THE JOURNAL OF CHEMICAL PHYSICS 128, 041103 共2008兲
Enhancement of the transverse conductance in DNA nucleotides
Vincent Meuniera兲 and Predrag S. Krstić
Oak Ridge National Laboratory, P.O. Box 2008, Oak Ridge, Tennessee 37831, USA
共Received 18 October 2007; accepted 26 December 2007; published online 29 January 2008兲
We theoretically study the electron transport properties of DNA nucleotides placed in the gap
between two single-wall carbon nanotubes capped or terminated with H or N. We show that in the
case of C-cap and H-termination the current at low electric bias is dominated by nonresonant
tunneling, similarly to the cases of gold electrodes. In nitrogen-terminated nanotube electrodes, the
nature of current is primarily quasiresonant tunneling and is increased by several orders of
magnitude. We discuss the consequence of our result on the possibility of recognition at the level of
single molecule. © 2008 American Institute of Physics. 关DOI: 10.1063/1.2835350兴
The possibility of genome sequencing by measuring the
transverse conductivity upon applied dc bias while a DNA
strand is translocated through a nanogap or nanopore has
recently been the subject of some debates in the theoretical
considerations,1–6 as well as of significant experimental
activities.4 The origin of the debate stems from the fact that
the Fermi energy of gold electrodes at 0 K is rather far
共⬃2 eV兲 from the molecular eigenlevels of the DNA
nucleotides.1 The electronic transport is therefore dominated
by nonresonant tunneling, which is highly dependent on the
difficult-to-control relative geometry between the molecule
and electrodes, while it is weakly dependent on the electronic structure of the molecule. In addition, this geometry is
influenced by aqueous and electrolytic environment, thermal
phenomena, and the effects of the applied transverse as well
as longitudinal 共translocating兲 electric fields. With a low
transverse voltage bias and picoampere and subpicoampere
tunneling currents the main difficulty remains in the poor
signal-to-noise ratio, weakening the predictive power of distinguishing various nucleotides, or detecting their
presence.1,2 These uncertainties are the essence of the recent
controversy in the literature.5,6 Higher bias is unacceptable in
an actual device realization: besides other possible destructive and nonlinear effects, electric forces at the negatively
charged backbone of the DNA move the molecule toward the
anode, disabling the translocation.
In this letter, we propose a new two-terminal setup for
molecular conductance measurements, based on the use of
specially prepared carbon nanotube 共CNT兲 electrodes, Fig. 1.
This approach enables a large, quasiresonant enhancement of
the current at low bias, with the possibility of removing
drawbacks of previous setups. Up to now, conductance measurements have been typically performed using standard metallic probes, such as gold.7 Unfortunately when DNA segments are sandwiched between this type of usually large
cross-section electrodes, the structural deformations at the
interface between electrodes and the base pairs can cause
unacceptable variations in the measured current.8,9 The intera兲
Author to whom correspondence should be addressed. Electronic mail:
[email protected].
0021-9606/2008/128共4兲/041103/4/$23.00
face sensitivity in quantum transport is not only limited to
measuring the current across DNA, it is a universal effect
that makes measurement in single molecule particularly difficult. For that reason, there is currently a strong interest in
the development of experimental apparatus that will alleviate
the difficulty of controlling the coupling between the elec-
FIG. 1. 共Color兲 Configurations of the CNT leads and molecules for the
calculation of electronic transmission. 关共a兲 and 共b兲兴 Empty nanogap between
carbon-capped and H-terminated CNTs. 关共c兲–共f兲兴 Detailed structure of nanogap created between N-terminated CNT with nucleotides shown explicitly
for adenine, cytosine, guanine, and thymine, respectively. In each case, the
DNA bases include a sugar-phosphate group as explained in Ref. 1. The
distance between the CNT tips was kept at 1.5 nm. In this figure, all lead
layers of the extended molecule used in the calculation are shown. Nitrogen,
carbon, hydrogen, oxygen, and phosphorus atoms are shown in yellow, blue,
white, red, and gold, respectively.
128, 041103-1
© 2008 American Institute of Physics
041103-2
V. Meunier and P. S. Krstić
trodes and the molecules. One attractive idea is to develop a
system where the coupling between the molecule and the
electrodes is better localized, in such a way as to ensure
higher reproducibility of measured current-voltage curves.
