in cement 9 “We can go beyond laboratory limitations, do thousands of simulations to test materials and calculate properties, and design cement from electrons to buildings” My 2005 self Calcium Silicate Hydrate: interfaces 16 Duque-Redondo, E., Masoero, E., & Manzano, H. (2022). Nanoscale shear cohesion between cement hydrates: The role of water diffusivity under structural and electrostatic confinement. Cement and Concrete Research, 154, 106716. Duque-Redondo, E., Masoero, E., & Manzano, H. (2022). Nanoscale shear cohesion between cement hydrates: The role of water diffusivity under structural and electrostatic confinement. Cement and Concrete Research, 154, 106716. Calcium Silicate Hydrate: interfaces 17 slip-stick mechanism Duque-Redondo, E., Masoero, E., & Manzano, H. (2022). Nanoscale shear cohesion between cement hydrates: The role of water diffusivity under structural and electrostatic confinement. Cement and Concrete Research, 154, 106716. Duque-Redondo, E., Masoero, E., & Manzano, H. (2022). Nanoscale shear cohesion between cement hydrates: The role of water diffusivity under structural and electrostatic confinement. Cement and Concrete Research, 154, 106716. Calcium Silicate Hydrate: interfaces 18 What changes at the interface? Calcium Silicate Hydrate: interfaces 19 What changes at the interface? Cohesion lost when water mobility increases Calcium Silicate Hydrate: interfaces 20 What changes at the interface? Control water mobility to test the mechanism Cohesion lost when water mobility increases Calcium Silicate Hydrate: interfaces 21 What changes at the interface? Control water mobility to test the mechanism Electrostatic confinement controls cohesion Cohesion lost when water mobility increases Calcium Silicate Hydrate: Microstructure Umar Hayat, Eduardo Duque-Redondo, Ming-Feng Kai, Hegoi Manzano, Muhammad Riaz Ahmad, You Dong, Jian-Guo Dai, Desorption of water from aqueous solution confined in CSH gel pore: A molecular dynamics study Construction and Building Materials 490, 142602 The Ca/Si ratio does not have significant impact 22 Calcium Silicate Hydrate: Microstructure Umar Hayat, Eduardo Duque-Redondo, Ming-Feng Kai, Hegoi Manzano, Muhammad Riaz Ahmad, You Dong, Jian-Guo Dai, Desorption of water from aqueous solution confined in CSH gel pore: A molecular dynamics study Construction and Building Materials 490, 142602 The Ca/Si ratio does not have significant impact 23 The Ca/Si ratio of the bulk is not be the key parameter The present of atomistic simulations in cement 24 “We can go beyond laboratory limitations, do thousands of simulations to test materials and calculate properties, and design cement from electrons to buildings” My 2005 self “In practice, the number of atomistic simulation studies that truly imply a practical advance or guide the design of cement towards enhanced performance is limited” 20 years later Duque-Redondo, E., de Souza, F. B., Geng, G., & Manzano, H. (2026). A critical review and perspectives on atomistic models of non-crystalline cementitious materials. Cement and Concrete Research, 199, 108067. Intrinsic scale limitations, lack of contextualization, incomplete models, lack of experimental focus,… Kaolinite to Metakaolin Adapted from: Cheng, et al. (2019). Dehydroxylation and structural distortion of kaolinite as a hightemperature sorbent in the furnace. Minerals, 9(10), 587 Zunino, F., & Scrivener, K. (2022). Oxidation of pyrite (FeS 2) and troilite (FeS) impurities in kaolinitic clays after calcination. Materials and Structures Zunino, F., & Scrivener, K. (2024). Reactivity of kaolinitic clays calcined in the 650 C–1050 C temperature range: Towards a robust assessment of overcalcination. Cement and Concrete Composites, 25 Metakaolin: work in progress Melting - quenching Al(VI) Al(IV) O-H Dehydroxylation protocols 32 Al(V) Same local structure Metakaolin: work in progress Melting - quenching Al(IV) Al(V) Al(VI) Dehydroxylation protocols 33 Metakaolin: work in progress Melting - quenching Al(IV) Al(V) Al(VI) Amorphization Change in Al coordination Dehydroxylation protocols 34 The present of atomistic simulations in cement 35 “We can go beyond laboratory limitations, do thousands of simulations to test materials and calculate properties, and design cement from electrons to buildings” My 2005 self “In practice, the number of atomistic simulation studies that truly imply a practical advance or guide the design of cement towards enhanced performance is limited” 20 years later Duque-Redondo, E., de Souza, F. B., Geng, G., & Manzano, H. (2026). A critical review and perspectives on atomistic models of non-crystalline cementitious materials. Cement and Concrete Research, 199, 108067. Intrinsic scale limitations, lack of contextualization, incomplete models, lack of experimental focus,… Case of study: belitic cements 36 C2S ( rdiss = 18!μmol!m−2!s−1 ) β C2S ( rdiss = 3x10-4!μmol!m−2!s−1) γ α α′ H α′ L β γ 1425ºC 1160ºC 630-680ºC <500ºC 690ºC 780 - 860ºC X αC2SH ~450ºC M. Miyazaki, et al. ‘‘Crystallographic Data of a New Phase of a Dicalcium Silicate,’’ J. Am. Ceram. Soc. (1998). Dicalcium Silicate: searching for new polymorphs 37 18K 315 12 α′ H Remove duplicates 9K Remove energies above alpha Thermodynamic stability (phonons) 122 Temperature annealing New dicalcium silicate metastable polymorphs Computational screening López-Zorrilla, J., Aretxabaleta, X. M., & Manzano, H. (2024). Exploring the polymorphism of dicalcium silicates using transfer learning enhanced machine learning atomic potentials. Dicalcium Silicate: searching for new polymorphs 38 López-Zorrilla, J., Aretxabaleta, X. M., & Manzano, H. (2024). Exploring the polymorphism of dicalcium silicates using transfer learning enhanced machine learning atomic potentials. The amorphous phases are likely to be the reactive ones A bright future ahead! 39 Better models! Better methods! Focused studies! Increasing contextualization and interaction with experiments Fly ashes and slags: current models Al/Si T-O-T network Ca-Mg network modifiersFull slag model ~70,000 atoms! 70 substitions! Molecular Dynamics + Reverse Monte Carlo Qi Zhai, Macro Bertani, Hegoi Manzano, Takayasu Ito, Koji Ohara, Kiyofumi Kurumisawa, The changes in the reactivity of synthetic aluminate silicate-based slag in alkaline environment induced by minor components, under review 40 Machine Learning potentials 41 López-Zorrilla, J., Aretxabaleta, X. M., & Manzano, H. (2024). Exploring the polymorphism of dicalcium silicates using transfer learning enhanced machine learning atomic potentials. 20K calculations; DFT @ 10h ~ 23 years CPU time Transfer Learning of ænet MLP 20K calculations; MLP @ 10’ ~ 0.4 years CPU time MLP - direct MLP -transfer ReaxFF