Multi-Copper-Mediated DNA Base Pairs Acting as Suitable Building

Jan 24, 2011 - Electronically, the equi-number H-by-Cu replacement not only leads to considerable reductions of the HOMO−LUMO gaps and ionization ...
0 downloads 0 Views 4MB Size
ARTICLE pubs.acs.org/JPCC

Multi-Copper-Mediated DNA Base Pairs Acting as Suitable Building Blocks for the DNA-Based Nanowires Genqin Li, Haiying Liu, Xiaohua Chen, Laibin Zhang, and Yuxiang Bu* The Center of Modeling & Simulation Chemistry, Institute of Theoretical Chemistry, Shandong University, Jinan 250100, People’s Republic of China

bS Supporting Information ABSTRACT: Transition-metal-mediated base pairs are under intense research because of their potential application in nanoscale molecular devices. To pursue suitable building blocks for DNA-based molecular wires, a three-copper-mediated guaninecytosine (G3CuC) and a two-copper-mediated adenine-thymine (A2CuT) base pair were designed by equi-stoichiometric H-by-Cu replacements in this Article. Their structural and electronic properties were examined by theoretical methods. Geometrically, G3CuC and A2CuT have great resemblances to the natural GC and AT with a size-expansion of about 1.0 Å due to the larger radii of Cu(I). Their significantly larger binding energies promise them to be structurally suitable for DNA helix construction. Electronically, the equi-number H-by-Cu replacement not only leads to considerable reductions of the HOMO-LUMO gaps and ionization potentials, but also enhances transverse electronic communication within isolated G3CuC and A2CuT pairs, revealed by the charge-transfer transitions in the UV absorption spectra of G3CuC and A2CuT. To further examine the effect of H-by-Cu substitution on conductivity, three-layer-stacked G3CuC and A2CuT of repeat and cross sequences were studied with positive results obtained. It can be reasonably concluded that the multi-Cu-mediated G3CuC and A2CuT pairs are promising candidates for building blocks of the Cum-DNA nanowires. This work would open a new prospective for rational design of the DNA-based molecular wires by multimetal incorporation.

1. INTRODUCTION DNA’s remarkable structural features of nanometer-scale and interbase stacking make it one of the most promising materials for nanodevice constructions, such as potential conductive molecular wires and so on. It has been proposed that the stacked aromatic bases of DNA may act as a “π-way” for efficient transfer of electrons.1,2 However, both experimental3-5 and theoretical6-8 explorations about DNA conductivity have obtained controversy results: insulating,9-11 or semiconductors,12 or large dc,13 or ac conductivity,14 or even superconductivity.15 Although the conductivity of the native DNA is still under debate, improvement of the DNA’s conductivity for the DNA-based molecular wires becomes feasible. Therefore, the functionalized DNAs with modified components are shifted into the limelight, and the inner nucleobases and base pairs have recently been the main targets for modification, often with applications of new bases or base pairs in the nanodevices.16-22 Transition metals are good carriers of many functions particularly in the nanoworld, and, therefore, they are also utilized for functionalization of DNA23 among various modification schemes. Transformation of nucleic acids into conducting wires, by means of metallization processes,11,24-26 is currently being intensively explored.27,28 It is believed that introduction of metalmediated base pairs into DNA would intrinsically confer a variety of metal-based properties and functions upon DNA. For example, in 1999, Aich et al observed a drastic change in conductivity of DNA when the imino-protons of the GC pairs were replaced r 2011 American Chemical Society