For instance, advanced two-probe electric systems have been
devised to measure geometrical and electronic properties of
DNA and DNA derivatives.10 In this device, the tip is replaced by a carbon nanotube in order to probe nanometerscale samples, since the probe must have a radius of curvature smaller than the size of the samples. Carbon nanotubes
have been under the spotlight since their discovery in 1990
by Iijima,11 sparking intense research in various fields of
materials sciences.12 It has been repeatedly shown that CNTbased nanotips are powerful tools to manipulate biological
molecules and to help decipher the relationship between
structure and function, in part owing to their high aspect ratio
that allows for imaging with higher spatial resolution.13,14
The appeal for carbon nanotubes does not only stem from
their unique morphology but also because their ends can be
conveniently functionalized in order to improve desired
properties, such as increased coupling or decreased work
function for field emission purposes.15–17 A similar idea was
recently developed where a CNT was used as an electrode
for dielectrophoretic trapping of DNA molecules, in order to
make it possible to apply relatively low trapping voltages
while still achieving high enough field gradient for trapping
purposes.18 Another report shows that single DNA chains can
be chemically grafted onto aligned CNT electrodes, thereby
paving the way toward the development of DNA-CNT sensors of high sensitivity and selectivity.19 Previous experimental achievements similar to those cited above most exclusively involve covalent bonding between the CNT tip and the
DNA strand. Here, we are interested in the lateral conductivity of DNA single nucleotides sandwiched between, but not
covalently attached to, CNT electrodes. The reason is that the
conductivity measurement is expected to be used as a sequencing method where the DNA strand can easily be
threaded inside the nanogap created between the nanotube
electrodes. At first glance, the geometry studied here is quite
appealing: given the size of the nanotube cross section, a
satisfactory resolution along the strand could be achieved.
Unfortunately, since we seek noncovalent bonding to allow
threading, the resulting current is probably too low to be
practical. As shown below, this is indeed an insurmountable
drawback for conventional 关i.e., capped or hydrogensaturated edges, Figs. 1共a兲 and 1共b兲兴 nanotubes.
It has been established that CNT ends can be modified
chemically.20,21 End doping can be done during growth as
long as the dopant has a surfactant behavior. In that respect a
nitrogen atom is a very attractive dopant. First, it was demonstrated theoretically and confirmed experimentally that nitrogen behaves as a surfactant during CNT growth, i.e., it
covalently attaches to the CNT outer rim, and does not get
buried inside the nanotube lattice, under growth conditions.22
Armchair CNTs terminated with nitrogen atoms have been
shown to be stable, at least when the N occupies two coordinated positions.22 In that case the presence of nitrogen does
not disrupt the sp2 bonding of the tube. In addition, the edge
saturation with N maximizes the number of double N v N
J. Chem. Phys. 128, 041103 共2008兲
bonds. Second, the presence of nitrogen atoms at the end cap
leads to a shift in the energy level of localized states at the
tip toward the Fermi level and to an overall increase of the
density of states around the Fermi level.23,24 For those reasons, we have considered N-doped 共5,5兲 nanotubes as potentially improved electrodes to measure increased currents
through the DNA nucleotides in the nanogap, Figs. 1共c兲–1共f兲.
Using density functional theory 共DFT兲, we demonstrate
that single-wall 共metallic兲 CNTs, properly terminated, can
favorably replace metal electrodes, supporting the transverse
electron transport across the DNA. The explicit circuitry engineering of connecting the nanotube electrodes into an external battery is not considered here, as was not considered in
previous works with metal electrodes.1–4 In actual devices,
nanotubes will eventually need to be connected to metallic
electrodes attached to an external battery. Band realignment
and charge injection in the vicinity of that “external” junction, as well as other imperfections in the external circuitry
have an effect on the magnitude of the direct current. However, nanotube leads are typically rather long 共tens of nanometers兲 and screening takes place over a distance much shorter
than the nanotube branch, supporting our assumption of the
infinite leads. Therefore this effect does not depend on the
nanotube chemical ending close to the DNA nucleotide. The
different behaviors for different endings will therefore be
preserved 共though the amplitude of the current might be
modified兲. From a different perspective, we note that the effect of nanotube junction to external electrodes will not be
expressed in a conventional four point measurement since it
eliminates the effect of that type of contacts. To illustrate the
enhancement of the conductivity with N doping, we calculate
electron transport for all four types of the DNA nucleotides
关adenine 共A兲, thymine 共T兲, cytosine 共C兲, and guanine 共G兲兴,
with C-capped, and H and N nanotube terminations, in geometry shown in Fig. 1. All the results in this letter correspond to the Landauer approach,25 which establishes that the
electron transmission probability, T, is proportional to the
molecular electronic conductance. T includes the detailed,
quantum mechanical description of the molecule and its interaction with the leads. Assuming nearly thermodynamic
equilibrium and a small voltage drop V across the system
共from 0 to 500 mV兲, the so-called linear-response, equilibrium regime is known to yield quantitative results compared
to a full nonequilibrium approach where the Poisson’s equation is solved self-consistently.26 Within these assumptions,
we describe the electronic structure of the system with the
self-consistent DFT calculation for the ground state. Allelectron DFT calculations were performed using the quantum chemistry package NWCHEM,27 with the 3-21G contracted Gaussian basis set.
