The age of information has placed unprecedented demands on data storage. In response, a rich hub of research activity has erupted in the search for new types of memory with improved capacity and data processing speed [1, 2]. There has even been progress in mimicking the subtleties of how the human neural system remembers and forgets with artificial synaptic devices [3, 4]. Yet many challenges remain. As researchers at Brigham Young University in the US point out, 'Current data storage technologies have limitations that prevent their use in archival data storage applications'. Anthony C Pearson and colleagues look at the potential role of graphene in data storage solutions. Graphene's wide range of extraordinary properties has recommended it for a number of applications. This issue reports how it can form the basis of nanofuses that may be used for permanent, write-once-read-many (WORM) data storage devices [5].Leon Chua's forecast of 'Memristor—the missing circuit element' [6] and the subsequent demonstration of memristance in nanoscale systems byresearchers at HP Labs [7] triggered a surge of studies investigating the application of memristor devices in non-volatile memory [8]. In fact thesestructures may have a still broader impact in nanotechnology. At the nanoscale statistical variations in defect concentrations give rise to vast differences in behaviour, and device variability and ageing can be profound. These characteristics led George Snider to observe, 'Nanodevices are crummy', when considering their application in conventional Boolean logic systems [9]. Yet Snider's grim assessment was merely a prologue to an insightful recount of simulations demonstrating how to side-step a number of nanotechnology's main challenges by using self-organizing networks based on memristive devices.Similar fault tolerance is demonstrated in the organic-nanoparticle transistors developed by researchers at the University of Lille, who demonstrate a weighting behaviour that mimics aspects of synaptic functions that may be key to implementing neuromorphic computing [10]. Developments in this field show increasingly more sophisticated imitation of human neural processing with analogue memory functionality [3] and circuits that both learn and forget [4]. Keep an eye out for the Nanotechnology special issue on synaptic electronics later this year with guest editors James Gimzewski from the University of California, Los Angeles, and Dominique Vuillaume from the University of Lille, for a one-stop update on the latest cutting edge developments in this field.Graphene has demonstrated excellent potential in a number of applications from supercapacitors [11] to photomechanical actuators [12]. However, so far its potential in memory has been notably less explored. In this issue Anthony C Pearson and colleagues at Brigham Young University report how they fabricate graphene 'bow-tie' structures for fuses and program them though thermal oxidation [5]. As the researchers point out graphene's low atomic mobility, high chemical resistance to oxidation and excellent electrical conductivity make it well suited as archival data storage fuse material. They also highlight a number of attractive attributes in the devices as a whole: 'graphene WORM devices can be read and written electronically, can potentially have the very high data densities of flash memory, appear to be highly stable in both the on and off states, have a high on/off current ratio, and can be programmed with low voltages and powers'.In fuses an electrical connection is destroyed by an applied voltage. The process sounds destructive yet fuses are incredibly useful. Edison once famously commented, 'Just because something does not do what you planned it to do does not mean it is useless' [13]. There are synergies of this philosophy throughout nanotechnology research where apparently awkward behaviour is put to good use. Adding the dynamical nature of nanodevices to the list of obstacles to scalability, Snider rounded up his assessment, 'So we are faced with the challenge of computing with devices that are not only crummy, but dynamical as well'. Yet it was that dynamic behaviour that he went on to exploit in his demonstration of robust self-organizing networks using memristor devices. Successful research is by nature continually revealing behaviour that is unexpected or unusual. Thework described here is just some of the many examples of how successful development in research does not tolerate the unexpected, but embraces it.
71
