Machine Learning on Microstructural Chemical Maps to Classify Component Phases in Cement Pastes

Document Type : Regular Article


1 Graduate student, School of Sustainable Engineering and the Built Environment, Arizona State University, Tempe AZ 85287, USA

2 Intel Corporation, Chandler, AZ 85224, USA

3 Adjunct Faculty, Computer Science and Engineering, Arizona State University, Tempe AZ 85287, USA

4 Professor, School of Sustainable Engineering and the Built Environment, Arizona State University, Tempe AZ 85287, USA


This paper implements machine learning (ML) classification algorithms on microstructural chemical maps to predict the constituent phases. Intensities of chemical species (Ca, Al, Si, etc.), and in some cases the nanomechanical properties measured at the corresponding points, form the input to the ML model, which predicts the phase label (LD or HD C-S-H, clinker etc.) belonging to that location. Artificial neural networks (ANN) and forest ensemble methods are used for classification. Confusion matrices and receiver-operator characteristic (ROC) curves are used to analyze the classification efficiency. It is shown that, for complex microstructures such as those of ultra-high performance (UHP) pastes, the classifier performs well when nanomechanical information augments the chemical intensity data. For simpler systems such as well-hydrated plain cement pastes, the classifier accurately predicts the phase label from the intensities of Ca, Al, and Si alone. The work enables fast-and-efficient phase identification and property forecasting from microstructural chemical maps.