All of our results presented in the paper have been carefully checked with three different exchange correlation functional, i.e., with local density approximation 共LDA兲,28 generalized gradient approximation 共GGA兲,29 and hybrid
methods 共B3LYP兲,30 remarkably all leading to qualitatively
the same conclusions, in spite of possible weak coupling
present here. The numerical results shown in this letter correspond to the GGA functional, motivated by the fact that
041103-3
J. Chem. Phys. 128, 041103 共2008兲
Transverse conductance in DNA
FIG. 2. 共Color online兲 Comparison of the I-V characteristics of guanine with
N-, C-, and H-terminated CNT leads.
GGA is a good compromise for extended electronic systems
and molecular systems.
We carefully chose subsystems containing a DNA nucleotide and a large part of the CNT electrodes, with the appropriate terminations. To describe the open boundary conditions appropriate for the leads, we used the Green’s function
matching method in conjunction with the generalized tightbinding approach to compute the transmission function
T.31–33 The integral of T共E , V兲 over the band energy E determines the current response I to the applied voltage V. The
transmission function T depends not only on the electronic
structure of the nucleotide but also on a number of other
factors, e.g., the strength of the coupling between the base
and the leads, which is a function of the base-lead geometry.
Specifically, T is very sensitive to the energy matching of the
asymptotic Bloch channels in the leads with the energy levels of the base, deformed 共i.e., shifted and broadened兲 by the
coupling with the leads and adapted to the chemical potential
drop across the molecule, resulting from the applied bias.
Our approach takes all of these effects into account at the
DFT level.
Typical I-V characteristics for various lead terminations
are illustrated with the example of guanine in Fig. 2. The
current response for the N-terminated leads is found to be
two to six orders of magnitude larger than that obtained with
C- and H-terminations in the whole range of the considered
biases. It is important to note that the current in the N-case
reaches nanoampere values at about 0.4 V, while it stays in
subpicoampere, or picoampere range for C- and
H-terminations, comparable to the values obtained using
gold
electrodes
with
similar
electrode-nucleotide
geometries.1 The obtained enhancement might lead to an increase in the signal-to-noise ratio in conductance measurement of the DNA nucleotides, if other sources of uncertainty
are unchanged.
FIG. 3. 共Color兲 Spectra of eigenenergies of the extended molecule, for various CNT terminations and various nucleotides. Black thin lines are orbitals
located at the CNT leads, while red-colored thick lines represent the orbitals
located at DNA nucleotides. The Fermi energy is set to zero 共dashed horizontal line兲 in each case.
Strong enhancement of the low-electric bias response
with N-termination is found for all DNA nucleotides, as illustrated for two voltages in Table I. The enhancement factors also increase orders of magnitude with increase of voltage from 0.1 to 0.5 V. All nucleotides have an enhancement
factor to the H-terminated leads, IN / IH, ranging between 105
and 106 for higher of the considered voltages. This factor to
the C-termination, IN / IC, is more varying, between 107 for
adenine, down to 103 for guanine.
While the leads define the boundary conditions and supply the electrons around the Fermi energy EF, the physical
mechanisms of transport across the interlead gap mainly depend upon the electronic structure of the molecule placed in
the gap, while its coupling to the leads defines the tunneling
characteristics of the junction. The existence of the electronic
states localized in the gap, energetically close to the Fermi
energy, is of a decisive importance for the overall conductance. In absence of such states, the mechanism of electron
transport is dominated at low biases by the nonresonant tunneling, causing a significant suppression of conductance.
