Electronic Fingerprints of DNA Bases on GrapheneTowfiq Ahmed,*,† Svetlana Kilina,‡ Tanmoy Das,§ Jason T. Haraldsen,§,∥ John J. Rehr,†

and Alexander V. Balatsky*,§,∥

†Department of Physics, University of Washington, Seattle Washington 98195, United States‡Department of Chemistry and Biochemistry, North Dakota State University, Fargo, North Dakota 58108, United States§Theoretical Division and ∥Center for Integrated Nanotechnologies, Los Alamos National Laboratory, Los Alamos, New Mexico87545, United States

*S Supporting Information

ABSTRACT: We calculate the electronic local density of states (LDOS) ofDNA nucleotide bases (A,C,G,T), deposited on graphene. We observesignificant base-dependent features in the LDOS in an energy range within afew electronvolts of the Fermi level. These features can serve as electronicfingerprints for the identification of individual bases in scanning tunnelingspectroscopy (STS) experiments that perform image and site dependentspectroscopy on biomolecules. Thus the fingerprints of DNA-graphene hybridstructures may provide an alternative route to DNA sequencing using STS.

KEYWORDS: STM, electronic DNA sequencing, graphene, tunneling conductance, DNA base fingerprints

The determination of the precise sequence of the fournucleotides [adenine (A), cytosine (C), guanine (G), andthymine (T)] in DNA molecules is an important goal for bothfundamental research interests as well as large number ofapplications in biomedical research,1 biotechnology,2,3 drugdelivery,4 and biomaterial growth.5 However, conventionalapproaches are generally complex and expensive.6 Numerousexperimental and theoretical7−9 attempts have been made toimprove such determinations. For example, efforts have beenmade to develop efficient techniques based on high-resolutionmicroscopic10,11 and spectroscopic12 probes that can yielddirect fingerprints of these biomolecules. However, thesetechniques require the biomolecules to be deposited on ahost substrate in an ultrahigh vacuum environment. Difficultiesin preparing high-quality samples and in obtaining reproduciblemeasurements have limited the utility of these approaches.13

For example, the electronic fingerprints of a given DNA basecan vary dramatically even for subtle changes in the relativeangle of a base with respect the substrate.14 On the other hand,a recent scanning tunneling microscopy/spectroscopy (STM/S) study12 by Tanaka et al. has shown that the guanine base ofDNA on a Cu(111) surface always exhibits a strong tunnelingpeak around a fixed bias voltage of −1.6 eV, thus providing areliable marker in tunneling measurements. This observationhas led to a significant advance in the field, since it opens anopportunity for electronic identifications of all bases viatunneling conductance from a local probe. However, thelocalized d-states and the dangling-bonds near the surface of thebulk Cu are among many effects that can complicate the

implementation of this approach, thus suggesting theimportance of finding a more suitable substrate.In this Letter, we show that an attractive substrate to

consider is graphene, a purely two-dimensional sheet of carbonatoms arranged in a honeycomb lattice.15 Among its uniqueproperties, graphene possesses linearly dispersed Diracquasiparticles with a semimetal density of states near theFermi level.16,15 Additionally, graphene combines a conductingsurface with remarkable mechanical strength. Also strong π−πinteractions between DNA bases and conjugated carbons ingraphene should favor mostly plane orientations of DNA baseswith respect to the surface17 and thus results in relativelyhomogeneous orientations of bases in the DNA strands when itadsorbs on graphene and offers optimal conditions for STMmeasurements. These characteristics make this novel materialan excellent surface for studying adsorbates of various organicmacromolecules. Several successful implementations of gra-phene have already been reported recently in electronic devicesand biomedical and bioassay applications.18−20 We aim todevelop a detailed understanding of the interaction andadsorption between graphene and biomolecules such as theDNA bases including local electronic structure, using anapproach that goes beyond the structural analysis of suchhybrid systems.

Received: November 14, 2011Revised: December 23, 2011Published: January 18, 2012

Letter

pubs.acs.org/NanoLett

© 2012 American Chemical Society 927 dx.doi.org/10.1021/nl2039315 | Nano Lett. 2012, 12, 927−931

pubs.acs.org/NanoLett

To achieve this goal, we have carried out extensive DFT-based, first-principles, numerical simulations of the electroniclocal densities of states (LDOS) of all four DNA bases ongraphene. Considering only the short-range interactionbetween the DNA bases and graphene, the van der Waalsinteraction was not included in our calculation. It waspreviously demonstrated that on distances of 2−3 Å, thelong-range dispersion interaction correction becomes negli-gible21,22,14 (see discussion in the Supporting Information). Wefind that the electronic LDOS of the nitrogen atom can serve asan electronic fingerprint of a particular DNA base. That is, wefind several distinguishing and dominant features in the LDOSthat can identify each base. This finding can be used inconjunction with STM/S measurements of atomic resolutionthat image and probe the local density of states (LDOS) ofsurfaces, as illustrated in Figure 1. In particular, (i) we deducethat the local chemical environment and the graphene baseleads to new electronic states and additional features in LDOSinside the parent highest occupied molecular orbital (HOMO)and lowest unoccupied molecular orbital (LUMO) insulatinggap structure for all bases are discussed below. (ii) For eachbase, the peak positions and peak heights exhibit a characteristicevolution as a function of their orientation with respect to thegraphene sheet. (iii) We also investigated the similarities anddifferences of the LDOS of DNA bases by placing the grapheneon the commonly used SiC substrate. The STM topographysimulations also add important insights into the character-ization of charge distributions and hybridization of statesbetween bases as well between base and graphene. Takentogether, we demonstrate that all four bases of DNA adsorbedon a graphene substrate can be differentiated quantitatively.This suggests the possibility of an efficient and cost-effectivesequencing approach.In an STS experiment, measurements are done in two steps.