with divalent metal ions (Zn2þ).29 Also, both experimental29-34 and theoretical12,35-40 efforts have been devoted to investigations of the metal-mediated base pairs. Examinations on electronic properties of the metal-mediated (e.g., Zn2þ, Co2þ, and Fe2þ) poly(dG)-poly(dC) complexes, denoted as M-DNA, reveal that when doped with suitable metals, M-DNA can be turned into semiconductors and even conductors.12 That is, by careful selection of appropriate nucleobases and candidate metals, MDNA could be converted into nanowires with good conductivity. Inspired by these previous studies, and with the aim at the pursuit of suitable building blocks for the DNA-based molecular wires, we herein report a design of two multicopper-mediated base pairs on the basis on the following considerations. First, the natural G, C, A, and T bases were adopted as metal ligands, and the N and O atoms involved in the Watson-Crick H-bonds become the bonding atoms with metal ions. Second, it is hoped that the inserted metal ions can chemically interact with the bases so that metal ions can intrinsically alter the electronic properties of the metal-mediated base pairs and M-DNA. For this purpose, Cu is selected in this design because it is the best candidate to mimic the covalent character of the H-containing bonds (X-H, X = C, N, and O). Finally, as our main objective, we hope that the enhanced conducting ability of the multicopper-mediated DNA Received: August 12, 2010 Revised: December 1, 2010 Published: January 24, 2011 2855

dx.doi.org/10.1021/jp107605k | J. Phys. Chem. C 2011, 115, 2855–2864

The Journal of Physical Chemistry C

ARTICLE

Figure 1. Optimized geometries of G3CuC and A2CuT and atomic numbering schemes. The relative changes for bond lengths in the purine and pyrimidine moieties of G3CuC and A2CuT to those of the natural GC and AT pairs are marked out, only for the changes larger than 0.010 Å (“þ” standing for elongating and “-” for shortening).

(Cum-DNA) origins from the intrinsic properties of building blocks and Cum-DNA, such as decrease of the gap between the highest occupied molecular orbital (HOMO) and the lowestunoccupied molecular orbital (LUMO), instead of the geometrical distortions of the Cu-mediated base pairs and Cum-DNA from the natural Watson-Crick H-bonded base pairs and B-DNA. Thus, all the Watson-Crick H-bond protons of the GC and AT base pairs are substituted by equi-stoichiometrical Cu(I), as called an equi-stoichiometrical H-by-Cu replacement scheme. Naturally, the two base pairs we design here are threeCu-mediated GC (G3CuC) and two-Cu-mediated AT (A2CuT) base pairs, respectively, as shown in Figure 1. It is expected that multicopper participation in DNA base pairings could introduce much stronger effect on the base pairs than does the single-metal participation. To confirm our initial hypothesis, a density functional theory (DFT) study on the structural and electronic properties of G3CuC and A2CuT was performed in the gas phase, and also for their three-layer stacks. Investigations reveal that either of two modified base pairs is planar and has great resemblance to their corresponding natural Watson-Crick ones geometrically, which is the basis for construction of the DNA-like architectures. More importantly, they possess much smaller HOMO-LUMO gaps and ionization potentials, and enhanced transverse electronic communication within the isolated G3CuC and A2CuT base pairs, implying their high charge transfer efficiencies along Cum-DNA. Although we did not investigate complete Cum-DNA duplexes, the results obtained here can provide valuable insights into the electronic properties and the charge-conducting ability of the Cum-DNA duplexes constructed by G3CuC and A2CuT or their hybrids with the native base pairs.

2. COMPUTATIONAL DETAILS Geometry optimizations of single G3CuC and A2CuT and the natural Watson-Crick GC and AT pairs were carried out in the gas phase without any geometrical restriction using the Gaussian 03 program.41 Because previous theoretical calculations have shown that the B3LYP approach is a cost-effective method for studying transition-method ligand systems,42 the hybrid threeparameter B3LYP method was used to fully optimize the G3CuC and A2CuT geometries at the 6-31þG*(H,C,N,O)/LANDL2DZ(Cu) hybrid basis set level, followed by frequency analyses to confirm that the obtained species represent true minima on the potential energy surfaces. To examine the stabilities of G3CuC and A2CuT, their binding energies were determined with corrections for the basis set superposition errors (BSSE) using the counterpoise procedure.

Figure 2. Occupied molecular orbitals of G3CuC and A2CuT associated with the Cu-N and Cu-O interactions.

