Modeling and Testing of Molecular Wire Sensors To Detect a Nucleic

Modeling and Testing of Molecular Wire Sensors To Detect a Nucleic Acid Base ... A first-principles study of aryloxyanthraquinone-based optical molecu...
0 downloads 0 Views 713KB Size
J. Phys. Chem. C 2007, 111, 3495-3504

3495

Modeling and Testing of Molecular Wire Sensors To Detect a Nucleic Acid Base Bidisa Das,*,† Shuji Abe,*,† Yasuhisa Naitoh,† Masayo Horikawa,† Tetsuo Yatabe,† Yasuzo Suzuki,† Takashi Funaki,† Seiji Tsuzuki,‡ and Yuji Kawanishi† Nanotechnology Research Institute, National Institute of AdVanced Industrial Science and Technology (AIST), Tsukuba, Ibaraki 305-8568, Japan, and Research Institute for Computational Sciences, National Institute of AdVanced Industrial Science and Technology (AIST), Tsukuba, Ibaraki 305-8568, Japan ReceiVed: September 13, 2006; In Final Form: NoVember 12, 2006

We report a theoretical study of molecular wires which are designed to detect a single nucleic acid base thymine, through electrical conductance change. Three model sensors with ethynylphenyl backbones having suitable detection units and capable of forming hydrogen bonds with thymine are proposed. A combination of density functional theory and nonequilibrium Green’s function formalism were used for the theoretical studies. The results show that subtle changes in electronic properties of the sensors due to hydrogen bond formation with the target may give rise to measurable changes in conductance. We also report experimental conductance studies for one model sensor, which agrees qualitatively with the theoretical results.

1. Introduction Sensitive and quick detection of DNA sequence is a very important area of research in biological and medical sciences. Significant effort has been devoted to develop a detection scheme with single base-pair precision. A possible approach is threading one strand of DNA through a nanoscale pore and reading the identity of each base as it passes a detector.1 In this manner, it may be possible to detect the base sequence of a single DNA, if one can design a nanoscale device that detects the target nucleic base by a molecular recognition process, which would then be translated into detectable signal. By using this idea, sensing of DNA by detecting one of its nucleic bases selectively, in the presence of appropriate receptors, can be designed by using the properties of hydrogen bond formation. A change in electronic conduction when bonded to the target forms the basis for this sensor, which then detects the target electrically. We focus on a few model sensors in this study, which have been designed in our group recently.2,3 It has been theoretically demonstrated earlier4 that molecules consisting of a conducting and detecting unit joined by a suitable connection can be used to detect a single nucleic base, like thymine. There are various structural schemes for this purpose, and 2,6-diaminopyridine derivatives were studied in particular.4 Initial effects of hydrogen bonding of thymine to the sensor were evaluated by using the change in atomic charge distribution and shifts in molecular orbital energies. The calculated energy shifts were about 0.1 eV in preferable cases. The effect of a local potential of about 0.1 eV on the conductance of a simple molecular wire was evaluated5 by using a tight-binding model. The change in conductance was estimated to be about 10-20%. Although this is not a negligible effect in principle, its realization in actual molecular junction devices is a challenging goal. In this study we mainly focus on detectability by hydrogen bond formation. * Corresponding author. E-mail: [email protected] (B.D.) and [email protected] (S.A.). † Nanotechnology Research Institute. ‡ Research Institute for Computational Sciences.

A natural extension of this study is selectivity which we intend to study as the next step. In the present study we consider molecules having ethynylphenyl-based backbones which constitute the conducting molecular wire portion. Appropriate detection units, which can form hydrogen bonds with thymine, are incorporated into these molecular wires. This makes the sensor molecules suitable for conductance measurements. We also perform theoretical conductance studies to predict their sensing abilities. The changes in electronic properties of the sensor upon complex formation triggers a conductance change in the system. Although hydrogen bonding interactions are rather weak to induce drastic changes in the electronic properties, the idea is unique and one can think of many other possibilities based on it. Measurement of conductance of a single molecule is generally done in a two-probe setup. Under this condition a metalmolecule-metal junction is formed and the change in the conductance of the system due to complex formation may be estimated from the transmission spectra and current vs voltage (I-V) characteristics. The ab initio studies of the free and complexed sensors present a qualitative understanding of how these molecules interact with thymine and what changes to expect in the molecule after complex formation. There have been many studies for sensitive detection of DNA molecules. DNA sensors based on field effect transistors have been used for biomolecule detection.6-9 This idea was extended for detection of DNA where a specific detection of a point mutation is achieved by combining the electronic measurement with an allele-specific polymerase chain reaction. DNA recognition through hybridization reactions is well-known, but appropriate and efficient transducers are needed to generate a physically measurable signal from the hybridization event.10-20 This may been done optically, for example, using molecular beacons, DNA-derivatized nanoparticles, or conjugated polymers. Electrochemical methods such as redox-active nucleic acids, redox polymers, and enzymatic systems have also been used for this purpose. Many of them rely on some form of chemical amplification by polymerase chain reaction. DNA sensing based on detection of single nucleic bases has also been

10.1021/jp0659709 CCC: $37.00 © 2007 American Chemical Society Published on Web 02/07/2007

3496 J. Phys. Chem. C, Vol. 111, No. 8, 2007

Das et al.

