Molecular rectifier composed of DNA with highrectification ratio enabled by intercalationCunlan Guo1†, Kun Wang1†, Elinor Zerah-Harush2, Joseph Hamill1, Bin Wang1, Yonatan Dubi2,3*and Bingqian Xu1*
The predictability, diversity and programmability of DNA make it a leading candidate for the design of functionalelectronic devices that use single molecules, yet its electron transport properties have not been fully elucidated. This isprimarily because of a poor understanding of how the structure of DNA determines its electron transport. Here, wedemonstrate a DNA-based molecular rectifier constructed by site-specific intercalation of small molecules (coralyne) intoa custom-designed 11-base-pair DNA duplex. Measured current–voltage curves of the DNA–coralyne molecular junctionshow unexpectedly large rectification with a rectification ratio of about 15 at 1.1 V, a counter-intuitive finding consideringthe seemingly symmetrical molecular structure of the junction. A non-equilibrium Green’s function-based model—parameterized by density functional theory calculations—revealed that the coralyne-induced spatial asymmetry in theelectron state distribution caused the observed rectification. This inherent asymmetry leads to changes in the coupling ofthe molecular HOMO−1 level to the electrodes when an external voltage is applied, resulting in an asymmetric changein transmission.
The field of molecular electronics, the goal of which is to func-tionally incorporate molecular components in electronicdevices1–3, has focused intensively on DNA. In addition to
the molecule’s high density of genetic information, its inherent struc-tural and molecular recognition properties render it ideal for mol-ecular electronics applications. In the past two decades DNA hastherefore attracted inordinate amounts of attention in the molecularelectronics and spintronics fields4–6 for its potential to transportcharge in molecular electronics applications such as DNA chips7,8.
Shown to be sequence-dependent, the conduction mechanism ofnative duplex DNA was found to be dominated by tunnelling whenthe guanine–cytosine (G–C) pairs are separated by three or feweradenine–thymine (A–T) pairs. If the number of A–T pairs isincreased, however, diffusive hopping becomes the main electronictransport mechanism9,10. To further complicate matters, researchersrecently demonstrated that DNA charge transport includes an inter-mediate tunnelling–hopping regime11. Achieving the goal of obtain-ing a functional electronic device based on DNA will thus requirethat its molecular structure be altered to obtain non-monotonicI–V behaviour. The conductivity of DNA was recently found to besensitive to subtle changes in DNA conformation caused by altera-tions in the ionic environment12 or by its modification with methyl-ation13,14 or with metal ions15,16. Remarkable in their own right,these achievements hint at the tantalizing possibility that the elec-tronic transport properties of DNA can be fine-tuned via structuralmodification for the development of a diversity of DNA-basedelectronic devices, especially an applicable DNA molecular rectifier.
First proposed by Aviram and Ratner in 1974 (ref. 17), the notionof a molecular rectifier continues to be a subject of intense interest inthe field of molecular electronics. To date, using conductancemeasuring methods based on self-assembled monolayers18,19
and single-molecule junctions20–22, rectification can be observedeither with asymmetric molecules23,24 or with inconsistentmolecule–electrode contacts25. Most rectifications have been
observed using small organic molecules, the structures of whichusually comprise electron-donor and electron-acceptor groupsseparated by an insulating group18,24.
Here, we report the successful creation of a DNA-based rectifier,accomplished by intercalating coralyne molecules into specificallydesigned duplex DNA. Electrical measurements of single DNA–coralyne complex molecular junctions made using the scanning tun-nelling microscopy break junction (STMBJ) technique (Fig. 1a)26,27
unexpectedly revealed rectification with a high rectification ratioof around 15 at 1.1 V, a completely counterintuitive finding inlight of the apparent structural symmetry of the DNA–coralynecomplex. Non-equilibrium Green’s function (NEGF) calculationsof a coherent, tight-bonding model based on density functionaltheory (DFT) suggested that the rectification behaviour was causedmainly by the coralyne-induced local spatial asymmetry of the distri-bution of electron states along the DNA chain, which, in turn, leads toan asymmetric change in the transmission function associated withthe second highest occupied molecular orbital (HOMO−1).
