Energy barriers at grain boundaries dominate charge carrier transport in an electron-conductive organic semiconductor

I Vladimirov,R T Weitz,M Kühn,T Geßner,F May

doi:10.1038/s41598-018-33308-y

Abstract

Semiconducting organic films that are at the heart of light-emitting diodes, solar cells and transistors frequently contain a large number of morphological defects, most prominently at the interconnects between crystalline regions. These grain boundaries can dominate the overall (opto-)electronic properties of the entire device and their exact morphological and energetic nature is still under current debate. Here, we explore in detail the energetics at the grain boundaries of a novel electron conductive perylene diimide thin film. Via a combination of temperature dependent charge transport measurements and ab-initio simulations at atomistic resolution, we identify that energetic barriers at grain boundaries dominate charge transport in our system. This novel aspect of physics at the grain boundary is distinct from previously identified grain-boundary defects that had been explained by trapping of charges. We furthermore derive molecular design criteria to suppress such energetic barriers at grain boundaries in future, more efficient organic semiconductors.

Highlights

Progress in understanding charge transport in organic semiconductors has benefited significantly from both experimental and theoretical works
The reason is that the energy barriers caused by such filled wells can be screened by ionized dopants or counter charges at the gate dielectric reducing the effective trap height[14,15,22,23]
In realistic thin film transistors, grain boundaries can dominate charge transport, since they can act as traps for charges reducing the charge carrier mobility[33,34], lead to increased bias stress[35,36] or lower long-term electrical stability[37,38]

Summary

Materials and Methods

The DOS at large negative energies (deep traps, between E = −100 meV and E = −70 meV for the raw data) is regarded, which can be described by an exponential. The fitted exponential term (see equation above) is subtracted from the measured DOS for all energies. A parabolic fit of the logarithmized data is used to obtain the parameters N and σ. This procedure is repeated applying different shifts of the experimentally determined energies E until the number of states with E < 0, i.e. the integral of the analytic function above from E = −∞ to E = 0, is equal to half the total number of states in the sample as found in our simulations.

Author Contributions

Additional Information