studied before. In one such study Ru(II) complexes of tetraazaphenanthrene ligand were used as luminescent photoprobes of DNA. Luminescence was turned on in the presence of DNA. The system also exhibited selectivity in luminescence: longlived excited species were formed with adenine-thymine rich sites in comparison to short-lived ones formed with guaninecytosine sites.21 Our study approaches the problem of DNA detection in a similar manner, i.e., detection of thymine base. But the conductance change due to sensing has not been used before. 2. Theoretical Methods The main aspect of this study is to understand how the electronic properties of the sensors change in the presence of the target molecule and then to compare between the bound and unbound situations. Ab initio electronic structure methods are employed to study the properties of the free and complexed sensors. Unconstrained structures of the free sensor and the corresponding thymine complexes were optimized by using Density Functional Theory (DFT) as implemented in Gaussian 03 software22 with the 6-311G** basis-set for all atoms. To mimic the two-probe situation in experimental conductance measurement systems, we have in some cases terminated the sensors by thiol-gold (-S-Au) bonds at the two ends of the model molecules. In these studies, we have used the Lanl2DZ basis-set and effective core potential23 for Au. The structures of the gold-terminated sensors are all spin singlet. For all studies the hybrid functional B3LYP formulated by Lee, Yang, and Parr24,25 was used. The magnitude of the changes in energy eigenvalues and the changes in charge distributions of the sensor after complexation with thymine are examined to see if they serve as intuitive parameters which control the change in conductivity. As a precautionary measure, we have also studied the geometries of the stable structures using Hartree-Fock (HF) methods, using the same basis-set. We found the geometries are only marginally different from that predicted by DFT. The molecular orbital energies, however, are very different for HF and DFT methods. To evaluate potential conductance channels, the nature of the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) of the molecule should be analyzed. The energy gap between the two orbitals (HOMO-LUMO gap or HLG) is an important factor controlling the tunneling current through the molecule. If it is large, the molecule shows lesser admittance toward an incoming electron as the rearrangement of the electron density becomes more and more difficult in the presence of the extra electron. The alignment of the HLG with the electrode Fermi energy is also another important factor. It is well-known that HF methods grossly overestimate the HLG of extended systems, whereas DFT underestimates it, but hybrid methods like B3LYP, B3P86, and B3PW91 give a better estimate;26 we have used B3LYP method for our studies. All the stable geometries are characterized by vibrational frequency calculations. We also have performed a detailed analysis of charge distribution following the scheme of electrostatic potential fitting methods by Merz-Singh-Kollman27,28 with increased point densities for better accuracies.29 Binding energies of the complex formation reactions with thymine are reported in all cases, using the B3LYP/6-311G** method after basis-set superposition error correction.30,31 We calculated the change in conductance of the system after complex formation of the sensors with the target. The electronic transport properties are calculated with ATK 2.0 software,32,33 which combines nonequilibrium Green’s function (NEGF)

Figure 1. Schematic structures of the model sensor molecules S1 (top), S2 (middle), and S3 (bottom), designed to form a hydrogen-bonded complex with thymine.

formalism and DFT. The DFT implementation uses numerical atomic basis-set to solve the Kohn-Sham equations.34-36 We have used two methods for conductance studies, mainly due to the large size of the molecules. For sensor S3 and S1 we have used SZP/LDA-PZ.36,37 For the comparatively smaller molecule S2 we used DZP basis-set with GGA-PBE38 for the exchange correlation functional. In NEGF theory, the current I at an applied bias voltage V through the contact is calculated by using the Landauer formula,39,40

∫µµ T(E,V) dE

I(V) ) G0

R

(1)

L

where G0 ) 2e2/h is the quantum unit of conductance, h is Planck’s constant, µL/R are the electrochemical potentials of the left and right electrodes (µL - µR ) V), and T(E,V) is the transmission spectrum. In the zero bias limit (V ) 0), µL and µR become identical with the Fermi energy EF of the metal, and the conductance is given by G ) G0T(EF,0). The left and the right electrodes in our study are modeled by two Au(111) surfaces with a unit cell of 5 × 3 atoms to accommodate the bridging molecule. Calculations of the electrodes are done under periodic boundary conditions so that they mimic bulk metal structures. The electrode unit cell contains three Au layers along the transport direction and two Au layers at both left and right sides are included in the contact region. A sensor molecule is attached to the two electrodes with -S-Au (thiol) bonds. The structure of the central molecule is pre-optimized by using the B3LYP/6-311G** methods in G03. The structure of the Au(111)-sensor-Au(111) system has not been relaxed due to the huge computational times required. Since the detection unit that makes hydrogen bonds with thymine is sufficiently far from the metal contact region, we do not expect large changes in geometry after optimization. With these structures, we calculated the transmission spectra and I-V characteristics for the free sensor and sensor-thymine complex. 3. Proposed Sensors for Thymine Detection The three model molecules which may be used for detection of thymine nucleic base on the basis of hydrogen bond formation are shown in Figure 1 schematically. Among the three structures,

Theoretical Study of Molecular Wires

J. Phys. Chem. C, Vol. 111, No. 8, 2007 3497

Figure 3. Optimized structure of S2 with important molecular orbitals given. Figure 2. Optimized structure of sensor S1 with important molecular orbitals.

S3 can form three hydrogen bonds with thymine whereas S1 and S2 can form only two hydrogen bonds. All three molecules have a phenylene-ethynylene-based molecular wire portion,41 which we denote as the conduction unit because this portion of the molecule is mainly responsible for electronic conduction. The portion that binds to the target molecule (in this case, thymine) is denoted as the detection unit. In these model sensors the detection units are different, 2-aminopyrimidine in S1, 2-aminopyridine in S2, and 2,6-diaminopyridine in S3.4 In S2 the detection unit is placed within the conduction unit whereas in S1 and S3 the detection units are connected to the conduction unit by -NH- and -CONH- bonds, respectively. It is important to mention here that -CONH- bonds can exist in cis and trans conformations of which trans forms are more stable, so in the case of S2 and S3 we consider only trans structures. Our previous study had shown that the rotational barrier for a trans amide bond to be converted to a cis amide bond requires an energy of about 15 kcal/mol,42 hence rapid conversions at room temperatures can be ruled out. The length of the conduction unit can be longer with more phenyleneethynylene units on both sides and in practice sensor molecules with appropriate lengths are synthesized to match the gap size of the electrodes. However, the molecular wire should not be too long, because it is known that the conductance decreases exponentially with its length.43 Using the model sensors, we try to understand theoretically how the various detection units and the linkages of the conduction and detection units change the electronic properties of the molecules as a whole. Sensor S1 can exist in only one stable conformation and Figure 2 shows its optimized structure with bond distances. It is an absolutely planar molecule and the average CdC bond distances in the phenyl rings of the wire portion ranges from 1.39 to 1.41 Å with average CtC distances being 1.21 Å. The HOMO (-5.47 eV) and LUMO (-1.94 eV) of the free sensor shown in Figure 2 are delocalized over the entire wire portion. Few other orbitals near HOMO and LUMO are found to have delocalized orbitals as well (not shown). Previous studies44-46 have shown that delocalized orbitals that span the entire molecule and are positioned close to the Fermi level of the metal electrodes facilitate charge transport across molecular junctions. HLG for free S1 is 3.53 eV. In the case of S2 the detection unit is placed within the conduction unit with a flexible -CONH- bond and we discuss only the trans conformer. Optimized structure and molecular

