In this paper, the researchers use several approximations that enable the calculation of all the important properties for thousands of molecular semiconductors stored in the CSD. This constitutes a two-order magnitude expansion of the commonly computable landscape – allowing a complete mapping of all semiconducting molecular materials deposited in the CSD.
In addition, the work establishes the physical range of all parameters contributing to charge mobility and their correlations. If used with a suitable model for computing mobility, these parameters and their distribution can identify new high-mobility candidates and construct a model for physically achievable maximum mobilities. For example, it can predict the optimal (hole) mobility achievable by molecular semiconductors, one of their most critical characteristics. (Holes in a semiconductor crystal lattice can move through the lattice similar to electrons, acting as positively charged particles.) This proves that coupling physical models of transport with high-throughput screenings of high-quality data can produce both specific target materials and a more general physical understanding of the materials space.
Based on all criteria and calculations, the authors reduced the organic semiconductor subset on the CSD to around 5,000 structures, on which they performed a more in-depth study. They ranked the data based on computed mobility and found 40 structures in the higher mobility range of 13 cm2 V−1 s−1. Based on previous studies, of those top 40 structures, only 15 have been considered as high-mobility materials for transport, 10 appear in other optoelectronic/photonic applications, and 15 have never been considered as semiconducting materials.
Top-ranked organic semiconductor on CSD refcode GEDBIC
In general, it was observed that the materials with the highest mobility have the best combination of properties, but not all the properties are in the optimal range. For instance, the structure with the highest mobility – CSD refcode GEDBIC – exhibits an undesirable electron-phonon coupling but that is compensated by other key mobility parameters.
The large, high-quality dataset stored in the CSD allowed the researchers to develop a new approach for calculating the mobility of a given crystal structure in a fully automated procedure that takes just a few CPU hours. (According to the authors, traditional approaches in past papers on just one or a few molecules required millions of CPU hours.) To save resources on otherwise computationally heavy calculations, they used approximations, which they validated with experimental results. Using magnitudes more data than other studies, they were able to identify which properties are important to charge mobility, whether such properties can be optimized independently and what a reasonably achievable optimum for each would be. This helped them produce a list of possible compounds for use as semiconductors and a new tool for the discovery of new materials. Their approach can be coupled with crystal structure prediction (CSP) methods to establish if, among the low-energy compounds predicted by CSP, there are some that justify synthesis.
Computing charge mobility generally comprises two steps: computing the materials parameters from its structure and computing the mobility from such parameters.
For the structures, they used about 40,000 molecular semiconductors extracted from the over one million structures in the CSD. They considered only organic materials (i.e., containing elements H, B, C, N, O, F, Si, P, S, Cl, As, Se, Br, and I) and excluded cocrystals, polymers, disordered solids, duplicate structures and materials containing molecules with more than 100 non-hydrogen atoms. To ensure they considered only semiconductors, they only looked at molecules with the computed gap between the highest occupied and the lowest unoccupied molecular orbitals (HOMO and LUMO) in the range of 2 and 4 eV. They used transient localization theory to evaluate mobility because it tends to agree with more complicated quantum dynamics calculations but is fast enough for thousands of different problems.
Read the full paper: "On the Largest Possible Mobility of Molecular Semiconductors and How to Achieve It" Adv. Funct. Mater. 2020, 30, 2001906.
Learn more about the CSD, which houses over one million, hand-curated structures from x-ray and neutron diffraction analyses.
Explore more examples of CCDC tools in action.
Just for fun, can you identify where Rubrene sits on the new ranking of organic semiconductors in the paper’s CSD subset? If so, send us the answer at email@example.com.