Results and discussionCoralyne is a small, planar molecule that has been shown to specifi-cally intercalate in poly(dA) and strongly bind with adenine–adenine (A–A) base pair mismatches28–30. Given the π conjugationnature of coralyne, its intercalation in native DNA is expected toperturb the π–π stacking interaction between the neighbouring basesof DNA strands, thereby modulating electron transport through thetreated DNA molecule. We prepared the DNA–coralyne complex byspecifically intercalating two coralyne molecules into a custom-designed 11-base-pair (bp) DNA molecule (5′-CGCGAAACGCG-3′)containing three mismatched A–A base pairs at the centre (Fig. 1b).UV–vis spectroscopy shows that the addition of the DNA moleculesto the coralyne alters the UV–vis absorption of the latter; instead ofa single maximum at 420 nm, the DNA–coralyne complex has twolocal maxima of 412 nm and 435 nm, an indication that the coralyne
1Single Molecule Study Laboratory, College of Engineering and Nanoscale Science and Engineering Center, University of Georgia, Athens, Georgia 30602,USA. 2Department of Chemistry, Ben-Gurion University of the Negev, Beer-Sheva 84105, Israel. 3Ilse-Katz Institute for Nanoscale Science and Technology,Ben-Gurion University of the Negev, Beer-Sheva 84105, Israel. †These authors contributed equally to this work. *e-mail: [email protected]; [email protected]
ARTICLESPUBLISHED ONLINE: 4 APRIL 2016 | DOI: 10.1038/NCHEM.2480
NATURE CHEMISTRY | VOL 8 | MAY 2016 | www.nature.com/naturechemistry484
© 2016 Macmillan Publishers Limited. All rights reserved
is intercalated between the adenine bases (Fig. 1c)29. The results ofcontinuous fraction analysis (Job plot, Fig. 1d), in which aninflection point of 1 was obtained for the single-stranded (ss)DNA to coralyne concentration ratio, indicate that two coralynemolecules were inserted into one formed DNA duplex. Furtherconfirmation of this outcome was provided by structural andenergy simulations as well as a circular dichroism (CD) spectrum(Supplementary Fig. 5), which showed that the most stablecomplex structure was symmetric and composed of B-form DNAwith two coralyne molecules inserted. Temperature-dependentultraviolet and CD measurements prove that the DNA–coralynecomplex is stable at room temperature (Supplementary Fig. 5).The DNA molecules used in this study were thiolated at their3′ ends to facilitate their contact with the Au(111) surface via aAu–S bond. Before electrical measurements, the morphologies ofthe native duplex DNA and of the DNA–coralyne complex on anAu (111) surface in PBS solution were studied using STMimaging (Supplementary Fig. 1). The electronic transport propertiesof the DNA–coralyne complex were then measured and comparedwith those of the native duplex DNA using STMBJ techniques(see Methods).
Conductance behaviour of DNA–coralyne complex molecularjunctions. Static differential conductance measurements of singleDNA–coralyne complex molecular junctions were first performedunder different bias voltages using STMBJ either in thecontinuous-stretching (CS) mode to form transient junctions or
under the stretching–holding (SH) approach to form stabilizedjunctions. Both techniques have been detailed elsewhere26,27. UsingCS-STMBJ, we generated conductance histograms comprising1,000–2,000 differential conductance traces. The first and secondpeaks obtained in the histogram correspond to one and twomolecules trapped in measured junctions, respectively. Therefore,the conductance of native duplex DNA was G = 2.69 × 10−6G0
under a bias of 0.3 V (Fig. 2a) and G = 3.30 × 10−6G0 under a biasof −0.9 V (Fig. 2b), where G0 = 2e2/h = 77.4 µS (e is the electroncharge and h is Planck’s constant). Repeating the conductancemeasurements for the DNA–coralyne complex using the samemethod and the same bias voltage of 0.3 V showed that itsconductance was 3.17 × 10−6G0 (Fig. 2c), that is, very close to that ofthe duplex DNA. However, under higher bias values, theconductance measurements of native DNA diverged from those ofthe DNA–coralyne complex, especially when under negative biasconditions. For example, at a bias of −0.9 V, the conductance of theDNA–coralyne complex was 10.77 × 10−6G0 (Fig. 2d), which is threetimes that of the native DNA (3.30 × 10−6G0) (Fig. 2b). This strikingphenomenon was also observed using the SH-STMBJ technique.
To elucidate the variable conductance behaviour of the DNA–coralyne complex under different bias voltages, we carried outI–V measurements of native DNA and of the DNA–coralynecomplex by continuously sweeping the bias voltages (−1.1 to 1.1 V)across single junctions stabilized by the free-holding action ofthe SH-STMBJ technique27. The I–V evaluation of native DNArevealed symmetrical current behaviour under opposite bias
=
+
360 390 420 450 480
0.0
0.2
0.4
0.6
0.6 1.2 1.8 2.4
1.23
1.26
1.29
V
5́ -CGC GAA ACG CG-3́ -SH
DNA–coralyneCoralyne
ssDNA/coralyne
Abs
orba
nce
Wavelength (nm)
A41
2/A
435
+N
O
O
O
O
Cl
a b
c d
Figure 1 | STMBJ set-up and UV–vis spectroscopy measurements. a, Schematic of the STMBJ system. The single-molecule junction is formed in the regionindicated by the dashed box. b, Molecular junction composed of the DNA–coralyne complex (magnified view of the dashed box in a). The DNA–coralynecomplex is formed with a coralyne molecule and a sample DNA molecule (5′-CGCGAAACGCG-3′-SH). c, UV–vis absorbance spectra of coralyne alone(black) and coralyne in the presence of DNA (red). The ratio of ssDNA to coralyne is 1. After adding DNA to a coralyne solution, the single absorption peakof coralyne at 420 nm splits into two peaks at 412 and 435 nm, indicating the intercalation of coralyne into native DNA. d, Job plot of DNA with coralyne,in which an inflection point of 1 is observed, indicating that two coralyne molecules have been inserted into the formed DNA duplex. ssDNA/coralyne(horizontal axis) represents the concentration ratio of ssDNA to coralyne. The total concentration of coralyne and ssDNA remained unchanged (at 10 µM)during the entire titration. A412/A435 (vertical axis) is the absorption ratio of coralyne at 412 nm versus 435 nm. Sample conditions for UV–vis absorptionwere 10 mM phosphate, 100 mM NaCl, pH 7.4.