orbitals of S2 are shown in Figure 3. It is not fully planar due to the presence of the -CONH- unit in the central portion: one-half of the molecule is twisted (by 23°) from the other half. The molecule is also not perfectly straight but is slightly bent in a “V” shape as can be seen in the figure. There can be another less stable trans amide form of S2 with opposite orientations of pyridine ring -N and amide group -N-H which cannot form two hydrogen bonds with thymine and we did not study that. HOMO, HOMO-1, LUMO, and LUMO+1 orbitals of the free sensor are shown in Figure 3. HOMO (-5.89 eV) and LUMO (-2.22 eV) in this case are extended but, due to the small twisting of the wire in the central portion of the molecule, the conjugation has been slightly disrupted. The HLG in this case is 3.67 eV. In Figure 4 the optimized structure of S3 is shown. Due to the presence of two flexible amide (-CONH-) units in S3 there can be many other conformations, depending on cis and trans conformations of the two -CONH- units, but we consider only the more stable trans structures. Relative energies of different conformations show structure A to be most stable with two trans -CONH- bonds. Conformer B (with two trans -CONHbonds, but a different orientation of the detection and conduction unit) is less stable with an energy 2.45 kcal/mol higher than that of A. The molecular wire portion of S3, A is nearly planar with average CsC bond distances in the phenyl rings ranging from 1.39 to 1.41 Å and average CtC distances of 1.21 Å, but the three phenyl rings are not placed linearly. The end ring, nearer to the detection unit, deviates from linearity slightly (-C-CtC- is 176°). The detection unit connected by the -CONH- bond is twisted with a dihedral angle of 20° with respect to the conduction unit of the molecule. The extended HOMO (-5.96 eV) and LUMO (-2.38 eV) of A are shown in Figure 4. HLG for this structure is 3.58 eV. The same figure also shows the optimized structure of the less stable conformer, B. In this case the three phenyl rings in the wire portion are not exactly coplanar, with one ring tilted out of plane by 10°. However, the molecular orbitals are spread over the whole conduction unit with HOMO at -5.84 eV, LUMO at -2.24 eV, and HLG of 3.60 eV. For the measurement of conductance, the sensor molecules are placed between gold electrodes connected by thiol-gold bonds. To understand the effect of thiol bonding to the gold electrodes, we have performed ab initio calculations for the molecules where the end hydrogen atoms of the conduction units are replaced by -S-Au groups. Such studies are useful in

3498 J. Phys. Chem. C, Vol. 111, No. 8, 2007

Figure 4. Optimized structure and molecular orbitals of S3, conformer A (top), and conformer B (bottom).

understanding the nature of the bonding of the molecule with the metallic electrodes and have been used before.39,45,46 Though attaching two isolated gold atoms at the two ends of the molecule hardly depicts the real situation, it can still give an idea about how the molecular orbitals may shift due to the presence of the metal electrodes. After -S-Au groups are bonded at the two ends of the free sensor the orbital energies indeed change and HLG becomes much smaller (2.06 eV for S1, 2.4 eV for S2, and 2.3 eV for S3). The molecular orbitals of the gold-terminated molecule also confirm that there is a reasonable amount of charge density in the contact region, which suggests that the contacts are well defined with strong covalent bonds. 4. Electronic Structure Calculations for Sensor-Thymine Complexes It is necessary to understand how the complex formation of S1, S2, and S3 affects their electronic properties for the evaluation of sensing efficiencies. Depending on the orientation of the methyl group in the target molecule (thymine), there can be two possible complex structures. In all situations we consider, both complexes behave in a similar fashion hence here we report results for only one complex in most cases. In thymine complex S1 the nitrogen atom of the pyrimidine ring and the hydrogen atom in the -NH- group are used for hydrogen bond formation. The optimized structure of the complex (with important bond lengths) is shown in Figure 5. The bond distances in the wire portion are found to remain almost unchanged (in comparison to Figure 2). In the same figure extended HOMO (-5.49 eV) and LUMO (-1.83 eV) orbitals of the sensor-thymine complex

Das et al.

