The Harvard Clean Energy Project: Large-Scale Computational Screening and Design of Organic Photovoltaics on the World Community Grid

Abstract

This perspective article introduces the Harvard Clean Energy Project (CEP), a theory-driven search for the next generation of organic solar cell materials. We give a broad overview of its setup and infrastructure, present first results, and outline upcoming developments.CEP has established an automated, high-throughput, in silico framework to study potential candidate structures for organic photovoltaics. The current project phase is concerned with the characterization of millions of molecular motifs using first-principles quantum chemistry. The scale of this study requires a correspondingly large computational resource which is provided by distributed volunteer computing on IBM's World Community Grid. The results are compiled and analyzed in a reference database and will be made available for public use. In addition to finding specific candidates with certain properties, it is the goal of CEP to illuminate and understand the structure-property relations in the domain of organic electronics. Such insights can open the door to a rational and systematic design of future high-performance materials. The computational work in CEP is tightly embedded in a collaboration with experimentalists, who provide valuable input and feedback to the project.

Boson sampling for molecular vibronic spectra

“…The corresponding vibronic transitions involve two electronic states and one can extract the molecular structural and force field changes from the spectra. The linear absorption spectra of molecules determines important properties such as their performance as solar-cells [29] or as dyes for either industrial processes [30] or biological labels [31]. The prediction of the linear absorption of molecules is challenging computationally, especially when complicated vibrational features (see e.g.…”

Lead candidates for high-performance organic photovoltaics from high-throughput quantum chemistry – the Harvard Clean Energy Project

“…4 It was combinatorially generated from 26 basic building blocks according to predetermined rules regarding possible connections. The fragments and connectivity rules were chosen with promise and synthetic feasibility in mind.…”

“…As detailed in ref. 4, we perform a series of DFT single point calculations on the different geometries using an array of model chemistries (i.e., functional and basis set combinations). The results are empirically calibrated to correct for some of the systematic errors in each theoretical model and account for the situation in a real material.…”

“…The calibration is based on linear regressions between a training set of known experimental data and the corresponding computational results, i.e., it aligns the predictions of the latter to the former. 4 Details are given in the ESI. † In order to obtain more robust values and reduce random errors introduced by failures of individual model chemistries or calibrations for particular data points, we average over all the independently acquired results.…”

Discovering a Transferable Charge Assignment Model Using Machine Learning

“…[11] Trained to reference datasets, ML models can predict energies, forces, and other molecular properties. [12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27] They have been 3 used to discover materials [28][29][30][31][32][33][34][35][36][37] and study dynamical processes such as charge and exciton transfer. [38][39][40][41] Most related to this work are ML models of existing charge models, [9,[42][43][44] which are orders of magnitude faster than ab initio calculation.…”