NATURE CHEMISTRY DOI: 10.1038/NCHEM.2480 ARTICLES
NATURE CHEMISTRY | VOL 8 | MAY 2016 | www.nature.com/naturechemistry 485
© 2016 Macmillan Publishers Limited. All rights reserved
polarities (Fig. 3a). In stark contrast, the DNA–coralyne complexexhibited decidedly asymmetrical I–V behaviour, with a sharpincrease in current under negative bias. The inset in Fig. 3a is agraph overlay of the continuously measured I–V curves and thestatic current measured under different bias voltages. Comparedwith directly measured I–V characteristics, the static current,obtained here by statistically averaging over a large number of mol-ecular junctions, may obscure some of the detail revealed by thedirect I–Vmeasurements. However, the I–V characteristics obtainedusing either of the two methods agreed well with each other, thusvalidating the measurements. The results therefore indicate thatthe DNA–coralyne complex junction functions as a molecular rec-tifier. We emphasize that the observed strong rectification behaviouris highly counterintuitive in light of the apparent symmetricalmolecular structure of the DNA–coralyne complex. Indeed, suchstrong rectification is an unprecedented feature exclusive toDNA-based molecular devices. A complementary study performedon poly d(CG)4 DNA, which eliminates the three mismatched basepairs, shows the absence of rectification (Supplementary Fig. 6),suggesting the specificity of the designed mismatch sequence forthe observed I–V rectification.
To describe rectification, two quantities are necessary: the recti-fication ratio (RR) and the switch-on voltage. Figure 3c presents RRversus bias curves following RR = |Iforward/Ireverse| (here –I(V–)/I(V+))for both the native DNA and the DNA–coralyne complex. Incontrast to the native DNA, the average RR of which was close tounity within the entire bias sweeping regime, the DNA–coralynecomplex exhibited small RR values slightly above unity under lowbias (∼0–0.7 V) that increased sharply when the bias exceeded0.7 V and then jumped to ∼15 at 1.1 V. We define the biasvoltage at which the RR value starts to deviate from unity towardhigher values as the switch-on voltage, which for this DNA–coralyne molecular rectifier is thus determined to be ∼0.7 V.Given the symmetrical nature of the molecular structure of theDNA–coralyne complex, its RR of 15—much greater than the RRvalues (∼2–10) typically reported for molecular junctions withasymmetric molecules or inconsistent molecule–electrode contacts—is encouraging24,25. This value is also close to the theoretically esti-mated upper limit of RR (∼20) that can be achieved in a coherent
transport molecular junction system31. A much higher RR canthus be expected when additional asymmetric factors, such as theasymmetric environmental control of the junction system32 orasymmetric gating33,34, are introduced into the demonstratedDNA–coralyne junction system. Based on our promising results,we believe that this DNA–coralyne junction system is an idealtestbed for pushing the limits of experimental molecular rectificationcloser to achieving the goal of functional molecular devices.
Rectification mechanism of DNA–coralyne complex junctions.To explore the mechanism of DNA–coralyne complex junctionrectification, we turned to theoretical calculations, central towhich is the assumption that the electron transport along theDNA molecular junction is fully coherent, which is reasonable inlight of the short lengths of the DNA molecules used in thepresent experiments (see Supplementary Fig. 8 for furtherdiscussion)35,36. We used a fully coherent, tight-binding model forthe DNA junction37–40, graphically depicted in Fig. 4a. Theelectronic states localized on the base units are defined as |n,s⟩,where n denotes the position of the base along the DNA chainand s denotes the strand (top or bottom). The double-strandDNA molecule is encoded into a tight-binding double ladder,described by the Hamiltonian
H =∑11
n=1
∑s=1,2
ϵn,s n, s⟩⟨n, s∣∣ ∣∣
+∑
n,s(αsn,n+1 n, s⟩⟨n + 1, s
∣∣ ∣∣ + h.c.)
+∑
n(βn n, 1⟩⟨n, 2
∣∣ ∣∣ + h.c.) (1)
where ϵn,s are the onsite energies of the (excess) charges localized onthe |n,s⟩ basis, αsn,n+1 are the intrastrand coupling matrix elementsbetween neighbouring bases, and βn are the interstrand couplingmatrix elements. These parameters for the native DNA, whichdepend on the bases and their sequence (that is, the order ofbases), were evaluated in earlier theoretical studies41–43. However,considering the possible electrostatic interaction between theelectrodes and the DNA chain, which is known to alter theenergetics of molecules in proximity to metallic surfaces,
1.00.5 1.50
3
6
0.05 0.10 0.15
0
10
20
30
0.05 0.10 0.150
3
6
9
1.00.5 1.5
0.05 0.10 0.151.00.5 1.50.05 0.10 0.151.00.5 1.50
3
6
9
0.3 V –0.9 V
–0.9 V0.3 V
Con
duct
ance
(×
10–6
G0)
Con
duct
ance
(×
10–6
G0)
Con
duct
ance
(×
10–6
G0)
Con
duct
ance
(×
10–6
G0)
Stretching (nm) Time (s) Stretching (nm) Time (s)
Stretching (nm) Time (s) Stretching (nm) Time (s)
a b
c d
Figure 2 | Conductance histograms and typical conductance traces of native DNA and DNA–coralyne complex molecular junctions. a–d, Native DNA (a,b)and DNA–coralyne complex (c,d) were measured with both CS-STMBJ and SH-STMBJ techniques under 0.3 V (a,c) and −0.9 V (b,d). A conductancehistogram and typical conductance traces obtained using both the CS-STMBJ and SH-STMBJ methods are shown to the left and right of the middle dashedline, respectively. All histograms were constructed from 1,000–2,000 traces. Note that under −0.9 V, the conductance of the DNA–coralyne complex is threetimes that of the native DNA, although they do not show a significant difference under 0.3 V.