Figure 5. Optimized structure of the sensor-thymine complex for S1. HOMO and LUMO of the hydrogen-bonded complex is shown (middle). The change in atomic charges after complex formation (bottom). Only differences greater than 0.01 are shown.

are shown (HLG: 3.57 eV). The binding energy with thymine due to hydrogen bonding is 6.23 kcal/mol. Free sensor S1 is fully planar (see Figure 2), but after complex formation the detection unit bends away from the conduction unit to facilitate hydrogen bond formation with thymine with a twist around the -NH- connection (of 35°). The decrease in planarity of the molecule after complex formation results in decreased delocalization of the extended π orbitals in the conduction unit. We find that there is only a small change in the energy of the HOMO (+0.06 eV), but a +0.11 eV shift in the energy of LUMO after complex formation. HLG after complex formation is 3.57 eV, which is not much different from HLG for the free complex (3.54 eV). The energy shifts of HOMO and LUMO are similar to those estimated in our previous study.4 The effects are moderate, as they are due to hydrogen bond formation, because the structure after complex formation is not very different. The changes in the charge distribution of S1 due to complex formation with thymine are displayed in the same figure. The maximum charge difference occurs near the pyrimidine ring after complex formation, but careful analysis shows that in the wire portion of the molecule there are some effects. The C-atom connected to the detection unit has a large change in atomic charge (0.116 au) with few adjacent atoms having noticeable changes. So the electronic changes due to complex formation are efficiently translated through the -NH- connection. There is an additional effect due to the presence of the thymine molecule near one benzene ring of the wire. This can be thought of as through-space interaction, which affects the atomic charge. For S2 the nitrogen atom of the pyridine ring and the hydrogen atom in the amide (-CONH-) group are used for

Theoretical Study of Molecular Wires

J. Phys. Chem. C, Vol. 111, No. 8, 2007 3499 3.66 eV in the complex from 3.58 eV in the free sensor. We expect the delocalization to decrease slightly due to the small twisting of the phenyl rings in the conduction unit. Comparatively, for conformer B geometrical changes are less significant after complexation and are only reflected in a slight change in bond distances. For this structure, the shift in HOMO and LUMO energies after complexation is +0.16 and +0.17 eV, respectively (hence HLG remains almost unchanged, 3.61 eV). The energy for thymine binding with conformer B is 9.1 kcal/ mol. Since in the case of S3 two conformers are possible (A and B), it is important that the changes caused by complex formation for these two cases do not nullify one another. Here the shifts in the molecular orbitals after complex formation with thymine are positive in both cases hence they reinforce each other. The changes in the charge distribution are most prominent in the detection unit, but they is also present in some part of the conductional wire portion. For A there is some change in charge distribution in one of the benzene rings in the molecular wire portion partially due to a through-space interaction with thymine and geometrical changes due to target binding. If we consider the changes in atomic charges when the central phenyl ring is directly connected to detection unit, it is found that conformer A has a more significant change in comparison to conformer B. 5. Conductance Calculations for Sensor Effects

Figure 6. Optimized structure of the sensor-thymine complex for S2. HOMO and LUMO of the hydrogen-bonded complex is shown (middle). The change in atomic charges after complex formation (bottom). Only differences greater than 0.01 are shown.

hydrogen bond formation with thymine. Figure 6 shows the HOMO (-5.81 eV) and LUMO (-2.05 eV) orbitals of one complex and they are all extended in nature with HLG of 3.76 eV. The binding energy with thymine due to hydrogen bonding is calculated to be 6.1 kcal/mol. After target binding the geometry of the molecule changes with slight flattening of the overall V structure and a decrease in the twist between the portions left and right of the amide bond (in comparison to Figure 3). The change in molecular orbital energies is similar to that for sensor S1. There is only a change of +0.08 eV in HOMO but there is a +0.17 eV shift in the LUMO level after complex formation. HLG after complex formation is 3.76 eV in comparison to 3.67 eV for the free complex. The central portion of the molecule shows the maximum charge difference after complex formation with the adjacent phenyl ring showing only a small charge difference. Thymine complexes of sensor S3 are formed by hydrogen bond formation between the pyridine N atom and the H atoms of the two adjacent amide groups. Both conformers A and B can form complexes with thymine and the molecular orbitals remain extended after complex formation. One of the thymine complexes of the more stable structure A is shown in Figure 7 along with molecular orbitals, important bond distances, and the change in atomic charges. The binding energy with thymine due to hydrogen bonding is calculated to be 10.8 kcal/mol for conformer A. After target binding, a phenyl ring in the conduction unit is somewhat twisted (in comparison to Figure 4). The twisting occurs near the target molecule and the angle of twist is about 15°. The three phenyl rings in the wire portion continue to be slightly nonlinear and the small changes in bond distances are reported in Figure 7. For conformation A, the shift in HOMO and LUMO energies after complexation is +0.2 and +0.29 eV, respectively. The HLG is increased marginally to

To understand the changes in conductance after complex formation of thymine, we calculated the transmission spectra for all three sensors in free and complexed conditions. Among the three types of sensors, the most conclusive experimental data were obtained for sensor S1, so we study sensor S1 in more detail and compare the results with the experimental data available. For S2 and S3 we report only the transmission at zero bias voltage and the conductance changes before and after complex formation with thymine. An example of the two-probe system for the free sensor and the sensor-thymine complex is shown in Figure 8. We have positioned the sensor (or the sensor-thymine complex) in between two electrodes in such a way that the sulfur atoms in both the termini are bonded to gold atoms. The Au-S bond has been very well studied for self-assembled monolayers and we have used bond lengths similar to such studies.47-49 We have considered situations for different adsorption sites on Au(111) as it has been reported50-52 that the molecule metal contacts are very important aspects of a two-probe system like this. In all cases we assume that the molecular wire is perpendicular to the metal surfaces, although we have also tried a slightly tilted orientation of the central molecule with respect to the metallic electrodes in one case without any significant effects. 5.1. Sensor S1. According to the literature47-49,53,54 available there can be on-top, fcc-hollow, hcp-hollow, bridge, fcc-bridge, and hcp-bridge sites for the sulfur atom in thiols to bond to gold surface. For each adsorption site Au-S bond distances and adsorption energies are reported to be different. So we have separately calculated the transmission for sensor S1 adsorbed on fcc-hollow, hcp-hollow, bridge, and on-top sites on Au(111) electrodes in both termini using Au-S bond distances already reported.47-49,53,54 These transmission spectra are then compared with the transmission spectra obtained for the corresponding sensor-thymine complex. From the difference of transmission at Fermi energy the conductance changes are calculated. Typical transmission spectra for the fcc-hollow site bound sensor along with the corresponding thymine complexes are shown in Figure 9a. Analysis of the transmission spectra in this

3500 J. Phys. Chem. C, Vol. 111, No. 8, 2007

Das et al.

