Probing Traps in Ta2O5/Al2O3 Memristive Switching Devices

Artikel i vetenskaplig tidskrift, 2026

Ta2O5/Al2O3 analog memristive devices are promising candidates for next-generation neuromorphic computing applications. This study presents a comprehensive investigation of the charge transport mechanisms in Ta2O5/Al2O3 devices, with a focus on identifying the properties and location of the defects involved in the formation of the conductive filament and analog resistive switching. By integrating experimental measurements with multiphonon charge transport simulations, we show that the relative thickness of the Ta2O5 and Al2O3 layers plays a pivotal role in determining the properties and location of the dominant defects that control the conduction through the RRAM stack. These defects are generated during the formation of the conductive filament and are, therefore, involved in the switching behavior. Sensitivity maps are utilized to pinpoint both the energy levels and spatial positions of defects within the oxide layers that contribute to the switching current. Temperature-dependent current-voltage (I-V) measurements allow us to extract key trap properties, including thermal ionization and relaxation energies, which are used to identify defects in the individual dielectric layers that contribute to the resistive switching. These insights offer valuable guidance for optimizing both the design and performance of these devices for neuromorphic applications.