ARTICLES NATURE CHEMISTRY DOI: 10.1038/NCHEM.2480
NATURE CHEMISTRY | VOL 8 | MAY 2016 | www.nature.com/naturechemistry486
© 2016 Macmillan Publishers Limited. All rights reserved
the onsite energies of the bases that are in contact with theelectrodes44, ϵ1,1 and ϵ11,2 , are treated as free fitting parameters.
To address the rectification behaviour, an additional term thatphenomenologically describes the voltage drop across the junction45
was added to the Hamiltonian
HV =∑11
n=1
∑sVn n, s⟩⟨n, s
∣∣ ∣∣ (2)
where Vn is the shift in the local energy. Assuming that the voltageprofile along the chain is linear and that a fraction φ of the totalvoltage drop V falls on the DNA chain, the local potential shiftcan be estimated via45
Vn = φn − 110
−12
( )V (3)
such that V1 = –(φV/2) and V11 = (φV/2), leading to a total voltagedrop of φV across the DNA chain and to a voltage-dependentHamiltonian. The value of φ is treated as a fitting parameter. Forthe electrodes, we calculated the current using the Landauer form-alism in the frame of the wide-band approximation(Supplementary equation (6))46.
Electron transport properties of native duplex DNA. The first stepin characterizing DNA junctions is to calculate the transportproperties of the native duplex DNA. Considering that the localonsite energies and coupling matrix elements are provided fromprevious calculations, there are four fitting parameters: the onsiteenergy shift S1 at the edge sites in contact with the electrodes, thelevel broadening Γ, the voltage drop fraction φ, and the electrode
chemical potential μ. We calculated the I–V curves and fit thetheoretical curve with the experimental data by minimizing (usinga dynamic metropolis Monte-Carlo minimization algorithm47) thesquared difference of Itheory(V) – Idata(V).
Figure 3b (inset) shows the excellent agreement between theexperimental (solid blue) and theoretical (dashed green) I–Vcurves and Supplementary Table 3 lists the fitting parametersobtained (with a deviation of <1% between experimental and theor-etical data). We note that the chemical potential of gold that wefound with our fitting procedure is close to its known value of∼–5.1–5.3 eV. However, we found surprising values for two of thefitting parameters. First, the level broadening was ∼1 eV, muchlarger than the values found in most molecular junctions (thatis, 10–3–10–1 eV). Second, the voltage drop along the junction wasextremely small. Because such strong molecule–electrode bindingusually results in a substantial voltage drop on the molecule, thesetwo unexpected fitting parameter values seem to contradicteach other.
To unravel this apparent contradiction, we plotted the squaredamplitude |ψn|
2 of the wavefunction of the DNA double strands(orange, top strand; green, bottom strand) as a function of its positionalong the DNA chain (Fig. 4b). Examining the wavefunctions closestin energy to the chemical potential of the electrodes (that is, the statethat contributes most to the charge transport) shows that the wave-functions are highly localized around the central area of the DNAchain. Moreover, from the centre of the chain to its edges, the wave-function weight drops by approximately eight orders of magnitude.Therefore, despite its large level broadening value, the DNA junctioneffectively behaves as a molecular level localized at the centre of the
0
–2
–4
–6–1.0 –0.5 0.0 0.5 1.0
–0.6
–0.4
–0.2
0.0
0.2
0.4
–1.0 –0.5 0.0 0.5 1.0 –1.0 –0.5 0.0 0.5 1.0
–1.0 –0.5 0.0 0.5 1.0–6
–4
–2
0
–1
0
1
0.0 0.2 0.4 0.6 0.8 1.00
3
6
9
0.0 0.5 1.00
5
10
15
–0.8 –0.4 0.0 0.4 0.80
2
4
6
–1.0 –0.5 0.0 0.5 1.0
0
50
100
150
DNA DNA–coralyne
Cur
rent
(nA
)
Cur
rent
(nA
)
Bias voltage (V) Bias voltage (V)
Bias voltage (V) Bias voltage (V)
Experiment
Experiment
Theory
Theory
dI/d
V (
×10
–5 G
0)
Rec
tific
atio
n ra
tio
a b
c d
Figure 3 | I–V characteristics of native DNA and DNA–coralyne complex. a, Average I–V curves (solid line) over 40 individual curves (light shadow) ofnative DNA (blue) and DNA–coralyne complex (red). Inset: Graph overlay of I–V curves with static current values (yellow circles, native DNA; cyan squares,DNA–coralyne complex) under different bias voltages. b, Experimental (solid red line) and theoretical (dashed orange line) I–V curves for DNA–coralynecomplex (main panel) and native DNA (inset). c, Average rectification ratios of native DNA (blue) and DNA–coralyne complex (red). Inset: Average rectificationratios of native DNA and DNA–coralyne complex over 40 individual rectifications versus bias voltage curves. d, Average differential conductance (dI/dV) ofnative DNA (blue) and DNA–coralyne complex (red). Inset: Average dI/dV of native DNA and DNA–coralyne complex over 40 individual dI/dV curves.