The separation patterns are shown in Figure 3 with the corresponding binding energies listed in parentheses. The electronic 2856

dx.doi.org/10.1021/jp107605k |J. Phys. Chem. C 2011, 115, 2855–2864

The Journal of Physical Chemistry C

Figure 3. Separation schemes and the corresponding binding energies (in the parentheses, in kcal/mol) for G3CuC and A2CuT. The binding energies for the natural GC and AT pairs are -28.39 and -13.34 kcal/ mol at the same level of theory, respectively.

properties of G3CuC and A2CuT were also calculated at the same level (B3LYP/6-311þþG**(H,C,N,O)/LANDL2DZ(Cu)). Because Cu has a slight smaller electronegativity relative to H, electronic rearrangement may take place for G3CuC and A2CuT as compared to the natural base pairs, as confirmed by the natural charges revealed by NBO analyses. The electronic properties of G3CuC and A2CuT were also determined to evaluate the suitableness of G3CuC and A2CuT as building blocks of the DNA-based molecular wires from the aspects of molecular orbitals and ionization potentials. Ultraviolet absorption spectra of G3CuC and A2CuT were also examined using the configuration interaction singles (CIS)43 method at the same 6-311þþG** (H,C,N,O)/ LANDL2DZ(Cu) basis set level, focused to explore chargetransfer transitions associated with the electronic communication between two pairing bases, which could facilitate the charge transport along DNA duplexes. In addition, we also investigated three-layer stacks of G3CuC and A2CuT to further clarify the conductivity of their assemblies. To simplify the calculations, structural parameters of the natural B-DNA were employed to construct repeat and cross three-layer stacked G3CuC, (G3CuC)3, and three-layer stacked A2CuT, (A2CuT)3, without any optimizations, that is, keeping the interplane-distance as 3.4 Å and the helical angle as 36°, and the phosphate-sugar backbones not included. Molecular orbitals and vertical ionization potential (IPv) of these stacks were computed at the B3LYP/6-31þG*(H,C,N,O)/LANDL2DZ(Cu) level.44 All the same properties of the natural GC and AT base pairs and their three-layer-stacked analogs were also investigated at the same levels of theory for comparison purpose to unravel the Cu incorporation effect.

3. RESULTS AND DISCUSSION Here, we make a detailed analysis to obtain a basic understanding of the changes of structural and electronic properties that take place in the presence of Cu. For the A, T, G, and C bases in G3CuC and A2CuT, we adopt the customary naming scheme of atomic sites as in the natural Watson-Crick pairs. As for Cu ions, we denote them as Cu1, Cu2, and Cu3 in G3CuC and Cu1, Cu2 in A2CuT, respectively (Figure 1). The references of the natural GC and AT pairs are always included to reveal the effects of Cu incorporation. 3.1. Geometric and Bonding Characters. Geometric Characters. Figure 1 provides the optimized geometries of G3CuC and A2CuT. Overall, G3CuC and A2CuT have great geometrical resemblances to the natural Watson-Crick GC and AT pairs. That is,