First, the tip scans over the surface to search for the locations ofthe maximum transmission currents at a fixed voltage (Figure1a) and constructs the topographic image. In the next step, thevoltage is varied over an energy range by keeping the tipposition fixed, and the differential current dI/dV, which

determines the local electronic structure (LDOS) of thesample, is then measured. A schematic illustration is presentedin Figure 1b,c.We first considered the behavior of isolated DNA base

molecules using DFT calculations. In Figure 2a, the molecular

energy levels of each base are reproduced, which are inagreement with previous calculations.14 These results confirmthat all nucleobases have a large HOMO−LUMO gap (Eg ≈ 5eV). Near the energy gap, all electronic features are qualitativelysimilar for all four bases. Such small variations in the electronicstructure of DNA bases might bring in to question whether

Figure 1. (a) Schematic illustration of STM experiment that can extract topographical image and dI/dV information of a sample lying on grapheneplane. (b,c) Illustrations schematically demonstrate tunneling between the tip and sample. On the basis of positive (b) or negative (c) bias voltage,electrons can transmit to or from the graphene and map the unoccupied or occupied LDOS correspondingly of the sample lying on the plane.

Figure 2. (a) Molecular energy levels of each isolated DNA basemolecule. (b) Upper panel shows LDOS of carbon atom in puregraphene and lower panel shows nitrogen peak LDOS in isolatedDNA bases. (c) Integrated (−3.0 eV to EF = 0) partial charge densityof DNA bases on graphene plane. Base types are organized in columnsand angles between base-ring and graphene plane are varied alongrows as indicated. This simulation of STM topography is done usingHIVE-STM23 code. (d) Molecular orientations of DNA basescorresponding to the images in panel c are displayed.

Nano Letters Letter

dx.doi.org/10.1021/nl2039315 | Nano Lett. 2012, 12, 927−931928

STM is able to resolve fingerprints features of DNA bases. Onthe other hand, the presence of a substrate and subsequentinteractions significantly affect the alignment of electronic levelsof bases compared to the isolated molecules as was theoreticallypredicted for adsorbed DNA bases on the Cu surface.14 In thisLetter, we have considered an experimental setup with freeze-dried conditions, that will result in significantly diminishednoise due to the changes of molecular configuration and due tolow temperature.To investigate the effect of graphene on the DNA bases, we

first calculate the LDOS of nitrogen atoms in the isolated DNAbases and the LDOS of carbon atoms in the bare graphene asshown in Figure 2b. For the bare graphene, our calculationreproduces the well-known and distinct electronic feature ofgraphene at EF, the presence of a Dirac cone,

24 as can be seenin Figure 2b.STM topographical images also provide important geometric

information. This is particularly important in systems wheremolecules of interest can interact with the surface at manydifferent orientations. Depending on the backbone config-uration, DNA bases in a single stranded ssDNA can beadsorbed onto the graphene surface at different angularorientations and further complicate the electronic identification.Thus, we constructed supercells where each DNA base lies onthe graphene surface with four possible angles (Figure 2d). Inorder to simulate a STM topographic image theoretically, weintegrated our DFT based partial charge density from −3.0 eVto EF using HIVE-STM

23 (Figure 2c). The average distancebetween the DNA bases and graphene plane was fixed at 3.0 Å.In the remainder of this study, we primarily focus on the dI/

dV (or LDOS) of DNA bases on graphene. Since nitrogen is acommon constituent of all four bases with the most diffuseorbitals, we analyze the dependence of the LDOS on both theDNA base and its spacial orientation when a featureless tip isplaced directly above a nitrogen atom. The calculated LDOS is

shown for isolated DNA bases (Figure 2b) and adsorbed bases(Figure 3). For the adsorbed bases, we identify the behavior ofthe dominant peak features within the range of [−3.0; 3.0] eV,and their dependence on the angle between the base-ring planeand graphene plane. These peaks are marked by the coloredshaded regions in the curves of Figure 3a−d. In Figure 3e, wesummarize these angle dependent results for the positive andnegative bias energies for all four bases.By examining Figure 3, one can recognize the dominant STS

