On the Development of Memsensors

Since the postulation of the experimental realization of memristive devices in 2008, a broad variety of concepts for the fabrication of memristive devices has been pursued and the underlying switching mechanisms have been studied in detail. The unique electronic properties of memristive devices inspire applications that go beyond conventional electronics, such as using memristive devices as programmable interconnects, to realize logics for in-array-computing or in neuromorphic engineering. A particularly interesting aspect of biological neural networks is the close connection between signal detection and processing at the neuron level, which is an essential contribution to their outstanding efficiency. This work evolves around the concept of memsensors, which unify the characteristic features of memristive devices and sensor devices and as such appear as promising candidates to realize a close connection between signal detection and processing on the device level. Memsensors are a highly interdisciplinary topic, bridging research in the fields of material science and electrical engineering and relating to insights from biology and medicine through neuromorphic engineering. The major objective of this thesis is to provide tools and building blocks and showcase pathways to incorporate memristive and sensitive properties into memsensor devices. For this purpose, motivated by an experimental point of view, a nanoparticle-based memristive device with diffusive memristive switching characteristics was developed and characterised in detail and sensors relying on semiconducting metal oxide thin films and nanostructures were thoroughly studied. In addition, in terms of modelling of memsensor circuits, emerging features such as amplitude adaptation are discussed, showcasing the particular eligibility of memsensors in the context of neuromorphic engineering.