Understanding Thermal Transport in Organic Semiconductors

Displays based on organic semiconductors can nowadays be found in nearly every high-quality cell phone, but this materials class also holds high promise for a variety of other applications. These include solid-state lighting, solar cells and photodetectors, flexible electronics, and also thermoelectric devices. In all these devices, heat transport is of crucial relevance. This, on the one hand, applies to removing waste heat generated by irradiation with IR light, through non-radiative carrier recombination, and as Ohmic losses as a consequence of charge transport processes. On the other hand, for thermoelectric applications it is crucial to maintain a sufficiently large temperature gradient, which can be achieved by a minimized thermal conductivity. Nevertheless, very little is known about the thermal conductivity of organic semiconductors. This in particular applies to the relationship between the thermal transport coefficients and the structure of the material. In this context, structure has several meanings, namely the atomistic structure of the used molecules, their relative arrangement in a molecular crystal, and the types of interaction between individual molecules within the crystals. To close this knowledge gap, the current project aims at developing detailed structure-to-property relationships for thermal transport in (crystalline) organic semiconductors, addressing especially the aspects mentioned above. The primary strategy for achieving that will be computer simulations, due their inherent flexibility and efficiency. Additionally, it is planned to perform project-relevant experiments with collaboration partners in Austria (Materials Center Leoben), Italy (University of Bologna), Belgium (Free University of Bruxelles), and Japan (University of Nara). As far as the simulations are concerned, we will pursue a dual strategy, studying heat transport in real and in reciprocal space. This means that we will study the thermal conductivity spatially resolved at an atomistic level employing molecular dynamics approaches to identify, which parts of the molecular crystals act as heat-transport bottlenecks. Additionally, we will relate the thermal conductivity to the properties of the (quasi)particles that actually carry the energy the so-called phonons. These phonons are intimately linked to the vibrational properties of the crystals and by changing the chemical structure of the molecules and their relative arrangement, it will be possible to modify these vibrational properties. This will result in modified energies, velocities and lifetimes of the phonons and, thus, in varying heat transport properties. As the key outcome of the present project, the developed structure-to-property relationships will allow a knowledge-based design of new materials with either maximized or minimized thermal conductivities, depending on the specific application the materials are meant for.

The transport of thermal energy is crucial in nearly all "real world" applications of materials, be it for the dissipation of waste heat, for preventing heat dissipation, or for providing thermal energy to trigger processes that require this energy. Nevertheless, heat transport is only poorly understood in complex materials like organic semiconductors or their kins, metal-organic-frameworks (MOFs). The current project significantly contributed to changing that situation. For example, a strategy based on artificial intelligence (AI) was devised to reliably predict thermal transport processes even in highly complex materials at an unprecedented level of accuracy. Consequently, atomistic simulations are now able to quantitatively describe experiments. Importantly, in the said approach AI is not merely used to look for correlations. Rather, it is applied to provide the tools that then make it possible to perform simulations that look for causalities at an unprecedented level of accuracy and at truly outstanding speeds. As an example, for the largest simulations performed in the current project, we estimated the speedup through the applied so-called machine-learned potentials compared to density-functional theory (the classical quantum-mechanical method applied to model materials) to amount to an amazing factor of 10^11. A rather broad range of insights was possible by analyzing heat transport based on the actual particle trajectories in so-called non-equilibrium molecular dynamics simulations and at the same time studying the dynamics of so-called phonons, the quasiparticles that carry quantized portions of thermal energy. This allowed us to identify the bottlenecks to thermal transport in organic semiconductors and MOFs. In the former case these are the interfaces between adjacent molecules and in the latter the covalently bonded interfaces between the metal-oxide nodes and the organic linkers. Additionally, we were able to show that traditional models that describe the migration of the aforementioned phonons merely as quasi-classical particles are insufficient for understanding thermal transport in low thermal-conductivity materials and that also phonon-tunnelling needs to be considered. This turns out to be particularly important, when one intends to understand, how the thermal conductivity of a material changes with its temperature and how heat transport differs in different crystallographic directions. Analyzing thermal transport via the migration of quasi-particles carrying thermal energy also provides an explanation, why the thermal conductivity steeply rises at elevated pressures: these increases both, the speed of the phonons as well as their lifetimes (i.e., the times between inelastic collision processes with other phonons).Finally, when studying classical and semiconducting polymers we find that the thermal conductivity along polymer chains are extremely high, provided that they are extremely well ordered (i.e., crystalline) and as long as they do not bear any side chains. This can again be traced back to the properties of phonons.