Figure 7. Optimized structure of the sensor-thymine complex for S3, A. HOMO and LUMO of hydrogen-bonded complex is shown (middle). The change in atomic charges after complex formation (bottom). Only differences greater than 0.01 are shown.

To characterize the conductance change due to complex formation, we define the conductance ratio

Y)

Figure 8. Two probe geometries for free sensor and (top) sensorthymine complex (bottom).

and other cases shows that there is one broad peak in the spectra corresponding to the HOMO orbital and a sharp peak corresponding to the LUMO transmission. Though the HOMO transmission peak varies, the transmission corresponding to the LUMO is sharp in all cases. The transmissions for the free sensor bound to the hcp-hollow site and the fcc-hollow site are almost the same. The transmission at EF is fairly small and difficult to see in Figure 9a, but the enlarged portion in the inset shows that the transmission decreases after complex formation. We find that the zero bias conductance decreases by 10.5% and 16.5% respectively for sensor-thymine complex 1 and sensorthymine complex 2. Complexes 1 and 2 reported here are the two possible complexes (with different orientation of thymine in space) that S1 can form with thymine. The calculated conductances of the sensors adsorbed at different sites on Au(111) electrodes are given in Table 1.

Gcomplex Gfree

(2)

where Gcomplex and Gfree are conductances with and without target molecules. Considering the results from different adsorption sites studied we find that in every case Y is in the range 0.83-0.9, i.e., 10-17% decrease in conductance after complex formation with thymine. From the data shown in Table 1 we can see that the on-top site adsorbed sensor shows more conductance (1.61 µS) than others. This value reduces to (1.43-1.45 µS) after complex formation, which corresponds to a decrease of about 11%. The decreases in the conductance of the system for hcphollow and fcc-hollow sites are similar. The I-V characteristics of the molecular junction in the case of the fcc adsorption site are shown in Figure 9b within a range of -1.0 to +1.0 V. It is nonlinear in nature with currents on the order of µA. This value of current is in agreement with previous calculations for Tour wires by Taylor et al.50 This plot also shows that at higher applied bias voltage the sensing can become somewhat better (nearly 19% decrease after complex formation). The overall shapes of the I-V curves and the magnitude of conductance change, about 10-17%, obtained here are in agreement with the previous estimates based on the tightbinding model.5 5.2. Sensors S2 and S3. We first consider the case of S2 bridged between two Au(111) electrode slabs. Since the molecule concerned here is neither linear nor planar as mentioned before, the contacts on both sides are not the same in general. We have studied a case where the sulfur atom is bound to a fcc hollow site of the left-hand-side electrode. The

Theoretical Study of Molecular Wires

J. Phys. Chem. C, Vol. 111, No. 8, 2007 3501

Figure 9. (a) Transmission spectra T(E,0) calculated for sensor S1 along with sensor-thymine complex in the case of the fcc adsorption site. The origin of the abscissa is set at the Fermi energy EF of Au. The inset shows an enlarged view around the origin. (b) Calculated I-V characteristics for the sensor before and after complex formation.

TABLE 1: Change in Conductance (calculated) after Formation of Sensor-Thymine Complex for S1 Bridging Au Electrodes with Various Adsorption Sites and for Both Possible Complexes conductance adsorption site (Au-S bond length)

two-probe system

G (µS)

fcc hollow (2.55 Å)

free sensor complex 1 complex 2 free sensor complex 1 complex 2 free sensor complex 1 complex 2 free sensor complex 1 complex 2

1.02 0.91 0.85 1.03 0.92 0.86 1.61 1.45 1.43 0.72 0.64 0.62

hcp hollow (2.55 Å) on top (2.40 Å) bridge (2.52 Å)

ratio Y 0.895 0.835 0.895 0.835 0.888 0.887 0.885 0.867

Au-S bond distance is 2.55 Å. Due to the bent nature of the molecule, the other end is close to one Au atom (near on-top site) of the right side electrode with a bond distance of 2.48 Å. When this bridged molecule forms a complex with thymine the molecule is distorted slightly, making the nearest Au-S bond distance 2.42 Å. Close inspection shows that the bonding site after complex formation is different from that of the free sensor. In this case, it was found that the zero bias conductance of sensor S2 in free condition is 0.053 µS, which is decreased to 0.033 µS after complex formation with thymine. The transmission spectrum is shown in Figure 10, which has peaks corresponding to the HOMO and the LUMO of S2. After complex formation an increased twist in the molecule possibly results in a poor overlap of the molecular orbitals of the wire portion thus decreasing the transmission peaks and consequently the conductance. Since in this case the contacts with the metallic electrodes were not identical we also performed another study where the nature of the two contacts was reversed, i.e., in the left side on-top Au-S bond (2.48 Å) and in the right side fcchollow site bonded Au-S (2.55 Å). In this case we find the conductance is 0.047 µS before complex formation and 0.037 µS after complex formation. For conductance studies of sensor S3 we have chosen the more stable conformation, i.e., A. The structure of conformer A in free condition shows (see Figure 4) that the molecular wire portion is slightly nonlinear. Because of this, it is difficult to make a molecular junction model with identical connections

Figure 10. Transmission spectra T(E,0) calculated for free sensor S2 along with the sensor-thymine complex.

