New “intelligent” electromechanical functionalities added to silicon chips
The information revolution is synonymous with the traditional quest to pack more chips into less space and so increase computing power. This quest is embodied by the famous “Moore’s law”, which predicts that the number of transistors per chip doubles every couple of years and has held true for a remarkably long time. However, as Moore´s law approaches its limit, a parallel quest is becoming increasingly important. This latter quest is nick-named “more than Moore”, and it aims to add new functionalities (not just transistors) within each chip by integrating smart materials on top of the ubiquitous and still indispensable silicon base. Among these so-called smart materials piezoelectrics stand out for their ability to convert a mechanical deformation into a voltage (which can be used to harvest energy to feed the battery) or, conversely, generate a deformation when a voltage is applied to them (which can be used, for example, in piezoelectric fans for cooling down the circuit). However, the integration of piezoelectricity with silicon technology is extremely challenging. The range of piezoelectric materials to choose from is limited, and the best piezo electrics are all lead-based ferroelectric materials, and their toxicity poses serious concerns. Moreover, their piezoelectric properties are strongly temperature-dependent, making them difficult to implement in the hot environment of a typical computer processor, whose junction temperature can reach up to 150 degrees Celsius. There exists, however, another form of electromechanical coupling that allows a material to polarize in response to a mechanical bending moment, and, conversely, to bend in response to an electric field. This property is called “flexoelectricity”, and though it has been known for nearly half a century, it has been largely ignored because it is a relatively weak effect of little practical significance at the macroscale. However, at the nanoscale flexoelectricity can be as big as or bigger than piezoelectricity; this is easy to understand if we consider that bending something thick is very difficult, but bending something thin is very easy. In addition, flexoelectricity offers many desirable properties: it is a universal property of all dielectrics, meaning that one need not use toxic lead-based materials, and flexoelectricity is more linear and temperature-independent than the piezoelectricity of a ferroelectric. Researchers from the Catalan Institute of Nanoscience and Nanotechnology (ICN2), in collaboration with Cornell University (USA) and the University of Twente (Netherlands), have now managed to produce the world’s first integrated flexoelectric microelectromechanical system (MEMS) on silicon.
They have found that, at the nanoscale, the desirable attributes of flexoelectricity are maintained, while the figure of merit (bending curvature divided by electric field applied) of their first prototype is already comparable to that of the state of the art piezoelectric bimorph cantilevers. Additionally, the universality of flexoelectricity implies that all high-k dielectric materials used currently in transistor technology should also be flexoelectric, thus providing an elegant route to integrating “intelligent” electromechanical functionalities within already existing transistor technology. The results have just been published in the journal Nature Nanotechnology. The project, led by Dr Umesh Bhaskar and ICREA Professor Gustau Catalan, from the ICN2 Oxide Nanoelectronics Group, was funded by a European Research Council (ERC) Consolidator Grant and a Spanish Project from the National Plan for Research Excellence, as well as by national grants for the US and Dutch teams.