NATURE CHEMISTRY DOI: 10.1038/NCHEM.2480 ARTICLES
NATURE CHEMISTRY | VOL 8 | MAY 2016 | www.nature.com/naturechemistry 487
© 2016 Macmillan Publishers Limited. All rights reserved
chain and very weakly coupled to the electrodes, which explains theobserved small voltage drop. In addition, the wavefunction exhibitedonly slight asymmetry, which conferred on the native DNA itsrelatively poor rectification behaviour.
DNA–coralyne complex electron transport properties. To accountfor the electron transport properties of the DNA–coralyne complex,we postulated that intercalation of the two coralyne moleculesbetween the three A–A base pairs located at the centre of the DNAstrands would only alter the local energies of the surroundingbases. Under this scenario, we can safely assume that parametersrelated to DNA–electrode coupling (Γ, μ, S1 and φ) are not affectedand should be the same for the DNA–coralyne complex and forthe native duplex DNA. In addition, their values should matchthose found above when fitting the I–V curves of the native DNA.On the other hand, the on-site energies of the central sites, that is,close to the intercalation points of the coralyne molecules, arealtered and the numerical values obtained from the literaturecannot be used. Therefore, we define the local energy shifts (S5,1,S6,1, S7,1, S5,2, S6,2, S7,2) due to the presence of the coralyne41–43.
The coralyne-induced local energy shifts are new fitting par-ameters that should be found by minimizing the difference betweenthe theoretical and experimental I–V curves. We note, however,that we are unable to fit the experimental curve with high accuracy(<1%) without breaking all the spatial symmetries in these sites. Forexample, one could imagine that the symmetry Sn,1 = Sn,2, n = 5,6,7exists, but we found that this is not the case, a logical outcome dictatedby the asymmetry of the coralyne molecule. The fitting results showthat the theoretical I–V curve and the corresponding experimentalI–V curve agree well with each other (Fig. 3b).
To understand the origin of the rectification, we again plotted thesquared wavefunction |ψn|
2 for the DNA level closest to the electrodeFermi level (Fig. 4c). In contrast to the wavefunction of native DNA,that of the DNA–coralyne complex exhibits striking asymmetry, suchthat a difference of four orders of magnitude was found between thewavefunction weights at the left and right edges. The spatial asym-metry induced by coralyne intercalation thus explains the strongrectification by the DNA–coralyne complex.
We also measured the current through the single-molecule junc-tion by cyclically sweeping the bias between −1.2 and 1.2 V(Supplementary Fig. 10), which showed a large hysteresis undernegative bias for the DNA–coralyne complex, a finding that is incontrast to the well overlapped forward and backward currents ofnative DNA and that suggests a bias-dependent structuralchange45,48. When the voltage drops along the chain, the responseto the voltage drop, which depends on the direction of the biasdrop due to the above-mentioned asymmetry, is different, thusleading to a dependence of the transmission function, and conse-quently the current, on bias direction. This observation is in goodagreement with the experimental dI/dV result, which shows a pro-nounced asymmetry in the local density of states that contributed tothe current under opposite bias directions (Fig. 3d). It is surprisingthat such strong rectification behaviour can be achieved with such asmall value of the voltage drop fraction φ ≈ 2 × 10–4, a finding thatdemonstrates the strong sensitivity of the transport properties tolocal conditions.
We explored the role of the voltage drop further by plotting thetransmission function T(ω,V) of the DNA–coralyne complex forthree voltage values, V = –1.1, 0 and 1.1 V (Fig. 5a). The first con-clusion that can be drawn from Fig. 5a is that electron transportthrough the DNA is indeed dominated by a single orbital (HOMOlevel, that is, the level at which the energy is closest to the chemicalpotential). Second, it can also be concluded that the voltage dropalong the junction (which is small due to the small value of φ) doesnot affect the position of the resonance. Considering that the coralynemolecules intercalated between three mismatched A–A base pairs, thisresult coincides with those of previous reports showing that theHOMO of DNA molecules is dominated mainly by the HOMO ofguanine bases, which are not expected to be perturbed in our case8,12.However, the substantial change induced by coralyne intercalationis conferred by the off-resonance background transmission.
A more detailed examination of the wavefunction reveals thatthe large background transmission enhancement is related to afour-order increase in the magnitude of the coupling between theelectrodes and a level lying about 1.65 eV away from the chemicalpotential, that is, the HOMO−1 level. This outcome is seen in
α1,2s = 2
α1,2s = 1
N
N
2
2
1
1
β1
β2
2 4 6 8 10 2 4 6
nn
8 10
10–8
10–5
10–2
10–11
10–9
10–7
10–5
10–3
10–1
Γ
Γ
DNAb c
a
DNA–coralyne
|ψn|
2
|ψn|
2
Figure 4 | Theoretical calculations based on the tight-bonding model. a, Schematic representation of the tight-binding model for the double-stranded DNAmolecular junction. b,c, Weight of the wavefunction (orange, top strand; green, bottom strand) closest to the chemical potential as a function of base pairposition along the chains of the native DNA chain (b) and the DNA–coralyne complex (c). Note that, for the DNA–coralyne complex, the asymmetricelectronic structure induced by the intercalation of coralyne shifts the orbital weights asymmetrically, resulting in a highly asymmetric effective coupling ofthe orbitals to the electrodes.
