For devices based on graphene, non-refractory metallic electrodes of Ti/Au or Ti/Pd are frequently used. However, to reduce the contact resistance between metal electrodes and graphene, an annealing step in vacuum at a temperature of 700º C is generally required, which is incompatible with non-refractory metals [1]. On the other hand, refractory metal electrodes, such as tantalum nitride (TaN), can tolerate the annealing step but are less studied. Furthermore, the many techniques established to obtain graphene and to build high quality devices based on graphene are severely limited for large scale integration (LSI). Here we attempt a new approach to reach LSI. In our approach, graphene was obtained in large quantity from dielectrophoretic (DEP) assembly [2] of a few layers of insulating graphene oxide followed by thermal reduction at 450ºC. Next, for the metal contacts with graphene we used tantalum nitride (TaN) electrodes deposited by DC reactive sputtering in N₂/Ar environment. FET devices were obtained with the graphene sheets deposited between TaN electrodes, which are employed as source and drain, and laying on top of a TaO_x high-k gate dielectric deposited on an n+ silicon substrate used as the back electrode. To improve the contacts between graphene and TaN, ac signals were applied between electrodes, followed by annealing at 700° C in high vacuum to prevent electrode oxidation.

Scanning electronic microscopy and current-voltage (I-V) measurements were carried out to obtain the physical and electrical characteristics of our devices, respectively. The resistance values measured before and after annealing differ by five orders of magnitude, in agreement with [2], resulting in good contact characteristics.

In parallel, ab initio calculations were performed to investigate the nature of graphene/TaN interface employing two models for the interface. While this simulation effort is still ongoing, we have found that for the Ta/N-rich TaN surface and for the Ta-rich TaN surface the absence of the Dirac cone in the BL for each case is not due the BL roughness caused by the strong interaction with the TaN surface, especially for the Ta/N-rich surface, but due to metal surface states induced in the BL. Because in graphene the inter-layer separation is large, the electronic properties of the second graphene sheet are graphene-like, displaying a clear linear dispersion near the Dirac point.

[1] Souza, J. F., Development of Materials and Methods of Fabrication of Chemical/Biochemical Sensors based on Silicion and Carbon Nanostructures (ISFET, CNTFET and GraFET), PhD, University of Campinas, 2012;

[2] Burg B R et al, Appl. Phys, Lett. 94 053110 (2009);

DNA-functionalized graphene field-effect transistors (GFETs) have been shown to be capable of highly sensitive chemical vapor detection [1]. However, two major problems have so far prevented this sensor technology from being adopted in practical applications: (1) the lack of a scalable fabrication process and (2) an approach for improving the selectivity of the sensors. This work demonstrates methods to address both issues.

First, graphene FETs are fabricated on a 4-inch wafer scale through the use of LPCVD graphene growth and a photolithography fabrication processes. As an advance over previous GFET fabrication methods, we address the issue of contamination of the graphene surface by resist polymers [2] through the use of a removable protective overlayer film. Using this method, hundreds of GFET devices can be realized with mobilities exceeding 1000 cm^2/V*s under ambient conditions. The use of such a protection layer is of critical importance for graphene sensors where the cleanliness of the graphene surface is of concern.

We have previously shown using individual devices that the sensor response of a DNA/graphene FET depends on the strand sequence of the DNA functionalization layer [1]. In this work, we extend this approach to a large scale GFET array of approximately 50 devices. Each device is separately functionalized with one of at least 5 different ssDNA sequences. The sensor response of the array of devices is read out using electrical measurements of the GFET transfer characteristics. In our approach, the entire response of the array will be used to determine a unique “fingerprint” so as to enable discrimination between various chemical vapors of interest. In particular, this work will quantify the selectivity and sensitivity performance of the array in detecting certain biological compounds, such as those found human breath or secretions.

[1] Lu , (97), 083107 2010.

[2] Dan , (9), 1472-1475 2009.

The ideal structure of graphene is a two-dimensional hexagonal lattice. The presence of defects in graphene leads to a disruption of this structure and to changes in physical and chemical properties of graphene. Using Metropolis Monte Carlo method, the formation of linear defects and stable configurations of graphene and nanographene (which has a small finite size with free boundary atoms) have been studied. The interatomic interactions have been described by the Brenner's reactive empirical bond order potential which was parametrized specifically for carbon and hydrocarbon systems [1].

Nanographene has been modeled by 112 carbon atoms arranged in a flat plane and in the appropriate configuration for graphene with a lattice constant of 2.46 Å. To simulate a graphene the periodic conditions were superimposed on the boundary atoms of nanographene along its plane. The presence of linear defects has been modeled as point vacancies, which is located along the same line in the lattice. We studied their effects on changes in the lattice structure and binding (cohesive) energy of atoms. In particular, it was found that linear defects results in decreasing the cohesive energy of studied systems. We have found how the energy of cohesive varies on the number of point vacancies in a linear defect for graphene and nanographene.

[1] D.W. Brenner, O.A. Shenderova, J.A. Harrison, S.J. Stuart, B. Ni, S.B. Sinnot. “A second-generation reactive empirical bond order (REBO) potential energy expression for hydrocarbons”, J.Phys: Condens. Matter, 14, pp.783-802 (2002)

Graphene, a two-dimensional form of carbon with atoms arranged in a honeycomb lattice, has attracted enormous attention because of its unique electronic structures. To realize graphene-based electronics, modulation techniques of its electronic properties are indispensable. The most fundamental approach to tailor the electronic properties of graphene is doping. Nitrogen atoms can be doped into graphene using chemical vapor deposition or NH₃ plasma exposure [1,2]. It has been confirmed that the nitrogen-doped graphene exhibits an n-type behavior [1,2]. However, there exists little knowledge about the well-defined atomic arrangements of nitrogen-doped graphene. In this work, we have investigated the energetic stability of boron- or nitrogen-doped graphene using first-principle calculations within the local density functional theory.

We have calculated the interaction energy between two substitutional N atoms. A (4 x 4) supercell for graphene is employed, in which two C atoms are replaced by N atoms. The interaction energy has been found to decrease with increasing distance between N atoms. It should be noted that N atoms prefer to locate at the third nearest neighbors with each other, as shown in the previous report [3].

Next, we have investigated the interaction energy for the homogeneously-arranged N-doped graphene using (n x n) and (n√3 x n√3)-R30 unit cells (n: integer), where substitutional N atoms form triangle or honeycomb grids on the graphene basal plane. It has been found that the interaction energy decreases with decreasing density of N atoms. However, anomalous stabilization is confirmed for the honeycomb-arranged N-doped graphene with N:C=1:3. We will also report on the band structures of N-doped and B-doped graphene.

[1] D. Wei et al., Nano Lett. 9, 1752 (2009)

[2] Y.-C. Lin et al., Appl. Phys. Lett. 96, 133110 (2010)

[3] Z. Hou et al., Phys. Rev. B 85, 165439 (2012)