ARTICLE

equi-substitutions of H-by-Cu do not induce any significant structural distortions for the two base pairs. Like the natural GC and AT pairs, G3CuC and A2CuT are both planar within chemical accuracy, as indexed by representative dihedral angles of C2(G/A)-N1(G/A)N3(C/T)-C4(C/T), being about 180°. As is known, the covalent radius of Cu is 1.17 Å, while that of H is 0.32 Å. Hence, G and C in G3CuC and A and T in A2CuT should be further separated away from each other than those in the natural GC and AT pairs. To illustrate the expansion sizes quantitatively, Table 1 lists the interatomic distances of N/O 3 3 3 N, which cross the Cu in G3CuC and A2CuT, and also those in the natural GC and AT pairs. Taking the corresponding values of GC as reference, the O6(G)-N4(C) distance in G3CuC is elongated to be 3.804 from 2.825 Å (GC), stretched by 0.979 Å, while the N1(G)-N3(C) and N2(G)-O2(C) distances are elongated by 0.877 and 0.875 Å, respectively. Clearly, equi-substitution of H-by-Cu introduces a slight bend toward the minor groove as evidenced by elongation of O6(G)-N4(C) slightly larger than that of N2(G)-O2(C). Similarly, taking the native AT as reference, N6(A)-O4(T) and N1(A)-N3(T) in A2CuT are stretched by 0.849 and 0.962 Å, respectively. Obviously, H-by-Cu equi-replacement makes A and T bend slightly toward the major groove. The H-by-Cu equi-replacement-induced bending direction in AT is opposite to that in GC. Another index of the Cuincorporation-induced expansion is the interatomic distances between N9(purine) and N1(pyrimidine), both of which are the sugar-linking atomic sites. The N9(G) 3 3 3 N1(C) distance in G3CuC is 9.934 Å, and that in A2CuT is 10.023 Å, respectively. Both of them are longer than the corresponding values in the natural GC and AT pairs by about 1.0 Å, as the nearly same magnitude in the N/O 3 3 3 N distances discussed above. Relative to G3CuC, separation in A2CuT is slightly larger. These observations have seemingly implied that slightly mechanical strain may be incurred upon incorporation of these G3CuC and A2CuT motifs into DNA. However, due to the flexibility of the phosphate-sugar backbone of DNA, this slight expansion would not affect the entire architecture of DNA duplexes constructed completely by G3CuC and A2CuT (the DNA duplexes consisting of G3CuC and/or A2CuT are denoted as Cum-DNA in the following) or the hybrid DNAs consisting of G3CuC and/or A2CuT with the native base pairs. Clearly, the diameter of Cum-DNA should be about 1.0 Å larger than that of B-DNA. Consequently, the expanded sizes of these multi-Cu-mediated base pairs may implicate a possible enhancement of the stacking forces between neighboring base pairs, which is thus favorable to stability of DNA duplexes and the electronic conduction along DNA.45 Considerable structural changes in G3CuC and A2CuT have also been observed as compared to their corresponding native base pairs, mainly about the bonds associated with Cu. A notable feature is that the bond lengths are averaged for the bonds in the Watson-Crick H-bond faces of the bases. That is, the double-bond-characterized bonds in the native base pairs get elongated toward the single bonds, while the single-bond-characterized bonds get shortened toward the double bonds. For example, all CdO bonds are elongated, being between the lengths of a CdO double bond and a C-O single bond, while their adjacent C-N bonds in the Watson-Crick H-bond zones are shortened, being more of double bond property. The C-N (associated with -NH2) bonds are also shortened slightly, such as C4(C)-N4(C) and C4(A)-N4(A). Clearly, these observations probably imply an enhancement of conjugated degree of each base moiety, consequently enhancing their aromaticity and 2857

dx.doi.org/10.1021/jp107605k |J. Phys. Chem. C 2011, 115, 2855–2864

The Journal of Physical Chemistry C

ARTICLE

Table 1. Selected Bond Lengths (Å) and Angles (deg) for G3CuC and A2CuTa G3CuC Cu1-O6(G)

1.890

Cu1-N4(C)

1.927

Cu2-N1(G)

1.904

Cu2-N3(C)

1.942

Cu3-N2(G)

1.880

O6(G) 3 3 3 N1(G) N1(G) 3 3 3 N2(G) N4(C) 3 3 3 N3(C)

2.317 (þ0.023) 2.343 (þ0.020) 2.321 (þ0.030)

N3(C) 3 3 3 O2(C) O6(G)-C6(G)-N1(G)

121.1 (þ1.3)

2.309 (þ0.014)

Cu3-O2(C)

1.957

N1(G)-C2(G)-N2(G)

117.5 (þ0.7)

Cu1 3 3 3 Cu2 Cu2 3 3 3 Cu3 O6(G)-N4(C)

2.486

N4(C)-C4(C)-N3(C)

118.7 (þ1.0)

