Reconfigurable Field-Effect-Transistors

点击翻译按钮获取中文摘要

During the last decades, a performance increase and thus increased functionality of highly integrated circuits (ICs) has been achieved by increasing the number of transistors of decreasing size onto the same chip area. The scalability of transistors, however, will soon come to an end. Therefore, an attractive alternative approach to further increase the functionality of ICs is to add functionality to the devices themselves. This can primarily be achieved with so-called reconfigurable contacts that allow changing the devices to work as a-type, p-type as well as band-to-band tunnel field-effect transistor (TFET) dynamically.Field effect transistors (FETs) with reconfigurable contacts consist of at least two gate electrodes, namely one that is the actual gate and the other gate electrode (the polarity gate) that allows to reconfigure the transistor. Based on such multiple-gate device architectures, a number of reconfigurable devices based on nanotubes, nanowires and two-dimensional materials have been demonstrated. In particular, switching between the operation as a conventional transistor and a TFET is very attractive since this would enable circuits that can operate either optimized to highest performance (conventional FET operation) or to lowest power consumption (TFET operation). The major drawback of all current reconfigurable FETs is the fact that the polarity gate does not change the work function of the contact metal and as a result, a Fermi level line-up at mid-gap is necessary in order to obtain similar injection of electrons and holes. Hence, carriers are always injected through a substantial Schottky-barrier which yields strongly deteriorated device characteristics. Within the present project, reconfigurable contacts with an injection probability similar to ohmic contacts for electrons and holes are realized. Furthermore, the contacts are made in a way that enables unipolar device operation which is important for proper TFET functionality. To this end, a gated sandwich consisting of a silicon channel layer, ultrathin SiNx and graphene will be employed. The graphene will be used as a metallic source electrode. While graphene-silicon diode/contact structures have been studied in literature the important addition here is the ultrathin SiNx layer in-between the graphene and the silicon. This layer is decisive in order to obtain a very low density of gap-states in the silicon which is necessary for unipolar device operation and, more importantly, for TFET operation with low off-state leakage. The low density of states in graphene around the Dirac point further decreases the density of silicon gap-states and, in addition, enables sufficient gating of the silicon through the graphene/ultrathin SiNx stack so that the conduction/valence bands can be shifted with an appropriate gate voltage. As a result, the current approach provides truly reconfigurable FETs that can operate as n-type, p-type and tunnel field-effect transistors.