peak features, except for a few cases where the experimentalresolution can play a critical role. For example, simulated LDOS(Figure 3e) shows distinctive features in the positive low energyregion (unoccupied states) for cytosine (the lowest among all)and guanine (the highest among all), independent of the baseorientation.Consequently, if several LDOS peaks are compared (for

example, both occupied and unoccupied states close to theFermi energy), the identification of the bases can be achieved.For the above example, the supercell consisted of a set of ∼130atoms, and a full geometry optimization on such systems iscomputationally expensive. To assess the interaction betweenthe graphene and the bases, we reduced the supercell andperformed a full geometry optimization using VASP25−27

software package. The details of this calculation are given in theSupporting Information.Our calculations reveal the dependence of the binding

energies on the angular orientation for a given base. Forexample, the 0° configuration of adenine is more bound thanthe 45° one. The preference in plane alignments of bases withrespect to the graphene is expected, since parallel orientation ofbases along the graphene surface allows better overlap in π-orbitals between the aromatic rings of DNA bases and aromaticcarbons in graphene increasing the π−π stacking. Thisinteraction results in relatively homogeneous and planarorientations of bases in the DNA strand adsorbed on the

Figure 3. (a) Solid lines are nitrogen LDOS in adenine adsorbed on graphene with different angles, and the shaded regions show the dominant peakfeatures near EF on both positive and negative energy side, 0 eV corresponds to Fermi energy. (b−d) Similar LDOS curves for cytosine, guanine andthymine correspondingly. (e) Dominant peak positions near EF for different angles and bases. Positions of the peaks are plotted along y-axis andangles along x-axis. The relative size of the symbols reflects the relative peak intensity for each case.

Nano Letters Letter

dx.doi.org/10.1021/nl2039315 | Nano Lett. 2012, 12, 927−931929

graphene, which offers favorable conditions for STM experi-ment and better resolution of dI/dV features. At small angles(

(9) Min, S. K.; Kim, W. Y.; Cho, Y.; Kim, K. S. Nat. Nanotechnol.2011, 6, 162−165.(10) Yarotski, D. A.; Kilina, S. V.; Talin, A. A.; Tretiak, S.; Prezhdo,O. V.; Balatsky, A. V.; Taylor, A. J. Nano Lett. 2009, 9, 12−17.(11) Seong, G. H.; Niimi, T.; Yanagida, Y.; Kobatake, E.; Aizawa, M.Anal. Chem. 2000, 72, 1288−1293, PMID: 10740872(12) Tanaka, H.; Kawai, T. Nat. Nanotechnol. 2009, 4, 518−522.(13) Clemmer, C.; Beebe, T. Science 1991, 251, 640−642.(14) Kilina, S.; Tretiak, S.; Yarotski, D. A.; Zhu, J.-X.; Modine, N.;Taylor, A.; Balatsky, A. V. J. Phys. Chem. C 2007, 111, 14541−14551.(15) Castro Neto, A. H.; Guinea, F.; Peres, N. M. R.; Novoselov, K.S.; Geim, A. K. Rev. Mod. Phys. 2009, 81, 109−162.(16) Grüneis, A.; Attaccalite, C.; Rubio, A.; Vyalikh, D. V.;Molodtsov, S. L.; Fink, J.; Follath, R.; Eberhardt, W.; Büchner, B.;Pichler, T. Phys. Rev. B 2009, 80, 075431.(17) Akca, S.; Foroughi, A.; Frochtzwajg, D.; Postma, H. W. C. PLoSOne 2011, 6, e18442.(18) Geim, A. K.; Novoselov, K. S. Nat. Mater. 2007, 6, 183−191.(19) Novoselov, K. S.; Geim, A. K.; Morozov, S. V.; Jiang, D.; Zhang,Y.; Dubonos, S. V.; Grigorieva, I. V.; Firsov, A. A. Science 2004, 306,666−669.(20) Su, Q.; Pang, S.; Alijani, V.; Li, C.; Feng, X.; MÃijllen, K. Adv.Mater. 2009, 21, 3191−3195.(21) Henkelman, G.; Jońsson, H. Phys. Rev. Lett. 2001, 86, 664−667.(22) Jurecka, P.; Sponer, J.; Cerny, J.; Hobza, P. Phys. Chem. Chem.Phys. 2006, 8, 1985−1993.(23) Vanpoucke, D. E.; Brocks, G. Phys. Rev. B: Condens. Matter2008, 77, 241308.(24) Siegel, D. A.; Park, C.-H.; Hwang, C.; Deslippe, J.; Fedorov, A.V.; Louie, S. G.; Lanzara, A. Proc. Natl. Acad. Sci. U.S.A. 2011,DOI: 10.1073/pnas.1100242108.(25) Kresse, G.; Furthmüller, J. Phys. Rev. B 1996, 54, 11169−11186.(26) Vanderbilt, D. Phys. Rev. B 1990, 41, 7892−7895.(27) Perdew, J. P.; Zunger, A. Phys. Rev. B 1981, 23, 5048−5079.

Nano Letters Letter

dx.doi.org/10.1021/nl2039315 | Nano Lett. 2012, 12, 927−931931