(799f) A Novel Computer-Aided Molecular Design Approach to Design New Non-Intuitive Shrinkage Reducing Admixtures (SRAs) for Cement: A Modeling and Experimental Study for Improved Performance

A ceramic has traditionally been defined as an inorganic, nonmetallic solid that is prepared from powdered materials, is fabricated into products through the application of heat, and displays such characteristic properties as hardness, strength, low electrical conductivity, and brittleness.[*] Cementitious materials are among the broad categories that make up the ceramic industry. Portland cement concrete is used in the construction of most buildings, foundations, bridge decks, and pavements. The desired performance is not always achieved, however. Some structures develop early-age cracks caused by the shrinkage of cementitious materials, associated with the hydration of cement, drying of pore solution, and water movement in the cement pores of the structure. Accordingly, such shrinkage cracks become pathways into the concrete for water and harmful water-borne chemicals and can lead to further damage within the structure. Shrinkage reducing admixtures (SRAs) are one way to mitigate shrinkage cracking in cementitious materials by reducing both drying and autogenous shrinkage.

The development of new admixtures for concrete is normally an experimental approach where one uses prior knowledge to suggest potential effective compounds. This approach is time-consuming, incremental and typically expensive. To address that, we propose a computer-aided molecular design (CAMD) algorithm that uses an inverse quantitative structure property relationship (I-QSPR) as a powerful alternative. Such a technique applies information about how known substances behave in a given application, deconstructs molecular structure into fragments (Signatures), correlates Signatures to its performance, and then identifies new molecules with optimal predicted performance. This contribution outlines the methodology for using the Signature molecular descriptor as a basis for designing molecular structures that aggressively reduce the surface tension of water in an effort to discover new shrinkage reducing admixtures. In particular, this work focuses on two classes of compounds, amines and glycol ethers, as they are surface-active agents and possess a capability of dramatically reducing the surface tension of water. After evaluating 14 compounds for initial surface tension reduction, a quantitative structure property relationship (QSPR) model for surface tension reduction was created, and then a structure enumeration algorithm was employed to generate non-intuitive structures outside of the original training set that have optimal predicted properties. Once newly identified surface tension reducing agents are tested and previously developed QSPR model is evaluated, optimal candidates are tested in application, in particular, for autogenous and drying shrinkage in cementitious materials. The existing patent literature indicates that glycol ethers are the active ingredients of commercially available SRAs. However, the CAMD algorithm identified non-intuitive structures with a variety of functional groups that have not been previously recognized as potential SRAs. 

Twelve additives were tested for their ability to inhibit both autogenous and drying shrinkage in portland cement. Among these, two were commercial SRAs one was an active ingredient in a commercial admixture, five were newly identified compounds predicted by using I-QSPR approach, and four were original training set compounds. Results for both drying shrinkage and autogenous shrinkage indicate that designed compounds perform similar to commercial admixtures yet have vastly different chemical functionalities. Accordingly, the selection of potentially new SRAs can be facilitated by the use of the I-QSPR strategy which indicates that the CAMD technique is a game changer.