Carbon nanotubes and graphene for various applications in electronics: Competition and synergy

Summary form only given: Although carbon nanotubes have been actively studied for electronics applications during the past 2 decades (1), transistors made with those materials are still plagued by a number of problems, such as contact reproducibility, large hysteresis in the transfer characteristics as well as low on current and large off-current values, to name a few. Single walled carbon nanotubes (SWNTs) were first considered as potential candidates for the replacement of Si MOS type transistors in VLSI circuits. However the main field of application of SWNTs is shifting towards large area electronics on flexible, plastic-type substrates (2, 3), a domain which is at present, less demanding in terms of device dimensions and integration density. In particular, random networks of SWNTs, which can be obtained by solution-processing or grown at low temperature by plasma enhanced chemical vapour deposition (PECVD), represent an attractive and viable option for the fabrication of electronic devices and circuits on non-refractory substrates (4). Also, since carrier injection in carbon nanotube field effect transistors (CNTFETs) is controlled by Schottky barriers at the nanotube/metal contacts, CNTFETs can be used for gas sensing applications (5). This is a consequence of poor "wetting" of most metals on the surface of SWNTs, thus allowing gas molecule to diffuse at the interface and modify the alignment of Fermi levels by introducing surface dipoles. Finally, SWNT random arrays will probably find applications in transparent conducting films (TCFs), see ref. 6, replacing some of the oxide-based films, particularly those incorporating indium, an element whose availability on earth is limited. Graphene is a one atom thick sheet of sp2-bonded carbon atoms, that was first isolated in 2004. Since then, the interest of the scientific community for this material has literally ex ploded and the reader is referred to recent reviews for detailed information (7, 8). Graphene is a zero gap semiconductor and its two-dimensionally (2D) confined carriers exhibit exotic properties, such as a linear energy dispersion relation (photon like), room temperature anomalous integer quantum hall effect, ambipolar charge transport, and a very high carrier mobility of 10
⁴
cm
²
/Vs at room temperature (2.10
⁵
cm
²
/Vs in liquid He). Because it exhibits no energy gap, graphene cannot be used directly to make transistor devices, particularly for logic applications. However, a further confinement of the electrons in one of the in-plane directions (resulting in graphene nanoribbons) can yield a kx or ky (wave vector) quantization, thus opening a forbidden gap in monolayer graphene (just like in SWNTs). It has been shown experimentally that the forbidden gap scales with the inverse of the ribbon width, w, according to Eg(eV) 0.8/w(nm), see ref. 9, which is in relatively good agreement with theoretical predictions. This means that the lateral dimension of graphene ribbons has to be below 10 nm if a significant gap is to be opened. Moreover, the lateral roughness of graphene nanoribbons (GNRs) should be controlled, in order to avoid deleterious effects on carrier transport. In ref. 9, Li and co-workers have used chemically-derived suspensions of narrow ribbons (down to less than 10 nm in width), exhibiting smooth edges, to fabricate graphene-based TFTs with hole mobility values in the range of 100-200 cm
²
/Vs. However, the process is far from mature, since the width of GNRs is not calibrated, meaning that the deposited films are not homogeneous. Another way to open a forbidden gap in a graphene-based material is to use bilayer graphene and a dual gate transistor configuration as has been shown recently (10). The graphene and CNT fields are highly entangled, as the use of graphene electrodes can improve CNT-based transistors (4) and lead to the fabrication of novel memories (11). Also, since controlling the width of nanoribbons below 10 nm is not very realistic, multiwalled carbon nanotubes (whose fabrication is well controlled) can be "unzipped", in order to produce GNRs with more uniform widths (12). Finally, GNRs can be added to SWNT random arrays, in order to improve the conductivity of TCFs (13). Future carbon-based films, devices and circuits will probably make extensive use of combinations of these two exceptional materials.