Main Subjects

[1]     Suda Y, Saeki T, Saito T. Relation between chemical composition and physical properties of CSH generated from cementitious materials. J Adv Concr Technol 2015;13:275–90.
[2]     Manzano H, Dolado JS, Ayuela A. Elastic properties of the main species present in Portland cement pastes. Acta Mater 2009;57:1666–74.
[3]     Wang D, Shi C, Farzadnia N, Shi Z, Jia H. A review on effects of limestone powder on the properties of concrete. Constr Build Mater 2018;192:153–66.
[4]     Mondal P, Shah SP, Marks LD, Gaitero JJ. Comparative Study of the Effects of Microsilica and Nanosilica in Concrete. Transp Res Rec J Transp Res Board 2010;2141:6–9.
[5]     Chen JJ, Sorelli L, Vandamme M, Ulm F-J, Chanvillard G. A Coupled Nanoindentation/SEM-EDS Study on Low Water/Cement Ratio Portland Cement Paste: Evidence for C-S-H/Ca(OH) 2 Nanocomposites. J Am Ceram Soc 2010.
[6]     Vance K, Aguayo M, Oey T, Sant G, Neithalath N. Hydration and strength development in ternary portland cement blends containing limestone and fly ash or metakaolin. Cem Concr Compos 2013;39:93–103.
[7]     Yu R, Spiesz P, Brouwers HJH. Effect of nano-silica on the hydration and microstructure development of Ultra-High Performance Concrete (UHPC) with a low binder amount. Constr Build Mater 2014;65:140–50.
[8]     Huang W, Kazemi-Kamyab H, Sun W, Scrivener K. Effect of cement substitution by limestone on the hydration and microstructural development of ultra-high performance concrete (UHPC). Cem Concr Compos 2017;77:86–101.
[9]     Stutzman P. Scanning electron microscopy imaging of hydraulic cement microstructure. Cem Concr Compos 2004;26:957–66.
[10]   Durdziński PT, Dunant CF, Haha M Ben, Scrivener KL. A new quantification method based on SEM-EDS to assess fly ash composition and study the reaction of its individual components in hydrating cement paste. Cem Concr Res 2015;73:111–22.
[11]   Rößler C, Steiniger F, Ludwig H-M. Characterization of C-S-H and C-A-S-H phases by electron microscopy imaging, diffraction, and energy dispersive X-ray spectroscopy. J Am Ceram Soc 2017;100:1733–42.
[12]   Rossen JE, Scrivener KL. Optimization of SEM-EDS to determine the C–A–S–H composition in matured cement paste samples. Mater Charact 2017;123:294–306.
[13]   Němeček J, Šmilauer V, Kopecký L. Nanoindentation characteristics of alkali-activated aluminosilicate materials. Cem Concr Compos 2011;33:163–70.
[14]   Hu C, Li Z. Property investigation of individual phases in cementitious composites containing silica fume and fly ash. Cem Concr Compos 2015;57:17–26.
[15]   Ford E, Arora A, Mobasher B, Hoover CG, Neithalath N. Elucidating the nano-mechanical behavior of multi-component binders for ultra-high performance concrete. Constr Build Mater 2020;243:118214.
[16]   Ford EL, Hoover CG, Mobasher B, Neithalath N. Relating the nano-mechanical response and qualitative chemical maps of multi-component ultra-high performance cementitious binders. Constr Build Mater 2020;260:119959.
[17]   Wilson W, Sorelli L, Tagnit-Hamou A. Unveiling micro-chemo-mechanical properties of C–(A)–S–H and other phases in blended-cement pastes. Cem Concr Res 2018;107:317–36.
[18]   J. Krakowiak K, Wilson W, James S, Musso S, Ulm F-J. Inference of the phase-to-mechanical property link via coupled X-ray spectrometry and indentation analysis: Application to cement-based materials. Cem Concr Res 2015;67:271–85.
[19]   Hu C, Li Z. A review on the mechanical properties of cement-based materials measured by nanoindentation. Constr Build Mater 2015;90:80–90.
[20]   Konstantopoulos G, Koumoulos EP, Charitidis CA. Testing Novel Portland Cement Formulations with Carbon Nanotubes and Intrinsic Properties Revelation: Nanoindentation Analysis with Machine Learning on Microstructure Identification. Nanomaterials 2020;10:645.
[21]   Gao X, Wei Y, Huang W. Effect of individual phases on multiscale modeling mechanical properties of hardened cement paste. Constr Build Mater 2017;153:25–35.
[22]   Brown L, Allison PG, Sanchez F. Use of nanoindentation phase characterization and homogenization to estimate the elastic modulus of heterogeneously decalcified cement pastes. Mater Des 2018;142:308–18.
[23]   Young BA, Hall A, Pilon L, Gupta P, Sant G. Can the compressive strength of concrete be estimated from knowledge of the mixture proportions?: New insights from statistical analysis and machine learning methods. Cem Concr Res 2019;115:379–88.
[24]   DeRousseau MA, Laftchiev E, Kasprzyk JR, Rajagopalan B, Srubar WV. A comparison of machine learning methods for predicting the compressive strength of field-placed concrete. Constr Build Mater 2019;228:116661.
[25]   Chou J-S, Chiu C-K, Farfoura M, Al-Taharwa I. Optimizing the Prediction Accuracy of Concrete Compressive Strength Based on a Comparison of Data-Mining Techniques. J Comput Civ Eng 2011;25:242–53.
[26]   Dao D Van, Adeli H, Ly H-B, Le LM, Le VM, Le T-T, et al. A Sensitivity and Robustness Analysis of GPR and ANN for High-Performance Concrete Compressive Strength Prediction Using a Monte Carlo Simulation. Sustainability 2020;12:830.
[27]   Nguyen KT, Nguyen QD, Le TA, Shin J, Lee K. Analyzing the compressive strength of green fly ash based geopolymer concrete using experiment and machine learning approaches. Constr Build Mater 2020;247:118581.
[28]   Koumoulos EP, Paraskevoudis K, Charitidis CA. Constituents Phase Reconstruction through Applied Machine Learning in Nanoindentation Mapping Data of Mortar Surface. J Compos Sci 2019;3:63.
[29]   Ford E, Kailas S, Maneparambil K, Neithalath N. Machine learning approaches to predict the micromechanical properties of cementitious hydration phases from microstructural chemical maps. Constr Build Mater 2020;265:120647.
[30]   Arora A, Aguayo M, Hansen H, Castro C, Federspiel E, Mobasher B, et al. Microstructural packing- and rheology-based binder selection and characterization for Ultra-high Performance Concrete (UHPC). Cem Concr Res 2018;103:179–90.