ARTICLES NATURE CHEMISTRY DOI: 10.1038/NCHEM.2480
NATURE CHEMISTRY | VOL 8 | MAY 2016 | www.nature.com/naturechemistry488
© 2016 Macmillan Publishers Limited. All rights reserved
Fig. 5b, where the HOMO and HOMO−1 contributions to thetransmission function (solid and dashed lines, respectively) areplotted for bias voltages V = –1.1 and 1.1 eV (blue and greenlines). We can thus conclude that although the overall conductanceis dominated by the level closest to the electrodes’ chemical poten-tials, the rectification behaviour is in fact due to a voltage-drop-induced change in the coupling (and resulting transmission) ofthe HOMO−1 level.
The picture that emerges from these calculations is thus thefollowing. Although the native DNA exhibits some asymmetry, itis not sufficient to generate substantial rectification, especially inlight of the small voltage drop on the molecule. When coralyne isintercalated between the DNA strands, the asymmetric structureof the coralyne molecule induces additional asymmetry onto theDNA orbitals. This asymmetry is manifested mainly in a shift inthe orbital structure of the wavefunctions but not in the orbitalenergies (see Supplementary Fig. 7 for further discussion). The moststriking consequence of this change is an increase in the effectivecoupling between the HOMO−1 orbital and the electrodes (which isin itself asymmetric, being very small for positive bias and large fornegative bias). Thus, the HOMO−1 conduction channel, which isoff-resonant and governed by the level broadening for all voltagesstudied, is the one mostly affected by the coralyne intercalation.
To the best of our knowledge, our finding that coralyne inter-calation only manipulates the HOMO−1 transmission function byincreasing its coupling strength without perturbing the frontierorbital of HOMO is a feature unique to DNA–coralyne complexmolecular junctions. We note that this rectification mechanism isessentially different from any models proposed in the past toaccount for the rectification behaviour of molecular junctionsystems that possess asymmetric molecular cores dominated by asingle orbital asymmetrically coupled to the electrodes19,49,50. Wepoint out, however, the resemblance between the mechanism wesuggest here and a similar mechanism in molecular junctionsresponsible for negative differential conductance, where a biasvoltage drop along an asymmetric molecule changes the effectivecoupling between the molecular orbitals and the electrodes45.
In summary, we have constructed a molecular rectifier based onintercalating specific, small molecules into designed DNA strands.Characterization by diverse STMBJ techniques of the electronic trans-port properties of molecular junctions consisting of native DNA or ofa DNA–coralyne complex showed drastic differences. The I–V curvesof the single DNA–coralyne complex junction device exhibited astrong rectification ratio of around 15 at 1.1 V. Using the NEGFmethod, calculations based on a coherent tight-binding modelrevealed that the rectification is predominantly caused by the
coralyne-induced spatial asymmetry, which leads to voltage-drop-induced changes in the coupling and transmission functions of themolecular HOMO−1 level. Not only do these results offer a newmethod for studying the DNA–molecule interaction, they alsosuggest a novel strategy for engineering molecular electronic elementsbased on a specifically designed functional DNA complex.
MethodsDNA–coralyne complex sample preparation. The DNA–coralyne complex wasprepared by mixing equivalent amounts of DNA (5′-CGCGAAACGCG-3′-SH) andcoralyne in PBS. The DNA–coralyne mixture was heated to 80 °C in a water bath andallowed to return slowly to room temperature for annealing and formation of theDNA–coralyne complex. The final concentration of DNA–coralyne complex used insubsequent electrical measurements was 3 µM. As a control, the complementaryduplex DNA was prepared in the same concentration using identical methods. Oncethe DNA–coralyne complex was acquired, it was applied to a freshly annealed Au(111)surface to form a self-assembled monolayer. Immediately after 40 min incubation, thesamples were measured. The Au substrate was made by thermally evaporating anAu layer (thickness of ∼100 nm) on a freshly cleaved mica surface. Before using theAu substrate, it was annealed by hydrogen flame to form a Au(111) surface.
Molecular junction formation and electrical measurements. Electricalmeasurements were performed in PBS buffer on a PicoPlus SPM system (MolecularImaging) with a PicoScan 3000 controller (Molecular Imaging). Sheared gold wirewith a diameter of 0.25 mm (99.999%, Alfa Aesca) coated with wax was used as anSTM tip. The measurements were carried out with both CS-SPMBJ and SH-SPMBJ.For CS-STMBJ, the STM tip was first driven by a piezoelectric transducer (PZT)close to the DNA–coralyne complex monolayer on the Au surface until the currentreached a preset value, which indicated that the Au–DNA–Au junctions had formed.Subsequently, a current–distance trace was recorded while simultaneously retractingthe STM tip. Typically, 1,000–2,000 traces were collected for a given experimentalcondition at a retrace rate of 24–30 nm s−1. For SH-STMBJ, the retraction wasdivided into two processes: abrupt stretching and free holding of the molecularjunctions by withdrawing the STM tip by ∼0.6 nm and then holding it stably inposition for 25 ms. The stretching–holding processes, which were controlled by ahomemade LabView computer program, were alternately implemented until themolecular junctions were completely broken. The current–distance traces weremeasured at different biases using these two different techniques to study theelectrical properties of the DNA–coralyne complex. In the present system, theexternal bias was applied on the substrate with the tip grounded.
