Memory materials: a unifying description

Memory can be defined as the ability to store the state of a system at a given time, and access such information, or part of it, at some later time(1). This state could be the spin polarization, or the doping profile, or some other physical characteristic of the system. Ultimately, however, the physical origin of memory can be traced back to the dynamical properties of the constituents of condensed matter, such as electrons and ions. Irrespective, it turns out that essentially all memory materials and systems show resistive, capacitive, and/or inductive properties that are hysteretic when subject to time-dependent perturbations(1). This means that their characteristics can be classified as memristive, memcapacitive, or meminductive (or a combination of these); typical of non-linear circuit elements with memory that go under the name of memristors(2), memcapacitors, and meminductors(3). Starting from a simple example we will introduce the general notion of memory circuit elements. In fact, we can further argue that even when the materials are not probed via electrodes but via, e. g., non-contact electromagnetic pulses, and they show memory features (such as certain metamaterials(4,5) or phase-change materials(6)) their generalized response can be characterized in a similar manner. We will then provide additional experimental and theoretical examples of memory materials and systems and show that they all fall within this general classification. We will show how this unifying description is a source of inspiration for possible digital and analog applications in diverse areas such as information storage, neuromorphic computing, etc.