De Novo Calculation of the Charge Carrier Mobility in Amorphous Small Molecule Organic Semiconductors

Abstract

Organic semiconductors (OSC) are key components in applications such as organic photovoltaics, organic sensors, transistors and organic light emitting diodes (OLED). OSC devices, especially OLEDs, often consist of multiple layers comprising one or more species of organic molecules. The unique properties of each molecular species and their interaction determine charge transport in OSCs—a key factor for device performance. The small charge carrier mobility of OSCs compared to inorganic semiconductors remains a major limitation of OSC device performance. Virtual design can support experimental R&D towards accelerated R&D of OSC compounds with improved charge transport. Here we benchmark a de novo multiscale workflow to compute the charge carrier mobility solely on the basis of the molecular structure: We generate virtual models of OSC thin films with atomistic resolution, compute the electronic structure of molecules in the thin films using a quantum embedding procedure and simulate charge transport with kinetic Monte-Carlo protocol. We show that for 15 common amorphous OSC the computed zero-field and field-dependent mobility are in good agreement with experimental data, proving this approach to be an effective virtual design tool for OSC materials and devices.

28‐1: Invited Paper: Bottom‐Up OLED Development by Virtual Design: Systematic Elimination of Performance Bottlenecks Using a Microscopic Simulation Approach

“…The resulting all-particle trajectories allow a detailed analysis of spatially resolved and material specific distributions of loss processes, as well as their dependency on material properties. (15,26,27,34). Details of the approach can be found in the references accordingly.…”

“…Following prior work, we apply a seamless bottom-up multiscale simulation workflow to compute OLED material properties and simulate device characteristics (12,14,15,(24)(25)(26)(27)(28)(29)(30): First, we compute molecule specific forcefields automatically with quantum chemistry. We then apply a simulation protocol mimicking physical vapor deposition to generate atomistic models of pristine or mixed thin films or multilayer structures with atomistic resolution (13,31,32).…”

“…Resulting mobilities are depicted in Figure 1 showing good agreement in comparison to experimental data from literature for both field dependency (left panel) and extrapolated zero-field mobilities over a wide value range. Further, mobilities for three isomers of the molecule BPD (middle panel) show that the mobility workflow can even resolve subtle differences in molecular structures (15). How to efficiently apply the multiscale approach to systematically improve the mobility of a certain compound was demonstrated for the molecule Alq3 (28).…”

“…The key, however, to maximize the impact of computational approaches in OLED R&D, are parameter-free models that predict the impact of molecular properties on device performance without parametrization with experiment (11)(12)(13)(14)(15)(16)(17). Despite recent advances (18) such predictive models to fully connect fundamental chemistry and device design still struggle due to the high level of complexity: a large ensemble of effects and processes needs to be considered, ranging from charge transport in pristine layers via injection processes towards complex excitonic processes in mixed co-emission systems (7,8,(19)(20)(21)(22).…”

Toward Accurate Prediction of Ion Mobility in Organic Semiconductors by Atomistic Simulation

“…With these quantities, the rate of charge carrier mobility ( k IJ ) between all combinations of monomer I and J in the models can be calculated. To predict the charge carrier mobility (μ sim ), the ion mobility kinetics may be simulated by Ehrenfest dynamics − or kinetic Monte Carlo (kMC) simulations. − Especially the recently reported kMC simulation for predicting mobility provided a good agreement with the experiment . Since density functional theory (DFT) has been typically adopted for the quantum mechanical part of the process, , it has been the most time-consuming step, limiting the overall complexity of the models.…”