Computational optimisation of photoactive dye pairs for designing novel, highly efficient dye-sensitised solar cells.
Lead Research Organisation:
University of Cambridge
Department Name: Physics
Abstract
The overarching goal of this project is to further the power conversion efficiency of dye-sensitized solar cells (DSCs) - a renewable energy technology that can be effectively integrated into windows and wearables. One tactic to improve output current from a DSC device is by using multiple materials that absorb light in complementary regions of the solar spectrum. In particular, the aims of this project are to optimise key light-harvesting materials and combinations thereof when used synergistically in the same device. Using a large, custom-made database of such light-harvesting materials, optimum partner materials for well known, highly performing ones have been identified. Electronic structure techniques are being used to model the identified material pairs in their working environment, yielding information that will then be used to predict device parameters purely by computational means. This is the first time, to our knowledge, that this predictive method will be applied to multiple light-harvesting materials within the same device. By demonstrating that key parameters, such as power conversion efficiency, can be accurately predicted without relying on experimental data for input, automating the design of complementary materials is enabled via computational screening. The following work will then involve designing a workflow that can predict improved, hitherto unseen DSC materials, potentially using machine learning to tackle the optimisation. This effort is motivated by the increasingly urgent need for a diverse set of efficient renewable energy technologies. With an estimated 40% of energy consumption occurring in an urban environment, integrating efficient DSCs in windows of urban buildings would maximise harvested energy from light falling on surfaces that would otherwise go unused.
Planned Impact
Computing, data and communications infrastructure have transformed modern life. They all require software, security and trained personnel to work effectively and it has become common to use the term e-Infrastructure to refer to the whole of this interconnected ecosystem. E-Infrastructure has become a major contributor to advances in science and technology and it is clear that no industry will be able to compete internationally unless it exploits e-Infrastructure at highest level. The proposed EPSRC Centre for Doctoral Training in Computational Methods for Materials Science is focused on the development of new functionality in existing software, and even entirely new codes that will address challenges in materials that cannot presently be addressed. The UK is making large capital investments in e-Infrastructure but these investments will only achieve a fraction of their potential impact unless investments are also made in software development. The CDT will primarily focus on software development for materials science, which is itself, of course, extremely broad. However, the training provided in numerical methods, modern software development techniques and the exposure to present and emerging computational hardware means that the students will have the skilled set to work in any area of software after their PhDs. Given the universally acknowledge lack of highly trained personnel in software development one of our most important impacts will be in providing nearly 80 people who can apply this training in industry, including the very large number of UK software-based SMEs, or academia. The emphasis of the training and subsequent research project in the CDT is on development of innovative new methods for materials modelling and these will have impact in both academia and industry in further expanding the capability of materials simulation and the range of phenomena and processes that can be simulated and/or the amount of information that can be extracted from experiments. Thus we expect the CDT to have a significant impact across a very broad spectrum of disciplines.
People |
ORCID iD |
Nikolaos Nikiforakis (Primary Supervisor) | |
Leon Devereux (Student) |
Publications
Devereux L
(2021)
Data Science Applied to Sustainability Analysis
Devereux L
(2022)
In-Silico Device Performance Prediction of Cosensitizer Dye Pairs for Dye-Sensitized Solar Cells
in Advanced Energy Materials
Mukaddem K
(2020)
Dye-Anchoring Modes at the Dye···TiO 2 Interface of N3- and N749-Sensitized Solar Cells Revealed by Glancing-Angle Pair Distribution Function Analysis
in The Journal of Physical Chemistry C
Yang Z
(2020)
Predicting Device Parameters for Dye-Sensitized Solar Cells from Electronic Structure Calculations to Reproduce Experiment
in ACS Applied Energy Materials
Description | The feasibility of a computational modelling pipeline designed to rapidly screen specialised solar cell materials has been confirmed. These materials are dye molecules for dye-sensitised solar cells (DSCs) - they are the primary light-harvesting component of such cells. The pipeline developed matches two complementary dyes together, that each absorb a different range of wavelengths, such that the net absorption is maximised across the solar spectrum. Computationally demanding calculations on leading dye pairs in their working environment have been carried out using density functional theory (DFT, a quantum-mechanical modelling technique). These have yielded important descriptors of the dyes' electronic properties, which can be used to predict the power conversion efficiency of a DSC they are used in. Research into the most accurate and general model to perform this prediction is ongoing, alongside experimental fabrication of these cells to test predictions (experimental tests are done in collaboration with research group members at the Rutherford Appleton Laboratory, Oxfordshire). The standard technique to calculate absorption properties (called linear response time-dependent density functional theory, LR TD-DFT) is too computationally costly when applied to the very large model photoelectrode systems. An alternative technique called real-time time-dependent DFT, that uses a different scheme to obtain the same information, was also trialled on these systems. Whilst more efficient than the linear-response variant, it was ultimately deemed unfeasible in computational cost as well. Smaller dye-pair models are thus used to compute the key absorption information and used in the prediction methods above. |
Exploitation Route | It is hoped that the steps towards a full computational prediction pipeline for complementary dye pairs will enable much faster design of new, optimised dyes, inspired by computational drug design. This award is ongoing, and so in the remaining time following, it is planned that machine-learning based tools will be investigated to assist in this computer-aided molecular engineering of new dyes. |
Sectors | Electronics Energy Environment |
Description | ANSTO collaboration for experimental and computational studies on adsorbed dye conformations in dye-sensitised solar cells |
Organisation | Australian Nuclear Science and Technology Organisation |
Country | Australia |
Sector | Public |
PI Contribution | Performed density functional theory calculations to obtain structures of select dye molecules, used in dye-sensitised solar cells, once bound to a titanium dioxide surface in multiple binding conformations. |
Collaborator Contribution | Carried out experimental characterisation techniques on the select dye layers bound to titanium dioxide (including x-ray reflectometry, infrared spectroscopy) and comparing consistency of computed structures with these results. |
Impact | The results elucidate the nature of preferred binding conformations of the studied dyes. Understanding this at the molecular level is critical when designing further variants of such dyes to optimise their performance in dye-sensitised solar cells. A manuscript for publication is currently being prepared. |
Start Year | 2019 |
Description | Argonne Leadership Computing Facility allocation of computational resources |
Organisation | Argonne National Laboratory |
Country | United States |
Sector | Public |
PI Contribution | Feedback on computational performance (e.g. parallelisation efficiency) of the Theta, Mira and Cooley supercomputer clusters when carrying out density functional theory calculations. |
Collaborator Contribution | Permitted use of the high-performance computing resources of the Theta, Mira (until Jan 2020) and Cooley supercomputer clusters, under the banner of the Argonne Data Science Programme. These have been used to carry out all the computationally demanding DFT modelling within the project thus far. |
Impact | See "Key Findings" section for the award entitled "Computational optimisation of photoactive dye pairs for designing novel, highly efficient dye-sensitized solar cells". |
Start Year | 2018 |