Figure 3 shows the eigenenergy spectra of the extended molecules of Fig. 1. The strong presence of states located mainly
at nitrogen and oxygen atoms of a nucleotide, extending
across the gap, is obvious for N-terminated leads, for all of
the nucleotides. The fact that no such states are present close
to EF in the case of the carbon- and hydrogen-terminations
explains the large increase of current in nitrogen-terminated
CNTs.
Our results can be further analyzed in terms of the improved coupling between the N-saturated edges of the nano-
TABLE I. Current enhancements for various CNT terminations and for four DNA nuelotide types, with 0.1 and
0.5 V transverse biases. The format a共b兲 means a10b.
Bias 共V兲
A
G
C
T
IN / TC
0.1
0.5
6.3 共3兲
1.1共7兲
1.4 共3兲
8.8共3兲
7.0 共1兲
4.1共4兲
5.6共3兲
2.8共6兲
IN / IH
0.1
0.5
1.8共2兲
2.0共5兲
1.1共3兲
1.3共6兲
6.1共2兲
2.7共5兲
1.2共4兲
5.5共6兲
041103-4
J. Chem. Phys. 128, 041103 共2008兲
V. Meunier and P. S. Krstić
tube with the heterocylic compounds present at one end of
each nucleotide, since they share chemical similarity. DNA
bases can be classified into two types: Adenine and guanine
are fused five- and six-membered heterocyclic compounds
called purines, while cytosine and thymine are six-membered
rings called pyrimidines. In the nanogap geometries used in
the present work, G and A nucleotides share similar current
profiles. This property is compatible with the chemical nature of the base. It follows that for all terminations these two
molecules carry the largest current. C and T nucleotides yield
a quite smaller current response. Again, this can be understood from the fact that C and T are pyrimidine compounds,
i.e., they have a single nitrogen-carbon heterocyclic group
which can account for a lower coupling than in the case of
purine bases. This feature makes the N-terminated CNT an
excellent sensor for the 共A,G兲 and 共C,T兲 groups of the DNA
bases. The issue of distinguishability within each of these
groups remains an open topic requiring further studies and
improvements.
Future work is required to understand the extent to
which the described current enhancement survives the refinement of the model systems toward more realistic representations of the DNA molecule in an aqueous and electrolytic
environment, averaging over the range of possible conformations of DNA while translocating through the nanogap
formed by the leads, and consideration of background Faraday currents, as well as other contributions to observed
signal-to-noise ratios. The choice of N-, C-, and
H-terminations of the CNT leads is guided by a possibility of
matching with atomic constituents of the DNA. We showed
that the presence of N establishes a natural connection with
purine and pyrimidines groups. Remarkably, the obtained selective enhancement of the current response, following from
transition from regime of nonresonant to the quasiresonant
tunneling, is not only limited to the DNA-related molecules.
The local chemistry at the tip of the electrodes has a dramatic
effect on the coupling with the molecule and functionalized
CNT ends offer interesting possibilities for molecular recognition using CNTs.
We would like to thank R. Zikic and B. G. Sumpter for
fruitful discussions. This research was supported by the U.S.
National Human Genome Research Institute of the National
Institutes of Health under Grant No. 1 R21 HG003578-01,
by U.S. Department of Energy 共DOE兲 Office of Fusion Energy Sciences 共P.S.K.兲, and by Division of Materials Sciences and Engineering 共V.M.兲 at ORNL managed by a UTBattelle for the U.S. DOE under Contract No. DEAC0500OR22725. A portion of this research was conducted at the
Center for Nanophase Materials Sciences, which is spon-
sored at ORNL by the Division of Scientific User Facilities,
U.S. DOE.
1
R. Zikic, P. S. Krstić, X.-G. Zhang, M. Fuentes-Cabrera, J. Wells, and X.
Zhao, Phys. Rev. E 74, 011919 共2006兲.
2
J. Lagerqvist, M. Zwolak, and M. Di Ventra, Nano Lett. 6, 779 共2006兲.
3
M. Zwolak and M. Di Ventra, Nano Lett. 5, 421 共2005兲.
4
Review of M. Zwolak and M. Di Ventra, e-print contains an extensive list
of references to recent experimental works, arXiv:cond-mat/07082724.
