FIRST PRINCIPLES COMPUTATIONS MQSPR modeling in materials informatics: a way to shorten design cycles?

We demonstrate applications of quantitative structure–property relationship (QSPR) modeling to supplement first-principles computations in materials design. We have here focused on the design of polymers with specific electronic properties. We first show that common materials properties such as the glass transition temperature (Tg) can be effectively modeled by QSPR to generate highly predictive models that relate polymer repeat unit structure to Tg. Next, QSPR modeling is shown to supplement and guide first-principles density functional theory (DFT) computations in the design of polymers with specific dielectric properties, thereby leveraging the power of firstprinciples computations by providing high-throughput capability. Our approach consists of multiple rounds of validated MQSPR modeling and DFT computations to optimize the polymer skeleton as well as functional group substitutions thereof. Rigorous model validation protocols insure that the statistical models are able to make valid predictions on molecules outside the training set. Future work with inverse QSPRs has the potential to further reduce the time to optimize materials properties. Introduction Discerning and exploiting patterns in chemical data are at the heart of any systematic program for materials design. MQSPR refers to the application of the science of quantitative structure–property relationship (QSPR) modeling to materials informatics. The existence of quantitative relationships between chemical structure and the properties of materials was first discerned through the study of linear free energy relationships [1–4] early in the last century. These studies quantified the effect of a substituent group on equilibrium or rate constants. In recent years, the tools of statistical learning and pattern recognition have been employed to discover more complex relationships hidden in the wealth of data produced by high-throughput experimentation and robotic assays. Such statistical methods typically use an array of computed structural descriptors and/or process parameters as input to a model that can be trained to predict the value of an experimental quantity. When employed instead to predict a computed rather than an experimental quantity, statistical modeling can also serve to complement and leverage the results from firstprinciples computations, such as those using ab initio quantum chemistry and density functional theory (DFT), thereby enabling quantitative predictions on many more systems than would be possible in the same time span with first-principles computations alone. This paper deals with applications of MQSPR modeling to supplement first-principles DFT computations in the design of polymers with specific electronic properties, such as high dielectric constant and band gap for capacitors, or a specific range of glass transition temperatures. Our approach consisted of multiple rounds of validated MQSPR modeling and DFT computations to optimize the polymer skeleton as well as functional group substitutions thereof. Electronic supplementary material The online version of this article (doi:10.1007/s10853-012-6639-0) contains supplementary material, which is available to authorized users. N. Sukumar (&) Department of Chemistry, Shiv Nadar University, Chithera, Dadri 203207, Uttar Pradesh, India e-mail: [email protected]; [email protected] N. Sukumar M. Krein Q. Luo C. Breneman Rensselaer Exploratory Center for Cheminformatics Research and Department of Chemistry and Chemical Biology, Rensselaer Polytechnic Institute, 110 Eighth Street, Troy, NY 12180, USA 123 J Mater Sci (2012) 47:7703–7715 DOI 10.1007/s10853-012-6639-0