Received 15 July 2015; accepted 22 February 2016;published online 4 April 2016
References1. Ratner, M. A brief history of molecular electronics. Nature Nanotech. 8,
378–381 (2013).2. Tao, N. J. Electron transport in molecular junctions. Nature Nanotech. 1,
173–181 (2006).3. Aradhya, S. V. & Venkataraman, L. Single-molecule junctions beyond electronic
transport. Nature Nanotech. 8, 399–410 (2013).4. Di Ventra, M. & Pershin, Y. V. Spin physics: DNA spintronics sees the light.
Nature Nanotech. 6, 198–199 (2011).
–2 –1 0 1 2–2
10–6
10–3
1
–1 0 1 2
10–7
10–5
10–3
10–1
Bias voltagea b
–1.1 eV
–1.1 eV
1.1 eV1.1 eV
0 eVT
(ω)
T(ω
)
ω–μ ω–μ
HOMOHOMO–1
HOMOHOMO–1
Figure 5 | Transmission function of the DNA–coralyne complex. a, Transmission function T(ω) as a function of ω for different values of bias voltageV= –1.1, 0 and 1.1 eV. The position and height of the transmission resonance are unaffected by the bias, indicating that the origin of rectification is a change in theoff-resonance background transmission. b, HOMO (solid lines) and HOMO−1 (dashed lines) contributions to the transmission function T(ω) as a function of ωfor different values of bias voltage V= –1.1 and 1.1 eV. The HOMO contribution, corresponding to the transmission resonance, is unaffected by the voltage drop, whilethat of HOMO−1 increases by three orders of magnitude. This demonstrates that, although for low bias the conductance and current are typically determined by theresonant transmission channel, the rectification behaviour of the DNA–coralyne complex at large biases is determined by the off-resonance transmission channel.
NATURE CHEMISTRY DOI: 10.1038/NCHEM.2480 ARTICLES
NATURE CHEMISTRY | VOL 8 | MAY 2016 | www.nature.com/naturechemistry 489
© 2016 Macmillan Publishers Limited. All rights reserved
5. Genereux, J. C. & Barton, J. K. Molecular electronics: DNA charges ahead.Nature Chem. 1, 106–107 (2009).
6. Livshits, G. I. et al. Long-range charge transport in single G-quadruplex DNAmolecules. Nature Nanotech. 9, 1040–1046 (2014).
7. Dekker, C. & Ratner, M. A. Electronic properties of DNA. Phys. World 14,29–33 (2001).
8. Mallajosyula, S. S. & Pati, S. K. Toward DNA conductivity: a theoreticalperspective. J. Phys. Chem. Lett. 1, 1881–1894 (2010).
9. Giese, B. et al. Direct observation of hole transfer through DNA by hoppingbetween adenine bases and by tunnelling. Nature 412, 318–320 (2001).
10. Lewis, F. D. et al. Direct measurement of hole transport dynamics in DNA.Nature 406, 51–53 (2000).
11. Xiang, L. et al. Intermediate tunnelling–hopping regime in DNA chargetransport. Nature Chem. 7, 221–226 (2015).
12. Wang, K. et al. Structure determined charge transport in single DNA moleculebreak junctions. Chem. Sci. 5, 3425–3431 (2014).
13. Joshua, H. et al. Effects of cytosine methylation on DNA charge transport.J. Phys. Condens. Matter 24, 164204 (2012).
14. Tsutsui, M. et al. Electrical detection of single methylcytosines in a DNAoligomer. J. Am. Chem. Soc. 133, 9124–9128 (2011).
15. Liu, H. et al. Electronic enhancement effect of copper modification of base pairson the conductivity of DNA. J. Phys. Chem. C 115, 22547–22556 (2011).
16. Liu, S. et al. Direct conductance measurement of individual metallo-DNAduplexes within single-molecule break junctions. Angew. Chem. Int. Ed. 50,8886–8890 (2011).
17. Aviram, A. & Ratner, M. A. Molecular rectifiers. Chem. Phys. Lett. 29,277–283 (1974).
18. Ashwell, G. J., Tyrrell, W. D. & Whittam, A. J. Molecular rectification: self-assembled monolayers in which donor–(π-bridge)–acceptor moieties arecentrally located and symmetrically coupled to both gold electrodes. J. Am.Chem. Soc. 126, 7102–7110 (2004).
19. Nijhuis, C. A., Reus, W. F. & Whitesides, G. M. Mechanism of rectification intunneling junctions based on molecules with asymmetric potential drops. J. Am.Chem. Soc. 132, 18386–18401 (2010).
20. Díez-Pérez, I. et al. Rectification and stability of a single molecular diode withcontrolled orientation. Nature Chem. 1, 635–641 (2009).
21. Elbing, M. et al. A single-molecule diode. Proc. Natl Acad. Sci. USA 102,8815–8820 (2005).
22. Batra, A. et al. Tuning rectification in single-molecular diodes. Nano Lett. 13,6233–6237 (2013).
23. Yoon, H. J. et al. Rectification in tunneling junctions: 2,2′-bipyridyl-terminatedn-alkanethiolates. J. Am. Chem. Soc. 136, 17155–17162 (2014).