5
J. Lagerqvist, M. Zwolak, and M. Di Ventra, Phys. Rev. E 76, 3 共2007兲.
6
R. Zikic, P. S. Krstić, X. G. Zhang, M. Fuentes-Cabrera, J. Wells, and X.
Zhao, Phys. Rev. E 76, 2 共2007兲.
7
M. A. Reed, C. Zhou, C. J. Muller, T. P. Burgin, and J. M. Tour, Science
278, 252 共1997兲.
8
K. Tagami, T. Matsumoto, T. Kawai, and M. Tsukada, Jpn. J. Appl. Phys.,
Part 1 42, 5887 共2003兲.
9
K. Tagami, L. G. Wang, and M. Tsukada, Nano Lett. 4, 209 共2004兲.
10
K. Shimotani, T. Shigematsu, C. Manabe, H. Watanabe, and M. Shimizu,
J. Chem. Phys. 118, 8016 共2003兲.
11
S. Iijima, Nature 共London兲 354, 56 共1991兲.
12
J. Bernholc, D. Brenner, M. Buongiorno Nardelli, V. Meunier, and C.
Roland, Annu. Rev. Mater. Res. 32, 347 共2002兲.
13
L. Q. Guo, R. Wang, H. M. Xu, and J. Liang, Physica E 共Amsterdam兲 27,
240 共2005兲.
14
L. Q. Guo, J. Liang, S. Dong, Z. W. Xu, and Q. L. Zhao, Appl. Surf. Sci.
228, 53 共2004兲.
15
J. C. Charlier, M. Terrones, M. Baxendale, V. Meunier, T. Zacharia, N. L.
Rupesinghe, W. K. Hsu, N. Grobert, H. Terrones, and G. A. J. Amaratunga, Nano Lett. 2, 1191 共2002兲.
16
V. Meunier, C. Roland, J. Bernholc, and M. Buongiorno Nardelli, Appl.
Phys. Lett. 81, 46 共2002兲.
17
A. Maiti, J. Andzelm, N. Tanpipat, and P. von Allmen, Phys. Rev. Lett.
87, 155502 共2001兲.
18
S. Tuukkanen, J. J. Toppari, A. Kuzyk, L. Hirviniemi, V. P. Hytonen, T.
Ihalainen, and P. Torma, Nano Lett. 6, 1339 共2006兲.
19
P. G. He and L. M. Dai, Chem. Commun. 共Cambridge兲 2004, 348 共2004兲.
20
M. Majumder, N. Chopra, and B. J. Hinds, J. Am. Chem. Soc. 127, 9062
共2005兲.
21
S. S. Wong, E. Joselevich, A. T. Woolley, C. L. Cheung, and C. M.
Lieber, Nature 共London兲 394, 52 共1998兲.
22
B. G. Sumpter, V. Meunier, J. M. Romo-Herrera, E. Cruz-Silva, D. A.
Cullen, H. Terrones, D. J. Smith, and M. Terrones, ACS Nano 1, 369
共2007兲.
23
H. S. Ahn, K. R. Lee, D. Y. Kim, and S. W. Han, Appl. Phys. Lett. 88,
093122 共2006兲.
24
L. Qiao, W. T. Zheng, H. Xu, L. Zhang, and Q. Jiang, J. Chem. Phys.
126, 164702 共2007兲.
25
R. Landauer, Philos. Mag. 21, 863 共1970兲.
26
W. C. Lu, V. Meunier, and J. Bernholc, Phys. Rev. Lett. 95, 206805
共2005兲.
27
R. A. Kendall, E. Apra, D. E. Bernholdt, E. J. Bylaska, M. Dupuis, G. I.
Fann, R. J. Harrison, J. L. Ju, J. A. Nichols, J. Nieplocha, T. P. Straatsma,
T. L. Windus, and A. T. Wong, Comput. Phys. Commun. 128, 260
共2000兲.
28
S. H. Vosko, L. Wilk, and M. Nusair, Can. J. Phys. 58, 1200 共1980兲.
29
J. P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. Lett. 77, 3865
共1996兲.
30
A. D. Becke, J. Chem. Phys. 98, 5648 共1993兲.
31
V. Meunier and B. G. Sumpter, J. Chem. Phys. 123, 024705 共2005兲.
32
P. S. Krstic, X. G. Zhang, and W. H. Butler, Phys. Rev. B 66, 205319
共2002兲.
33
M. Buongiorno Nardelli, Phys. Rev. B 60, 7828 共1999兲.