2.493 3.804 (þ0.979)

N3(C)-C2(C)-O2(C) O6(G)-Cu1-N4(C)

124.4 (þ0.1) 170.9

N1(G)-N3(C)

3.845 (þ0.877)

N1(G)-Cu2-N3(C)

179.1

N2(G)-O2(C)

3.819 (þ0.875)

N2(G)-Cu3-O2(C)

169.1

N9(G) 3 3 3 N1(C)

9.934 (þ0.884)

Cu1-Cu2-Cu3

178.9

C2(G)-N1(G)-N3(C)-C4(C)

180.0

A2CuT

a

Cu1-N6(A)

1.890

Cu1-O4(T) Cu2-N1(A)

1.912 1.941

N6(A) 3 3 3 N1(A) O4(T) 3 3 3 N3(T) N6(A)-C6(A)-N1(A)

2.297 (þ0.015) 121.2 (þ1.7)

2.362 (þ0.031)

Cu2-N3(T)

1.942

O4(T)-C4(T)-N3(T)

121.2 (þ0.8)

Cu1 3 3 3 Cu2 N6(A)-O4(T)

2.571

N6(A)-Cu1-O4(T)

173.1

3.795 (þ0.849)

N1(A)-Cu2-N3(T)

172.5

N1(A)-N3(T)

3.876 (þ0.962)

C6(A)-N1(A)-N3(T)-C4(T)

180.0

N9(A) 3 3 3 N1(T)

10.023 (þ1.020)

Values in parentheses are the relative values of the bond lengths and angles of G3CuC and A2CuT relative to those of natural GC and AT base pairs where “þ” denotes increased and “-” stands for decreased.

Table 2. Natural Charges for Some Atoms of G3CuC and A2CuT Obtained from NBO Analysis at the Hybrid 6-311þþG** (H, C, O, N)/LANL2DZ (Cu) Basis Set Level with Corresponding Values of the Natural GC and AT Pairs for Comparison G3CuC

GC

A2CuT

AT

O6(G)

-0.797

-0.681

N6(A)

-0.908

-0.765

N1(G)

-0.756

-0.645

N1(A)

-0.702

-0.620

N2(G)

-0.988

-0.791

O4(T)

-0.761

-0.655

N4(C) N3(C)

-0.848 -0.724

-0.738 -0.655

N3(T) Cu1/H61

-0.754 0.719

-0.660 0.435

O2(C)

Cu2/H3

0.699

0.462

-0.753

-0.688

Cu1/H42

0.741

0.450

Cu2/H1

0.639

0.447

Cu3/H22

0.726

0.437

further affecting the electronic properties of G3CuC and A2CuT. The structural changes may be attributed to the Cu-induced charge rearrangements, which can consequently lead to the structural alterations. Hence, a natural bond orbital (NBO) analysis was done. As expected, charge rearrangements in G3CuC and A2CuT occur on the atoms near Cu, as listed in Table 2. X-Cu(I)-Y Bonds. One important characteristic of G3CuC and A2CuT is the X-Cu(I)-Y (X and Y stand for N or O atoms neighboring Cu atoms) bonds. The H-by-Cu replacements completely change the bonding patterns of the Watson-Crick H-bond zones of the natural base pairs. Different from the geometrical characteristic of the X-H 3 3 3 Y Watson-Crick

