AbstractOrganic semiconductors are quietly revolutionising the modern world, finding application in light-emitting diodes, thin-film transistors and photovoltaics. However their short device lifetimes and reduced power-conversion efficiencies have limited their commercial uptake. Both experimental and computational work to find new materials and architectures and study of their charge transport properties continues.
Computational studies of the charge transport properties of organic semiconductors depend heavily on the structure/morphology. Structures are often assumed to be a regular lattice, even in the case of amorphous materials, with some groups using molecular dynamics (MD) and Monte Carlo (MC) methods to capture the disorder. However these approaches have significant drawbacks, some of which are technical (MD and MC can be computationally arduous at large system size) and some of which are physical (below the glass transition, the energy landscape makes it difficult to sample a large number of states). Other approaches such as coarse-grained and basin-hopping methods attempt to overcome these problems but often create new problems such as the need to develop new force-fields, having to regain full-atomistic representation of molecules afterwards, and still being hindered by the landscape.
In this work, a new method, Simulation of Atomistic Molecular Structures using an Elastic Network (SAMSEN), is proposed and applied to molecular and polymeric systems. SAMSEN contains both a structural and dynamical model that individually attempt to overcome the technical and physical problems of current methods. The structural model requires that molecules are split into rigid sections, which retain their atomistic representation and are constructed to interlock with their neighbouring sections, and restrict the maximum displacement of atoms from their locally optimised positions. Atomic collision rules, limiting minimum separations, are also enforced. This combination allows SAMSEN to recreate the short-range structure of weakly-interacting non-polar small molecules. The dynamical model creates an elastic network between the rigid sections and displaces them across the low-frequency vibrational modes to achieve collective large-scale motion and computationally-fast structural relaxation.
SAMSEN is applied to systems of spheres to study the structural and dynamical parameters and a regime is found where a band of collective low-frequency modes can be found, sampling rate can be increased without altering structure and the mean overlap of atoms can be controlled without altering the sampling rate. The structural parameters are determined entirely by the class of system being studied, leaving the dynamical parameters to be chosen to maximise sampling rate. SAMSEN is then applied to systems of small molecules and is shown to be widely applicable, providing good approximations to the short range structures produced by full atomistic force-field methods. SAMSEN phenyl-C61-butyric acid methyl ester (PCBM) states are found to have structures close to those of the higher energy minima in a potential energy landscape of an all-atom potential and the distribution of these potential minima is found to be near-Gaussian. SAMSEN outputs are then used as inputs of MD simulations, recreating the full-MD structures and quickly finding the lower-energy minima. This method provides a useful pathway for those interested in sampling the distribution of morphologies at equilibrium. The diffusion of electrons is simulated for each state in the distribution of MD and SAMSEN inherent structures and it is found that the diffusion constant is near independent of the potential of the minima, despite short-range structural differences.
Turning to polymers, the coarse-graining into rigid sections now controls the relative structure between repeat units. The structure and vibrational modes of poly(3-hexylthiophene) (P3HT) is studied in the pure amorphous phase and also when blended with PCBM. Increased persistence of the P3HT backbone is found upon mixing with PCBM but the density of neighbouring chains is unaffected up to the miscibility limit. Studying the long time (low frequency) modes, the rate of relaxation of the P3HT backbone is slowed by the addition of PCBM due to a shift in frequencies, rather than a change of collective behaviour of the individual modes. The excitonic transport properties of P3HT are then studied in the amorphous and crystalline phase in a transfer matrix approach, finding the exciton diffusion length in the low charge limit and comparing to experimental and previous computational work on P3HT nanoparticles. Low disorder polymer, Indacenodithiophene-co-benzothiadiazole (IDT-BT), is also studied, with a morphology generated, the inter-molecular transfer integrals calculated using a quantum chemistry package and the diffusion constant for holes determined for a range of intra-molecular transport rates. Comparing the hole diffusion constant to experimental and computational work, an estimate of the intra-molecular transport rate is made.
This work serves as a framework for researches interested in determining the structures, long-time vibrations and charge transport properties of weakly-interacting amorphous organic semiconductors where samples of states if required. It can achieve this without molecule-specific parameters or forces and does so at modest computational cost and therefore opens the possibility of sample a large number of states of simulating bigger system sizes (approaching device scale) without requiring access to high-performance computing resources. Further opportunity exists to constrain the dihedral angles, consider electrostatic interactions, as well as to study the charge transport properties of polymer systems incorporating an accurate model for intra-chain charge-transport.
|Date of Award
|13 May 2020
|Alison Walker (Supervisor) & Enrico Da Como (Supervisor)
- organic semiconductor
- charge transport