Using a single silicon nanowire, we have made the first prototype of a multipurpose device which can be used as a field effect transistor, a Schottky diode or a p-n diode. We obtain these functions using electrostatic gates to control the interfaces between the metallic source and drain contacts, and the semiconductor channel.
Nanometer-scale electronic devices are expected to be at the forefront of next-generation electronics. These devices will likely rely on the quantum nature of electrons, unlike the classical principle of operation of today's devices. Silicon nanowires with typical diameters of several tens of nanometres and lengths of several tens of micrometers (Fig. 1) provide both versatile and relatively simple systems that enable the investigation of complex quantum phenomena, due to electron confinement and interactions.
Contacting silicon nanowires to metal electrodes is challenging, due to the presence of a Schottky barrier at the metal/silicon interface. The Schottky barrier, with a typical height of ~0.5 V, acts as an obstacle to the injection (or extraction) of carriers into (or out of) the silicon nanowire. As a result, contacts with a small interface surface area can be highly resistive. A common approach to suppress the Schottky barrier consists in strongly doping the semiconductor close the interface. Doping can be difficult to achieve and control when the semiconductor is a single nanometric wire. We have developed an alternative approach, based upon the use of a local electrostatic field. This approach is applied to non-doped silicon nanowires having diameters as small as 20 nm.
Fig. 2: a) Scanning electron micrograph of a single nanowire device. S = source, D = drain, with corresponding gate electrodes GS and GD. All of the device fabrication has been carried out at the Plateforme Technologique Amont (PTA). b) Schematic of the device cut. c) Use as bipolar field-effect transistor d) Use as a p-n diode
Manufacturing the device
Figure 2 shows a scanning electron micrograph of a single nanowire device as well as its schematic. After their synthesis through a catalytic CVD growth in collaboration with SP2M, Leti and CNRS, the nanowires are first dispersed in ethanol and then transferred onto an oxidised silicon substrate with the aid of a micropipette. Using electron beam lithography, nickel contacts are deposited at the ends of a selected nanowire. When annealed at 500°C, nickel penetrates into the nanowire and forms a metallic binary phase of nickel silicide. During the annealing process, the metallic phase can grow over several hundreds of nanometres in the nanowire while the interface between silicon and the silicide remains well defined. This step is crucial since the metal semiconductor interfaces can be moved away from the contacts allowing them to be surrounded by control gate electrodes (GS and GD in Fig. 2a).
Transistor or diode on demand
When the GS and GD grids are not polarised, the source-drain resistance is very large (off state) because of the dominant resistance from the source and drain Schottky barriers. These barriers can be selectively suppressed by applying a few volts at the corresponding gate. When both gates are simultaneously polarized at the same gate voltage, the device behaves as a transistor. Negative voltages enable conduction through the silicon valence band (hole transport). Positive voltages favour transport through the silicon conduction band (electron transport). This bipolar behaviour is shown in Fig. 2c.
If only one of the Schottky barriers is suppressed, the device behaves as a Schottky diode. A more interesting configuration is obtained when opposite voltages are applied to the two gates. In this case, half of the channel is populated with holes (p type) and the other half with electrons (n type). The resulting electronic properties correspond to a p-n diode (Fig. 2d). In principle, this configuration could be also used for the fabrication of a nanoscale photodiode.
Maj : 20/02/2014 (974)