H-bonds in which the proton resides at one side, forming a covalent bond (X-H) and a H-bond (H 3 3 3 Y), each monovalent cation Cu resides at the center between its linked N and O with the Cu(I)-N/O bond lengths of 1.87-1.96 Å. Inspection of the associated molecular orbitals of G3CuC and A2CuT reveals the presence of the σ-type metal-ligand coordinative bonds between Cu and its adjacent N and O atoms. From the molecular orbitals of G3CuC shown in Figure 2, it can be easily known that (a) σ-type coordinative bonds exist between Cu and its neighboring N or O atoms; (b) these Cu-N/O coordinative bonds are mainly contributed by N and O atoms; and (c) the dx2-y2 and dxy orbitals of Cu participate in the formation of these Cu-N/O coordinative bonds. When necessary, they adjust themselves to appropriate directions to effectively interact with the p orbitals of N or O to yield larger overlaps, as shown in HOMO-7, HOMO-18, HOMO-24, and HOMO-25. It is the same for A2CuT. Clearly, each Cu atom in G3CuC or A2CuT has identical coordinative environments, that is, linearly two-coordinated by one N (O) atom of purine and an opposite N (O) of pyrimidine, and interacts with them almost equally with the Cu-N/O distances ranging from 1.87 to 1.96 Å, the typical Cu-ligand distances.46,47 Furthermore, IR frequencies assigned to the Cu-N and Cu-O stretching vibrations can also be found in the IR spectra of G3CuC and A2CuT, as listed in Table 3. The symmetric and antisymmetric stretching frequencies of X-Cu-Y in G3CuC are identified at 244 and 355 cm-1, respectively; besides, stretching vibrations of some Cu-X bonds are found at 301 cm-1. In A2CuT, the antisymmetric frequencies of X-Cu-Y are identified at 225 and 288 cm-1, while those of some Cu-X bonds are at 374 cm-1. Clearly, these stretching 2858

dx.doi.org/10.1021/jp107605k |J. Phys. Chem. C 2011, 115, 2855–2864

The Journal of Physical Chemistry C

ARTICLE

Table 3. IR Frequencies Assigned to Cu-N/O Stretching Vibrations (Symmetric, νs, and Asymmetric, νas) and the Cu 3 3 3 Cu Stretching Vibrations of G3CuC and A2CuT Calculated at the 6-311þþG**(H,C,N,O)/LANDL2DZ(Cu) Level of Theory frequencies/cm-1

stretching vibration assignments

intensity

G3CuC 152 200 244

νs (Cu1-Cu2-Cu3) νas (Cu1-Cu2-Cu3) νs (O6(G)-Cu1-N4(C))

0.12 1280 4.99

νs (N1(G)-Cu2-N3(C)) νs (N2(G)-Cu3-O2(C)) 301

νs (Cu1-N4(C))

2.97

νs (Cu2-N3(C)) νs (Cu3-O2(C)) 355

νas (O6(G)-Cu1-N4(C)) νas (N1(G)-Cu2-N3(C))

28.66

νas (N2(G)-Cu3-O2(C)) A2CuT 160

νs (Cu1-Cu2)

0.73

190

νs (Cu1-Cu2)

1.70

225

νas (N6(A)-Cu1-O4(T))

0.11

288

νas (N1(A)-Cu2-N3(T))

16.13

374

ν (Cu1-O4(T))

14.90

ν (Cu2-N3(T))

vibrational frequencies reflect the bonding modes and strengths of the possible Cu-X or X-Cu-Y (X = N, O) bonds.48 In short, the X-Cu-Y bond can be understood as a resonant average result between Xδ--Cuδþ r :Y and X: f δþCu-δ-Y, which are similar to the proton transfer tautomers of the H-bonded units, but they are really stronger than the H-bonded units (X-H 3 3 3 Y or X 3 3 3 H-Y) due to the multiple coupling interactions of Cu atomic orbitals with the bonded atoms (X, Y = N or O). Binding Energies. The σ-type metal-ligand coordinative bonds between Cu and its adjacent N and O atoms imply that purine and pyrimidine bases should connect more tightly in G3CuC or A2CuT than in the natural GC and AT pairs where they bind with each other through H-bonding. Calculations of binding energies quantitatively prove this point. As for the separation schemes, we only adopted those dividing G3CuC and A2CuT into two neutral pyrimidine and purine parts, as shown in Figure 3. For G3CuC, the binding energies are -167.01, -137.75, and -141.16 kcal/mol, respectively, depending on different separation patterns. Although they differ from each other, they are in the same order of magnitude. The differences may be attributed to different numbers of the Cu-O and Cu-N bonds each pattern involves. As a whole, G3CuC is more stable than the natural GC whose binding energy is -28.39 kcal/mol. A similar effect also takes place on A2CuT. Depending on dividing patterns, the binding energies of A2CuT are -91.43 and -112.87 kcal/ mol, respectively, basically in the same order of magnitude. Clearly, the stability of A2CuT is higher than that of the natural AT pair. As a summary, no matter for any separation patterns, it can be sure that the pyrimidine and purine bases in G3CuC and A2CuT are bonded together more tightly than in the natural GC and AT pairs. Because H-bonding in the Watson-Crick

