Preface

The development of our modern society is inherently related to the development of some key sectors of science and technology, namely energy, medicine, space and information technologies. The latter is becoming especially prominent due to the pervasive role of the internet, social networks and digital communications in everyday life. Commonplace devices such as smartphones, smartwatches, etc. utilise, in general, conventional CMOS-based (nano)electronics where the consumer demands for connectivity, speed, streaming, exchanged data volume and memory capacity continuously increase. Thus, the microelectronics industry is focused on creating smaller and smaller units with higher and higher integration density so as to be able to process, transmit and save more and more data. However, this shrinking in size has physical limits and further miniaturization of conventional electronics is restricted by basic physics. In addition the opportunity to move to smaller and smaller elements is limited; for example in transistors the thickness of the oxide barriers used becomes insufficient to stop statistical relativistic processes such as the escape of the electron from the small volume. As a result, a qualitative change, or in other words a new paradigm, is required. One of the most promising opportunities for this is through the introduction of components/devices that use atoms or ions as a fundamental operating principle instead of electrons. Ions have the advantage of much higher mass and can be more easily characterized and manipulated compared to electrons. In addition, different atoms and ions have inherently different properties in terms of mass, charge, chemical and physical behaviour. This allows us to select a particular sort or particular sorts of species depending on the target application. Conventional nanoelectronics now needs to incorporate nanoionics.

Memristors or memristive cells are typical examples of mixed nanoelectronic/nanoionic devices. In these cells redox reactions and/or transport of ions leads to significant changes in the electronic properties of the material, for example of its resistance. This change is then used to distinguish between the Boolean 0 and 1 that represents digital data storage. Nowadays memristors are capable of a variety of functionalities such as digital and/or analog data storage, selector abilities, sensing, learning and forgetting, etc. These devices are scalable to almost atomic level, being extremely fast (ps responses have been reported), energy efficient, robust and reliable. An important advantage is their stability towards high energy particles and electromagnetic waves, and their ability to operate at temperatures as low as 4 K and as high as 1000 K. These properties make them ideal for application in space technologies and/or harsh environments. Memristors also represent the smallest building units for devices used in alternative logic operations and neuromorphic computing, targeting much broader and more complex visions such as the Internet of Things (IoT) and artificial intelligence (AI). Memristive-based hardware is already on the market and is incorporated as field programmable gate arrays in space satellites and humanoid robots.

Despite broad scientific and public interest in these devices their industrialization is still progressing relatively slowly. One important reason for this is that there is a missing component in our knowledge, namely the relation between the materials used in memristors and the corresponding device properties and functionalities. Understanding this relationship will allow for rational design that will ensure the achievement of particular functionalities, by selecting materials and material combinations.

The Faraday Discussion Meeting “New memory paradigms” brought together physicists, chemists, electrochemists, engineers and computer scientists to discuss recent developments in this field and to provide a forum for discussion and exchange of ideas and expertise to overcome existing problems and to generate new ideas and solutions.

The Discussion was focused on the most promising memristive systems: namely electrochemical metallization memories (ECM, or also called CBRAM), valence change memories (VCM or OxRAM) and phase change memories (PCM). The discussions addressed issues in clarifying the reaction/switching mechanisms, the influence of external factors such as atmosphere, humidity, light etc., as well as the additional criteria for materials selection. Finally memristive-based neuromorphic computing was introduced and critically discussed, including fundamental questions about the requirements in terms of the properties and characteristics of devices for effective data processing.

The field of memristive applications and memristors is one of the most interdisciplinary areas of science, providing examples of a variety of unexpected properties of matter, and involving scientists with different expertise, presenting them with unexpected problems such as field-enhanced kinetics, overlapping space charge layers, nanoscale reactivity, materials interactions, Rayleigh instabilities, etc. Problems of this type can only be resolved on the basis of complex approaches based on a broad and interdisciplinary discussion.

---