on both sides. In our model, the left end of the molecule is bound to Au(111) in a fcc-hollow site (Au-S 2.55 Å), whereas the right side is bound to an on-top site (Au-S 2.45 Å). The molecule is placed perpendicular to the metallic electrodes. For this structure we find the conductance to be 1.14 µS. When the central molecule is allowed to form a complex with the thymine molecule the wire portion becomes slightly twisted in addition to being nonlinear. Hence if the electrodes are assumed to be placed at identical distance as in the free case, the right end of the molecule is not bonded (nearest Au-S ) 2.65 Å). Under such conditions the conductance of the metal-molecule-metal system is reduced to 0.89 µS only. The transmission spectra for S3 and corresponding complex are shown in Figure 11. The decrease may be attributed to the increased distance from the electrode in the right side and also to the twisting of the wire portion, which makes the overlap between the molecular orbitals smaller. We have also studied a case where the central molecule is bonded (Au-S ) 2.52 Å) to two bridge sites on the two sides and the conductance was found to decrease from 1.20 µS to 1.02 µS after complexation with thymine. 6. Experimental Observation of Conductance Changes Sensor molecules of types S1 and S2 have been synthesized2,3 for conductance measurements. Efforts have been made to synthesize a molecule of type S3 as well but it resulted in a structure different from the original design. Detection of thymine was tested for these molecules, and the best results have been obtained for the S1-type model molecule. Therefore, in this

3502 J. Phys. Chem. C, Vol. 111, No. 8, 2007

Das et al.

Figure 11. Transmission spectra T(E,0) calculated for free sensor S3 along with the sensor-thymine complex.

section we discuss the experimental results for the S1-type model sensor only. Selective detection of different DNA bases has not been checked in this study. 6.1. Sample Preparation and Measurement Methods. The direct conductivity measurements of an S1-type sensor were performed with use of gold molecular-sized nanogap junctions which were bridged by the sensor molecules. The molecular structure of the acetyl-protected S1-type sensor (S1′) is shown in Figure 12a. The nanogap junction structures were constructed on a Si substrate coated with a 300 nm thick thermally oxidized layer. The fabrication procedure for the junctions are reported elsewhere.2,55,56 The experimental process of the bridging involves immersing a clean gold surface of nanogap junction into a 0.1 mM 1,2-dichloroethane solution of S1′ for 20 h. At the same time, a 1 mM 1,2-dichloroethane solution of pyrrolidine, with nearly twice the amount of S1′, was added to the S1′ solution to cleave the acetyl groups. After removal from the solution, the junctions were immersed in a 1 mM 1,2dichloroethane solution of pyrrolidine for an hour again to cleave the protections completely. The junctions were then sonicated in ethanol for 5 min (for removal of aggregations of S1′ and cleaving pyrrolidines which were bonded with the sensor part on S1′) and dried with nitrogen. Next, the conductance measurements of S1′ were done in a few cycles of addition and removal of target 1-n-butylthymine (BT). The process of addition involves immersing the Au-S1′Au junctions into a 1 mM C2H4Cl2 solution of BT for 30 min. Since hydrogen bonds between S1′ and thymine are easily cleaved in an ethanol solution, we immersed the junctions bound to BT into an ethanol solution for the removal of BT. We used the same experimental conditions as used in a separate reference experiment with mass spectroscopy, by which we confirmed the addition and removal of BT in the organic solvent. The conductance measurements were performed in a vacuum probe station, using a Keithley 4200 Semiconductor Characterization System. All conductances reported in this section were observed at +1.0 V between two electrodes. 6.2. Results. After the bridging of S1′, nine out of twelve cases showed a clear decrease in resistance for nanogap junctions. However, the resistances of two among the twelve junctions were still over several GΩ after bridging S1′ and the resistance of another junction suddenly increased to over 10 GΩ during the cycles. These three samples were removed from the study, because stable Au-S1′-Au junctions were possibly not constructed in these cases. We therefore measured the conductances of nine samples with three cycles of addition and

Figure 12. (a) Synthesized sensor molecule S1′ protected by acetyl groups on both sulfur ends. (b) Example of observed I-V characteristics of S1′ between gold electrodes before and after the addition of target BT molecules (thin and thick curves, respectively). (c) Statistics of observed conductance changes for addition (black) and removal (gray) of target BT to S1′. The histogram is plotted for the conductance ratio Y defined in the text.

removal of BT for each sample. Moreover, since resistance changes of the first step, i.e., just after bridging of S1′ and the first addition of BT, were scattered in every sample, analysis of the results was done from the second cycle. The reason why the changes of the fist step were scattered might be explained by two possibilities: one is unstable Au-S bonds and the other is remains of pyrrolidines on sensor parts of S1′ in the first step. We estimated the sensing effects of S1′ using the nine molecular junctions. Figure 12b shows typical I-V characteristics of an Au-S1′Au junction and the junction with added BT. The I-V curves clearly indicate that the conductance of the Au-S1′-Au junction was decreased after BT was added. Table 2 shows typical conductance changes during the three cycles. The results indicate reversible and reproducible changes between BT-added and BT-removed states after the second cycle. To analyze the conduction changes we use the conductance ratio Y defined in

Theoretical Study of Molecular Wires

J. Phys. Chem. C, Vol. 111, No. 8, 2007 3503

TABLE 2: Example of Sequential Change in Conductance Observed in Repeated Addition and Removal of Target 1-n-Butylthymine (BT) Molecules in a Sample of S1′ Sensor Molecules Introduced in Nanogap Electrodes sample treatment

conductance G (10-9 S)

bare electrodes sensor molecules introduced

0.502 3.23

first cycle, BT added BT removed second cycle, BT added BT removed third cycle, BT added BT removed

