The ever-growing demand for higher bandwidth broadband communication has driven transistor operation to higher and higher frequencies. However, achieving cut-o frequencies in the terahertz regime have been unsuccessful with the current state-of-the-art transistors exhibiting no better than 800 GHz. While the high-frequency transistor eld is dominated by III-V semiconductors, it has been proposed that graphene may be a competitive material. Graphene exhibits electron and hole mobilities orders of magnitude larger than conventional semiconductors and has an atomically thin form factor. Despite these benets, high-frequency graphene transis tors have yet to realize high-frequency characteristics better than III-V's.
This thesis expands on the current limitations of graphene transistors in terms of improved fabrication techniques (to achieve higher carrier mobilities and lower contact resistances) and fundamental, band structure limitations (like quantum capacitance and the zero energy band gap).
First, graphene, fully encapsulated in hexagonal boron-nitride crystals, transistors are fabricated with self-aligned source and drain contacts with sub-100 nm gate lengths. The encapsulation technique shields the graphene from the external environment so that graphene retains its intrinsic high mobility characteristic. In this short-channel regime, transport is determined to be ballistic with an injection velocity close to the Fermi velocity of graphene. However, the transconductance and output conductance are only 0.6 mS/mm and 0.3 mS/mm, respectively. This lack-luster performance is due to a relatively thick (3.5 nm) eective oxide thickness but also due to the eects of quantum capacitance which diminishes the total gate capacitance by up to 60%. Furthermore, the output conductance is increased due to the onset of hole conduction which leads to a second linear regime in the I-V characteristic. This is a direct consequence of graphene's zero energy band gap electronic structure. Finally, the source and drain contact resistances are large, which leads to poorer output current, transconductance and output conductance.
Second, improvement to the contact resistance is explored by means of using graphite as the contact metal to graphene. Since graphite is atomically smooth, a pristine graphite-graphene interface can be formed without grain asperities found in conventional metals. Graphite is also lattice matched to graphene and exhibits the same 60 symmetry. Consequently, it is discovered that the graphite-graphene contact resistance exhibits a 60 periodicity, with respect to crystal orientation. When the two lattices align, a contact resistivity under 10 Wmm² is observed. Furthermore, contact resistivity minima are observed at two of the commensurate angles of twisted bilayer graphene.
Though graphene transistor performance is band structure limited, it may still be possible to achieve competitive high-frequency operation by use of h-BN encapsulation and graphite contacts.
| Identifer | oai:union.ndltd.org:columbia.edu/oai:academiccommons.columbia.edu:10.7916/D8FB535W |
| Date | January 2016 |
| Creators | Chari, Tarun |
| Source Sets | Columbia University |
| Language | English |
| Detected Language | English |
| Type | Theses |
Page generated in 0.0022 seconds