Research
Systems
A brief overview of the systems we have studied in the past few years is shown in the cartoon. We are always looking for new problems with these characteristics: systems (i) interesting to large community of experiementalists and (ii) whose fundamental physics is not fully understood. After the move to Liverpool, we are currently working more in the area of materials discovery with focus on molecular and polymeric materials for electronics, bioelectronics and energy applications. Machine learning and high-throughput virtual screening have been added to the toolbox.
Approach
We use a range of methodology from computational chemistry and physics (electronic structure calculations of molecules and solids, classical molecular dynamics simulations) and combine them with phenomenological or analytical theories. We often use the former group of methods to develop better analytical models and from the analytical models we formulate predictions. For example we worked few years on the theory of charge transport in molecular crystals, starting with the computational observation that dynamic disorder was important and developing an appropriate model of transport. In a key paper (with S. Fratini and S. Ciuchi) we built a map of all molecular semiconductors to drive the discovery of new materials. We recently completed an overview of this field.
The idea the general models can be derived by gradually simplifying detailed ones is the general concept behind model reduction. We proceeded in this way to build simplified models from detailed models of organic solar cell. We used the simplified model to propose a general principle for the design of electron acceptors. Few years later we analyzed a large body of the literature and verified that the prediction was correct in a statistical sense. We are now making a greater use of large data sets and probe the validity of hypotheses using the modern framework of machine learning.
Our interest in machine learning comprises two aspects. On one hand we use it to predict the performance of new materials using experimental datasets. However we are also interested in the formal limitations of this approach, for example to determine under what conditions the adoption of machine learning methods is useful for materials discovery. We are also interested in the limitations that prevent machine learning from discovering materials with completely new chemistries.
An increasingly important fraction of our work revolves around high-throughput virtual screening for molecules and molecular crystals that we have applied it to singlet fission, TADF, and the identification of inverted singlet-triplet molecules. We are part of the Diadem project, whose goal is to develop a general purpose platform for the discovery of organic electronics materials, from identification of candidates to experimental verification.
We used the same model reduction approach to study the problem of charge transport in polymeric systems. We build atomistic models of different polymers and computed their electronic structure in bulk. We used this information to build a very general model of polymer from which we extracted the universal properties of transport. Going back to the chemistry we predicted the structure of monomers that would make a polymer insensitive to structural disorder. We are continuing with the study of polymers, including the construction of hybrid coarse-grained and atomistic models and approximate methods for the study of the polymer electronic structure. We have recently started an ERC funded project on organic bioelectronics with the first collaborative work appeared in JACS.
Different methodologies but similar philosophy is used on our recent series of work on exited states dynamics of molecular materials. We considered the fluctuation of the excitonic coupling due to molecular fluctuations and discussed their role in different regimes of exciton transport in molecular crystals. We pay attention to the computation of the excitonic coupling to include the short range interactions. We study the open quantum system dynamics by extending the surrogate Hamiltonian approach into a vibronic basis. We are active in the study of singlet fission focusing on physical models that allow the co-existence of localized and delocalized states.