4.07 9.09 4.78 7.25 3.30 3.88

Section 5.1 commonly for both addition and removal of target molecules. Figure 12c shows a histogram of Y obtained from the results of nine samples with two cycles each. Although there are rare events with extraordinarily large values especially in the case of removal, Y is mostly in the range between 0.4 and 0.9, and its average is 0.86, corresponding to a decrease in conductance by 14%. 7. Discussion In the theoretical results, the conductance turned out to depend on the adsorption sites of the molecular wire. However, decrease in conductance after complex formation is found to be a common feature in all cases, as shown in Table 1. This is important, because in actual experiments the device may contain many molecules in parallel and presumably bonded to various adsorption sites. Even if molecular hopping among different sites is assumed, the overall conductance is not expected to change if the statistical distribution of adsorption sites remains the same. Then we can observe a net decrease in conductance due to complex formation. If we compare the theoretical results for S1 and the experimental results for S1′, there is an overall qualitative agreement. The conductance changes due to complex formation in both cases are about 10-20%. But this comparison can only be qualitative, as the experimental molecule S1′ has a longer molecular wire unit with five phenyl rings instead of the three in the model sensor, S1. This is the reason why the observed conductances are 2 orders of magnitude smaller than the calculated ones. The conductance of a molecular wire is known to decrease exponentially with its length L as ca. exp(-βL), where β is about 0.5 Å-1 for conjugated wires.43 The additional length of two phenylene-ethynylene units in S1′ is about 14 Å, which roughly corresponds to the reduction of conductance by 3 orders of magnitude. Although it is not known how many molecules are responsible for actual conduction, the observed small conductances suggest that a fairly small number of molecules actually contribute to the conduction. Our ab intio studies of the molecules in free and complexed conditions have revealed the changes in geometries, molecular orbitals, and changes in atomic charges for the three sensors after complex formation. It turned out that these calculations were not conclusive enough to predict the conductance changes, since we did not find any significant correlation between the shifts in HOMO and LUMO with changes in conductances after complex formation. The primary reason for this is that the conductance changes are rather small and that the effect of the metal molecule interface becomes comparatively important. It is interesting to note that for all three model sensors we find the conductance to decrease after complex formation with thymine. In reality there can be numerous microscopic details

of a metal molecule contact during device operation like the binding site, the metal-molecule bond distance, orientation angle, etc. which can dynamically change during the current flow. Hence theoretical estimates of current transport through a molecular junction may well differ from real values as they are too ideal in nature. In spite of these unpredictable effects, the results of conductance studies and the calculated I-V curve for S1 qualitatively match the experimental results discussed here. In conclusion, we have proposed three model sensors which can detect the thymine base of DNA. We have studied the changes in geometries, molecular orbitals, atomic charge distributions, and conductances of the sensors after complex formation with thymine. It was possible to synthesize, fabricate the device with use of gold nanogap junctions, and perform conductance studies in the case of model molecular-wire sensor S1 and it exhibited a conduction change that is in qualitative agreement with theoretical results. This demonstrates the feasibility of the proposed sensor scheme, although much work is still necessary to make the sensor more efficient and reliable. In this study we did not address the issue of selectivity, which is an important aspect of any sensor. Also, here only hydrogen bond formation with thymine is taken into account. However, one can think of situations where the binding of a target has a more drastic effect, like rotation of a benzene ring in the wire portion, effect of coordinate bond formation with the target,57 appropriate substitution in the wire portion, etc. Under such conditions the change in conductance due to complex formation may become large. Different target nucleic bases can also be detected by changing the receptor unit. Acknowledgment. We thank Dr. Takako Iizuka-Sakano, Dr. Yukihiro Shimoi, and Dr. Takao Ishida for useful discussions. This work was supported by NEDO under the Nanotechnology Program (Synthetic Nano-Function Materials Project). References and Notes (1) Vercoutere, W.; Akeson, M. Curr. Opin. Chem. Biol. 2002, 6, 816822. (2) Naitoh, Y.; Horikawa, M.; Yatabe, T.; Funaki, T.; Abe, H.; Liang, T.-T.; Suzuki, Y.; Shimizu, T.; Kawanishi, Y.; Mizutani, W. Nanotechnology 2006, 17, 2406-2410. (3) Yatabe, T. Unpublished. (4) Tsuzuki, S.; Kawanishi, Y.; Abe, S. Biosens. Bioelectron. 2005, 20, 1452-1457. (5) Abe, S.; Iizuka-Sakano, T. Synth. Met. 2005, 152, 281-284. (6) Souteyrand, E.; Cloarec, J. P.; Martin, J. R.; Wilson, C.; Lawrence, I.; Mikkelsen, S.; Lawrence, M. F. J. Phys. Chem. B 1997, 101, 29802985. (7) Kharitonov, A. B.; Wassermann, J.; Katz, E.; Willner, I. J. Phys. Chem. B 2001, 105, 4205-4213. (8) Cui, Y.; Wei, Q.; Park, H.; Lieber, C. M. Science 2001, 293, 12891292. (9) Barbaro, M.; Bonfiglio, A.; Raffo, L. IEEE Trans. Electron DeVices 2006, 53, 158-166. (10) Fodor, S. P. A.; Read, J. L.; Pirrung, M. C.; Stryer, L.; Lu, A. T.; Solas, D. Science 1991, 251, 767-773. (11) Tyagi, S.; Kramer, F. Nat. Biotechnol. 1996, 14, 303-308. (12) McQuade, D. T.; Pullen, A. E.; Swager, T. M. Chem. ReV. 2000, 100, 2537-2574. (13) Drummond, T. G.; Hill, M. G.; Barton, J. K. Nat. Biotechnol. 2003, 21, 1192-1199. (14) Storhoff, J. J.; Lucas, A.; Garimella, V.; Bao, Y. P.; Muller, U. R. Nat. Biotechnol. 2004, 22, 883-887. (15) Nam, J. M.; Stoeva, S. I.; Mirkin, C. A. J. Am. Chem. Soc 2004, 126, 5932-5933. (16) Liu, B.; Bazan, G. C. Anal. Chem. 2004, 76, 1824-1831. (17) Ho, H. A.; Boissinot, M.; Bergeron, M.; Corbeil, G.; Dore, K.; Boudreau, D.; Leclerc, M. Angew. Chem., Int. Ed. 2002, 41, 1548-1551. (18) Gaylord, B. S.; Heeger, A. J.; Bazan, G. C. PNAS 2002, 99, 1095410957. (19) Nilsson, K. P. R.; Inganas, O. Nat. Mater. 2003, 2, 419-424.