[31]   Arora A, Almujaddidi A, Kianmofrad F, Mobasher B, Neithalath N. Material design of economical ultra-high performance concrete (UHPC) and evaluation of their properties. Cem Concr Compos 2019;104:103346.
[32]   Wilson W, Sorelli L, Tagnit-Hamou A. Automated coupling of NanoIndentation and Quantitative Energy-Dispersive Spectroscopy (NI-QEDS): A comprehensive method to disclose the micro-chemo-mechanical properties of cement pastes. Cem Concr Res 2018;103:49–65.
[33]   Wilson W, Rivera-Torres JM, Sorelli L, Durán-Herrera A, Tagnit-Hamou A. The micromechanical signature of high-volume natural pozzolan concrete by combined statistical nanoindentation and SEM-EDS analyses. Cem Concr Res 2017;91:1–12.
[34]   Hoover CG, Ulm F-J. Experimental chemo-mechanics of early-age fracture properties of cement paste. Cem Concr Res 2015;75:42–52.
[35]   Sorelli L, Constantinides G, Ulm F-J, Toutlemonde F. The nano-mechanical signature of Ultra High Performance Concrete by statistical nanoindentation techniques. Cem Concr Res 2008;38:1447–56.
[36]   Stutzman PE, Clifton JR. Specimen preparation for scanning electron microscopy. Proc. Int. Conf. Cem. Microsc., vol. 21, International Cement Microscopy Association; 1999, p. 10–22.
[37]   Stutzman PE. Microscopy of Clinker and Hydraulic Cements. Rev Mineral Geochemistry 2012;74:101–46.
[38]   Ulm F-J, Vandamme M, Jennings HM, Vanzo J, Bentivegna M, Krakowiak KJ, et al. Does microstructure matter for statistical nanoindentation techniques? Cem Concr Compos 2010;32:92–9.
[39]   da Silva WRL, Němeček J, Štemberk P. Methodology for nanoindentation-assisted prediction of macroscale elastic properties of high performance cementitious composites. Cem Concr Compos 2014;45:57–68.
[40]   Oliver WC, Pharr GM. An improved technique for determining hardness and elastic modulus using load and displacement sensing indentation experiments. J Mater Res 1992;7:1564–83.
[41]   Oliver WC, Pharr GM. Measurement of hardness and elastic modulus by instrumented indentation: Advances in understanding and refinements to methodology. J Mater Res 2004;19:3–20.
[42]   Stutzman PE, Bullard JF, Feng P, Stutzman PE. Quantitative imaging of clinker and cement microstructure. US Department of Commerce, National Institute of Standards and Technology; 2016.
[43]   Lura P, Trtik P, Münch B. Validity of recent approaches for statistical nanoindentation of cement pastes. Cem Concr Compos 2011;33:457–65.
[44]   Haecker C-J, Garboczi EJ, Bullard JW, Bohn RB, Sun Z, Shah SP, et al. Modeling the linear elastic properties of Portland cement paste. Cem Concr Res 2005;35:1948–60.
[45]   Puerta‐Falla G, Balonis M, Le Saout G, Falzone G, Zhang C, Neithalath N, et al. Elucidating the Role of the Aluminous Source on Limestone Reactivity in Cementitious Materials. J Am Ceram Soc 2015;98:4076–89.
[46]   Li X, Liu Z, Cui S, Luo C, Li C, Zhuang Z. Predicting the effective mechanical property of heterogeneous materials by image based modeling and deep learning. Comput Methods Appl Mech Eng 2019;347:735–53.
[47]   Zhang J, Huang Y, Wang Y, Ma G. Multi-objective optimization of concrete mixture proportions using machine learning and metaheuristic algorithms. Constr Build Mater 2020;253:119208.
[48]   Konstantopoulos G, Koumoulos EP, Charitidis CA. Classification of mechanism of reinforcement in the fiber-matrix interface: Application of Machine Learning on nanoindentation data. Mater Des 2020;192:108705.
[49]   Tieleman T, Hinton G. Lecture 6.5-rmsprop: Divide the gradient by a running average of its recent magnitude. COURSERA Neural Networks Mach Learn 2012;4:26–31.
[50]   Chollet F. Keras, GitHub repository, 2015. [Online]. Available: n.d.
[51]   Srivastava N, Hinton G, Krizhevsky A, Sutskever I, Salakhutdinov R. Dropout: a simple way to prevent neural networks from overfitting. J Mach Learn Res 2014;15:1929–58.
[52]   Oey T, Jones S, Bullard JW, Sant G. Machine learning can predict setting behavior and strength evolution of hydrating cement systems. J Am Ceram Soc 2020;103:480–90.
[53]   Huang BFF, Boutros PC. The parameter sensitivity of random forests. BMC Bioinformatics 2016;17:331.
[54]   Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, et al. Scikit-learn: Machine learning in Python. J Mach Learn Res 2011;12:2825–30.
[55]   Chen T, Guestrin C. XGBoost: A Scalable Tree Boosting System. Proc. 22nd ACM SIGKDD Int. Conf. Knowl. Discov. Data Min., New York, NY, USA: ACM; 2016, p. 785–94.
[56]   Tharwat A. Classification assessment methods. Appl Comput Informatics 2021;17:168–92.
[57]   Fawcett T. An introduction to ROC analysis. Pattern Recognit Lett 2006;27:861–74.
[58]   Hand DJ, Till RJ. A simple generalisation of the area under the ROC curve for multiple class classification problems. Mach Learn 2001;45:171–86.
[59]   Su C, Ju S, Liu Y, Yu Z. Improving Random Forest and Rotation Forest for highly imbalanced datasets. Intell Data Anal 2015;19:1409–32.
[60]   Bergstra J, Bengio Y. Random search for hyper-parameter optimization. J Mach Learn Res 2012;13.
[61]   Vandamme M, Ulm F-J, Fonollosa P. Nanogranular packing of C–S–H at substochiometric conditions. Cem Concr Res 2010;40:14–26.
[62]   Yu Z, Zhou A, Lau D. Mesoscopic packing of disk-like building blocks in calcium silicate hydrate. Sci Rep 2016;6:36967.
[63]   Wei Y, Gao X, Liang S. A combined SPM/NI/EDS method to quantify properties of inner and outer C-S-H in OPC and slag-blended cement pastes. Cem Concr Compos 2018;85:56–66.
[64]   Kim H, Ahn E, Cho S, Shin M, Sim S-H. Comparative analysis of image binarization methods for crack identification in concrete structures. Cem Concr Res 2017;99:53–61.
[65]   Moon J, Yoon S, Wentzcovitch RM, Monteiro PJM. First-principles elasticity of monocarboaluminate hydrates. Am Mineral 2014;99:1360–8.