We measure the channel potential of a graphene transistor using a scanning photocurrent imaging technique. We show that at a certain gate bias, the impact of the metal on the channel potential profile extends into the channel for more than one-third of the total channel length from both source and drain sides; hence, most of the channel is affected by the metal. The potential barrier between the metal-controlled graphene and bulk graphene channel is also measured at various gate biases. As the gate bias exceeds the Dirac point voltage, VDirac, the original p-type graphene channel turns into a p-n-p channel. When light is focused on the p-n junctions, an impressive external responsivity of 0.001 A/W is achieved, given that only a single layer of atoms are involved in photon detection.

 These authors contributed equally to this work. * avouris@us.ibm.com, 914-945-2722 We investigate polyethylene imine and diazonium salts as stable, complementary dopants on graphene.Transport in graphene devices doped with these molecules exhibits asymmetry in electron and hole conductance. The conductance of one carrier is preserved, while the conductance of the other carrier decreases. Simulations based on nonequilibrium Green's function formalism suggest that the origin of this asymmetry is imbalanced carrier injection from the graphene electrodes caused by misalignment of the electrode and channel neutrality points.

The objective of this paper is to describe a simple phenomenological approach for including incoherent dephasing processes in quantum transport models. The presented illustrative numerical results show this model provides the flexibility of adjusting the degree of phase and momentum relaxation independently that is not currently available in mesoscopic physics and in device simulations while retaining the simplicity of other phenomenological models.Introduction: Although models for coherent quantum transport are fairly well established 1 , approaches for including incoherent dephasing processes represent an active area of current research. Much of the work is motivated by the basic physics of conductance fluctuation in chaotic cavities 2 3 4 . However, there is also a strong motivation from an applied point of view 5 . For example, if we calculate the transmission T(E) through a 2D conductor (which could be the channel of a nanotransistor) with a random array of scatterers (due to defects or surface roughness), we see fluctuations as a function of energy which arise from quantum interference. Such fluctuations however are seldom observed at room temperature in real devices (when the device length is larger than phase relaxation length), because interference effects are destroyed by dephasing processes. Clearly, realistic quantum transport models for nanotransistors need to include such dephasing processes. A common way of including dephasing is through additional Buttiker probes 6 whose effects can then be modeled within the Landauer-Buttiker framework for coherent transport theory. Such probes typically introduce an additional resistance, since the probes themselves destroy momentum of itinerant electrons by partially reflecting them. By using a pair of unidirectional probes, Buttiker introduced phase relaxation without introducing momentum relaxation 7 . However, we are not aware of any work extending this to a continuous distribution of probes as needed to model a long conductor.

The objective of this paper is to point out that contact induced states can help explain the structure dependence of the minimum conductivity observed experimentally even if the samples were purely ballistic. Contact induced states are similar to the well-known metal induced gap states (MIGS) in metalsemiconductor Schottky junctions, which typically penetrate only a few atomic lengths into the semiconductor, while the depth of penetration decreases with increasing band gap. However, in graphene we find that these states penetrate a much longer distance of the order of the width of the contacts. As a result, ballistic graphene samples with a length less than their width can exhibit a resistance proportional to length that is not 'Ohmic' in origin, but arises from a reduced role of contact-induced states. While actual samples are probably not ballistic and involve scattering processes, our results show that these contact induced effects need to be taken into account in interpreting experiments and minimum conductivity depends strongly on the structure and configuration (two-vs. four-terminal) used.