3504 J. Phys. Chem. C, Vol. 111, No. 8, 2007 (20) Dore, K.; Dubus, S.; Ho, H. A.; Levesque, I.; Brunette, M.; Corbeil, G.; Boissinot, M.; Boivin, G.; Bergeron, M. G.; Boudreau, D.; Leclerc, M. J. Am. Chem. Soc. 2004, 126, 4240-4244. (21) Guerzo, A. D.; Mesmaeker, A. K.-D. Inorg. Chem. 2002, 41, 938945. (22) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A., Jr.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian 03, Revision C.02; Gaussian, Inc.: Wallingford, CT, 2004. (23) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299-310. (24) Becke, A. D. J. Chem. Phys. 1993, 98, 5648-5652. (25) Becke, A. D. Phys. ReV. A 1998, 38, 3098-3100. (26) Salzner, U.; Lagowski, J. B.; Pickup, P. G.; Poitier, R. A. J. Comput. Chem. 1997, 18, 1943-1953. (27) Sing, U. C.; Kollman, P. A. J. Comput. Chem. 1984, 5, 129-145. (28) Bezler, B. H.; Merz, K. M.; Kollman, P. A. J. Comput. Chem. 1990, 11, 431-439. (29) Sigfridsson, E.; Ryde, U. J. Comput. Chem. 1998, 19, 377-395. (30) Boys, S. F.; Bernardi, F. Mol. Phys. 1970, 19, 553-566. (31) Simon, S.; Duran, M.; Dannenberg, J. J. J. Chem. Phys. 1996, 105, 11024-11031. (32) Brandbyge, M.; Mozos, J. L.; Ordejon, P.; Taylor, J.; Stokbro, K. Phys. ReV. B 2002, 65, 165401. (33) Taylor, J.; Guo, H.; Wang, J. Phys. ReV. B 2001, 63, 245407. (34) Soler, J. M.; Artacho, E.; Gale, J.; Garcia, A.; Junquera, J.; Ordejon, P.; Sanchez-Porta, D. J. Phys.: Condens. Matter 2002, 14, 2745-2779. (35) Troullier, N.; Martins, J. L. Phys. ReV. B 2001, 43, 1993-2006. (36) Perdew, J. P.; Zunger, A. Phys. ReV. B 1981, 23, 5048-5079.

Das et al. (37) Ceperley, D. M.; Alder, B. J. Phys. ReV. Lett. 1980, 45, 566-569. (38) Perdew, J. P.; Bruke, K.; Ernzerhof, M. Phys. ReV. Lett. 1996, 77, 3865-3868. (39) Xue, Y.; Dutta, S.; Ratner, M. J. Chem. Phys. 2001, 115, 42924299. (40) Datta, S. Electronic Transport in Mesoscopic Systems; Cambridge University Press: New York, 1996. (41) Tour, J. M.; Kozaki, M.; Seminario, J. M. J. Am. Chem. Soc. 2001, 120, 8486-8493. (42) Das, B.; Abe, S. J. Phys. Chem. B 2006, 110, 4247-4255. (43) Ishida, T.; Mizutani, W.; Aya, Y.; Ogiso, H.; Sasaki, S.; Tokumoto, H. J. Phys. Chem. B 2002, 106, 5886-5892. (44) Seminario, J. M.; Zacarias, A. G.; Tour, J. M. J. Am. Chem. Soc. 2000, 122, 3015-3020. (45) Derosa, P. A.; Seminario, J. M. J. Phys. Chem. B 2001, 105, 471481. (46) Seferos, D. S.; Trammell, S. A.; Bazan, G. C.; Kushmerick, J. G. PNAS 2005, 102, 8821-8825. (47) Hayashi, T.; Morikawa, Y.; Nozoye, H. J. Chem. Phys. 2001, 114, 7615-7621. (48) Cao, Y.; Ge, Q.; Dyer, D. J.; Wang, L. J. Phys. Chem. B 2003, 107, 3803-3807. (49) Felice, R. D.; Selloni, A.; Molinari, E. J. Phys. Chem. B 2003, 107, 1151-1156. (50) Taylor, J.; Brandbyge, M.; Stokbro, K. Phys. ReV. Lett. 2002, 89, 138301. (51) Strokbro, K.; Taylor, J.; Brandbyge, M.; Mozos, J.-L.; Ordejon, P. Comput. Mater. Sci. 2003, 27, 151-160. (52) Crljen, Z.; Grigoriev, A.; Wendin, G.; Strokbro, K. Phys. ReV. B 2005, 71, 165316. (53) Cyganik, P.; Buck, M.; Azzam, W.; Woll, C. J. Phys. Chem. B 2004, 108, 4989-4996. (54) Jang, S. S.; Jang, Y. H.; Kim, Y.-H.; Goddard, W. A., III; Flood, A. H.; Laursen, B. W.; Tseng, H.-R.; Stoddart, J. F.; Jeppesen, J. O.; Choi, J. W.; Steuerman, D. W.; DeIonno, E.; Heath, J. R. J. Am. Chem. Soc. 2005, 127, 1563-1575. (55) Naitoh, Y.; Linag, T.-T.; Azehara, H.; Mizutani, W. Jpn. J. Appl. Phys. 2005, 44, L472-L474. (56) Liang, T.-T.; Naitoh, Y.; Horikawa, M.; Ishida, T.; Mizutani, W. J. Am. Chem. Soc. 2006, 128, 13720-13726. (57) Das, B.; Abe, S. J. Phys. Chem. B 2006, 110, 23806-23811.