The behaviour of perovskite memories, solar cells and LEDs is strongly influenced by defects, namely deep and shallow traps and mobile ions. Their characterisation is difficult because their signatures overlap in conventional electrical measurements. We develop methods to distinguish these contributions, validate the methods against device simulation, and establish the conditions under which each interpretation is reliable.
Approach
For each device that we fabricate or simulate, the aim is to understand the underlying physics and to express it in analytical or semi-empirical models. We study generation and recombination, and both electronic and ionic charge transport, using nano-scale imaging, deep-level transient spectroscopy, impedance spectroscopy, and low-temperature I–V and C–V measurements. Device structures are simulated with Synopsys Sentaurus and Silvaco 2D/3D TCAD, Lumerical and MATLAB, both to support the experiments and to develop new models.
Key results
- Trap characterisation from dark J–V characteristics. Trapped charge near the injecting contact causes band bending, which shifts the diffusion-to-drift transition voltage (the peak of d log J / d log V) to lower values. In simulation, this shift follows the trap density and depth in the presence of mobile ions, and distinguishes films more reliably than VOC.
- CEISIM, an in-house simulator. CEISIM (Coupled Electronic–Ionic Transport Semiconductor Device Simulator) is a one-dimensional drift–diffusion simulator that treats trap occupancy and mobile ions as dynamical variables, with DC, small-signal and transient solvers in a single framework. It provides access to the internal state of the device during impedance and transient measurements.
- DLTS of ionic defects in perovskite memristors (methylammonium vacancies, 0.48 eV) and LEDs (a bromide-vacancy-related trap, 0.30 eV).
- Compact models. An analytical model of the double-gate MOSFET for the channel potential, subthreshold current and drain current, and a simulation study of a pore-confined phase-change memory that uses current crowding to reduce programming energy.
Methods and tools
Sentaurus and Silvaco TCAD · CEISIM (our in-house 1D drift–diffusion simulator) · DLTS · impedance and capacitance spectroscopy · low-temperature I–V and C–V · MATLAB