Unravelling Polarity Preference and Electrode Layer Interactions on Wurtzite Aluminum Nitride-Based Memristors for Neuromorphic Synapses

Aluminum nitride (AlN) memristors show great promise for neuromorphic computing due to their exceptional properties, including ultrafast switching, low current requirements, high on/off ratios, and CMOS compatibility. These characteristics make them ideal for emulating artificial synapses, crucial components in neuromorphic systems. Optimizing device performance depends heavily on AlN surface polarity and understanding the interactions between AlN and electrode layers.<br/>This study investigates the properties of electrode layers on AlN slabs and their influence on surface stability and polarity preferences. Using first-principles simulations, we calculate formation energies of various electrode layers on AlN slabs of each polarity and analyze the resulting trends. We also examine Bader charges of the electrode layers to elucidate interfacial interactions. Our results indicate that the interaction between electrode layers and AlN slabs is thermodynamically favorable and primarily electrostatic or ionic in nature. On Al-polar AlN slabs, the preferred stacking sequence of electrode layer elements aligns with the Wurtzite structure of the AlN slabs, minimizing lattice distortion and destabilization. We observe that the Born effective charge of interfacial atoms on AlN slabs varies with the electronegativity of electrode layer elements, potentially impacting interface polarization and piezoelectric properties, which could influence neuromorphic device performance.<br/>Our findings suggest that electrodes with stabilized interaction can be designed on AlN slabs with preferred polarity by considering the electronegativity or Bader charges of the electrodes. This study provides valuable insights for optimizing AlN-based memristive devices in neuromorphic computing applications, contributing to the development of bioinspired information processing systems and adaptive bio-interfacing.