(524e) An Application of Computer-Aided Molecular Design (CAMD) Using the Signature Molecular Descriptor: Designing New Water-Reducing Admixtures for Cement through Cement Paste Rheology Evaluation

Concrete is a material that forms the literal foundation of our modern society. Concrete is no longer simply cement, water and aggregates, but rather, it also typically contains a complex array of chemical admixtures that modify its many properties to produce a cost effective, durable, formable and mechanically sound material. Water-reducing admixtures (WRAs) have been developed to reduce the required water content of concrete for a given workability or increase the workability for a given water content.

            The development of new chemical admixtures for concrete is normally an experimental approach wherein one uses prior knowledge to suggest potential effective compounds. This approach is time-consuming, incremental and typically expensive. To accelerate the process, we propose a computer‐aided molecular design (CAMD) approch using the Signature molecular descriptor as a powerful alternative. CAMD is the application of computer-implemented algorithms that are utilized to identify and design new molecules with optimally predicted properties such that they can be tested and evaluated for efficacy. In this work, we are interested in the flow properties of cement paste, described by its rheological properties. Accordingly, a parallel-plate rheometer was used to measure the effect of eighteen low molecular weight organic chemicals belonging to different classes (sugars, sugar alcohols, and sugar acids) on rheology by directly obtaining yield stress and viscosity data. Workability, measured by a simple industry standard flow (mini-slump cone) test, was also obtained. Setting time, the time required for cement paste to stiffen, was measured since a higher workability is normally accompanied by an extended setting time, which is an undesired side-effect in some applications. After creating and refining four different quantitative structure–property relationship (QSPR) models for our four properties of interest (yield stress, plastic viscosity, workability, and setting time), a structure enumeration algorithm wass employed to generate structures outside of the original training set that have optimally predicted properties such as high workability, low yield stress and viscosity, and short setting time. Over 670 structures were newly generated and screened against a non-commercial chemical database, ZINC, to identify commercially available compounds. As a result, 6 compounds were purchased and tested for their effect on the flowability (i.e. workability, yield stress, and viscosity) and the setting time of cement paste for evaluation. Experimental results validated the previously developed QSPR models. Results confirm that the CAMD strategy is a powerful tool for generating novel and non-intuitive structures with the ability to increase flowability of cement paste with controlled setting time.