Dielectric permittivity modulation at nanoscale in plasma synthesized silver nanoparticles based nanocomposites for in-memory computing

The intense work on development of unconventional approaches for computing and signal processing involves efforts on design and engineering of materials with tunable dielectric properties and switchable electrical state as conduction state. This is the case of in-memory computing using emerging non-volatile memories which has successfully opened up new prospects for neuromorphic computing via the option of high volume data traffic between processor and memory units but faces materials-related challenges mostly attributed to the intrinsic and non-ideal device properties and expresses complexity in hardware implementation. In the effort to advance on the concept we describe here a way for controlled modulation at nanoscale of the dielectric response of plasma synthesized silver nanoparticles (AgNPs) based nanocomposites and a method for mapping their dielectric permittivity via Electrostatic Force Microscopy. By embedding a 2D-network of AgNPs close to the surface of thin SiO 2 -layers, one can locally modulate the relative dielectric permittivity ( $\varepsilon _{r}$ ) of the device in a large range. The presence of AgNPs in the dielectric layer leads to a nanostructuration of the relative dielectric permittivity, with lower $\varepsilon _{r}$ -values above the AgNPs and higher ones in-between them, when compared to the $\varepsilon _{r}$ -value of homogeneous SiO 2 . A nanostructuration factor is introduced to account for this effect. The nanostructured dielectric response is related to modulation of the electric field inside these AgNPs-based nanocomposites. The results in this work generate important contributions towards the practical applicability of such AgNPs-based nanocomposites for neuromorphic computing, which is considered as an important step towards device engineering.