Modelling of Device Physics
Devices used in neuromorphic circuits operate taking advantage of more and more complex phenomena than the standard CMOS technology. In addition to charge flow, the device characteristics are controlled by the heat generation and transfer and by motion of ions all driven by the gradients of concentration, temperature, and electrostatic potential. Solving the appropriate differential equations with boundary conditions corresponding to complex device geometries can only be done numerically.
Most of the modelling is done by a finite element approach meshing the complex shape into simple elements and finding solutions in each one of them.
Once validated by the experiment, simulations allow for fast exploration of large parameter space of device design and material selection.