24. Yee, S. K. et al. Inverse rectification in donor–acceptor molecularheterojunctions. ACS Nano 5, 9256–9263 (2011).
25. Wang, K. et al. Measurement and understanding of single-molecule breakjunction rectification caused by asymmetric contacts. J. Chem. Phys. 141,054712 (2014).
26. Xu, B. & Tao, N. Measurement of single-molecule resistance by repeatedformation of molecular junctions. Science 301, 1221–1223 (2003).
27. Zhou, J., Chen, F. & Xu, B. Fabrication and electronic characterization of singlemolecular junction devices: a comprehensive approach. J. Am. Chem. Soc. 131,10439–10446 (2009).
28. Persil, Ö. et al. Assembly of an antiparallel homo-adenine DNA duplex by small-molecule binding. J. Am. Chem. Soc. 126, 8644–8645 (2004).
29. Polak, M. & Hud, N. V. Complete disproportionation of duplex poly(dT)·poly(dA) into triplex poly(dT)·poly(dA)·poly(dT) and poly(dA) by coralyne. NucleicAcids Res. 30, 983–992 (2002).
30. Ren, J. & Chaires, J. B. Sequence and structural selectivity of nucleic acid bindingligands. Biochemistry 38, 16067–16075 (1999).
31. Stadler, R., Geskin, V. & Cornil, J. A theoretical view of unimolecularrectification. J. Phys. Condens. Matter 20, 374105 (2008).
32. Capozzi, B. et al. Single-molecule diodes with high rectification ratios throughenvironmental control. Nature Nanotech. 10, 522–527 (2015).
33. Huang, C. et al. Break junction under electrochemical gating: testbed for single-molecule electronics. Chem. Soc. Rev. 44, 889–901 (2015).
34. Capozzi, B. et al. Tunable charge transport in single-molecule junctions viaelectrolytic gating. Nano Lett. 14, 1400–1404 (2014).
35. Genereux, J. C., Wuerth, S. M. & Barton, J. K. Single-step charge transportthrough DNA over long distances. J. Am. Chem. Soc. 133, 3863–3868 (2011).
36. Tan, B. et al. Charge transport in DNA nanowires connected to carbonnanotubes. Phys. Rev. B 92, 075429 (2015).
37. Brisker-Klaiman, D. & Peskin, U. Coherent elastic transport contribution tocurrents through ordered DNA molecular junctions. J. Phys. Chem. C 114,19077–19082 (2010).
38. Cuniberti, G. et al. in Charge Migration in DNA (ed. Chakraborty, T.) 1–20(Springer, 2007).
39. Yi, J. Conduction of DNA molecules: a charge-ladder model. Phys. Rev. B 68,193103 (2003).
40. Grib, N. et al. Distance-dependent coherent charge transport in DNA: crossoverfrom tunneling to free propagation. J. Biophys. Chem. 1, 77–85 (2010).
41. Jortner, J. et al. Superexchange mediated charge hopping in DNA. J. Phys. Chem.A 106, 7599–7606 (2002).
42. Voityuk, A. A. et al. Energetics of hole transfer in DNA. Chem. Phys. Lett. 324,430–434 (2000).
43. Voityuk, A. A. et al. Electronic coupling for charge transfer and transport inDNA. J. Phys. Chem. B 104, 9740–9745 (2000).
44. Perrin, M. L. et al. Large tunable image–charge effects in single-moleculejunctions. Nature Nanotech. 8, 282–287 (2013).
45. Xu, B. & Dubi, Y. Negative differential conductance in molecular junctions: anoverview of experiment and theory. J. Phys. Condens. Matter 27, 263202 (2015).
46. Verzijl, C. J. O., Seldenthuis, J. S. & Thijssen, J. M. Applicability of the wide-bandlimit in DFT-based molecular transport calculations. J. Chem. Phys. 138,094102 (2013).
47. Vanderbilt, D. & Louie, S. G. A Monte-Carlo simulated annealing approach tooptimization over continuous variables. J. Comput. Phys. 56, 259–271 (1984).
48. Nowak, A. M. & McCreery, R. L. In situ Raman spectroscopy of bias-inducedstructural changes in nitroazobenzene molecular electronic junctions. J. Am.Chem. Soc. 126, 16621–16631 (2004).
49. Kornilovitch, P. E., Bratkovsky, A. M. & Stanley Williams, R. Currentrectification by molecules with asymmetric tunneling barriers. Phys. Rev. B 66,165436 (2002).
50. Liu, R. et al. Organometallic molecular rectification. J. Chem. Phys. 124,024718 (2006).
AcknowledgementsThe authors acknowledge the US National Science Foundation for funding this work(ECCS 0823849 and ECCS 1231967).
Author contributionsB.X. conceived the experiment. C.G., K.W., J.H. and B.W. performed the experiment andanalysed the data. Y.D. supervised the theoretical calculation. E.Z.-H. and Y.D. carried outthe calculations. C.G., K.W., Y.D. and B.X. co-wrote the paper.
Additional informationSupplementary information is available in the online version of the paper. Reprints andpermissions information is available online at www.nature.com/reprints. Correspondence andrequests for materials should be addressed to Y.D. and B.X.
Competing financial interestsThe authors declare no competing financial interests.
ARTICLES NATURE CHEMISTRY DOI: 10.1038/NCHEM.2480
NATURE CHEMISTRY | VOL 8 | MAY 2016 | www.nature.com/naturechemistry490
© 2016 Macmillan Publishers Limited. All rights reserved