H-bonded pairs is one of the factors responsible for DNA stability, it is reasonable to predict that Cum-DNA assembled by these G3CuC and A2CuT motifs should be more thermodynamically stable than B-DNA. Cu 3 3 3 Cu Cuprophilic Bonds. Another interesting phenomenon is there are short Cu 3 3 3 Cu bonds in G3CuC and A2CuT. The interatomic Cu 3 3 3 Cu distances of 2.49 Å in G3CuC and 2.57 Å in A2CuT are larger than the 2-fold covalent radius of Cu (ca. 1.17 Å) but smaller than the 2-fold van der Waals radii of Cu (ca. 2.00-2.27 Å).49 The Cu 3 3 3 Cu distances may be a clear indicator for d10-d10 Cu(I) 3 3 3 Cu(I) interaction between adjacent Cu(I). Recently, a topic regarding the Cu 3 3 3 Cu bonding in the multi-Cu(I) complexes has stimulated a great deal of discussions and controversies.50 Cotton et al. established the nonexistence of Cu(I)-Cu(I) covalent bonding for the complexes in which transition metals have partially occupied d-orbitals, and the short Cu(I) 3 3 3 Cu(I) distances are ascribed to strong Cu(I)-ligand bonding and very small bite distances (e.g., 2.2 Å) of the ligands.51,52 Nevertheless, some other studies suggested genuine intramolecular cuprophilicity on the basis of spectroscopic evidence in which the observed resonance Raman bands at 104 cm-1 could be attributed to the Cu-Cu vibrations (νCu-Cu).53-55 Clearly, the cuprophilic bonding is not covalent bonding with a bond order of one. That is, no direct molecular orbital interaction is present between the two adjacent Cu (I, d10) with a d10 electronic configuration. In general, the cuprophilic bonding, one kind of metallophilicity like other metallophilic bondings between adjacent d10 metal ions such as Au(d10) or Ag(d10), is attributed to correlation effects that are strengthened by relativistic effects.56,57 Herein, for G3CuC and A2CuT, the Cu 3 3 3 Cu interaction falls into the category of the ligand-assisted cuprophilic bonding (strong Cu-X bonding and small bite sizes of 2.297-2.362 Å) as some other similar systems,50-52,58,59 supported by the corresponding stretching vibrational frequencies of Cu-Cu in the IR spectra of G3CuC and A2CuT (Table 3). In G3CuC, the symmetric stretching vibration of Cu1-Cu2-Cu3 is at 152 cm-1, while the asymmetric one is at 200 cm-1. For A2CuT, the stretching vibrational frequencies of Cu1-Cu2 are 160 and 190 cm-1. These values are in the typical cuprophilic bonding zone,60 exactly indicating the existence of the cuprophilic bonding in G3CuC and A2CuT. 3.2. Electronic Property Analysis for Potential Conductivity. It is reasonable to expect that electronic properties of G3CuC and A2CuT could be quite different from those of the natural GC and AT pairs. Now, the main question we wish to address is if those differences can make G3CuC and A2CuT more appropriate as building blocks for the DNA-based nanowires than the native GC and AT pairs. To gain insight into this question, we have explored electronic properties of the isolated G3CuC and A2CuT motifs, as well as three-layer stacks of G3CuC and A2CuT with repeat and cross sequences. Electronic Properties of Isolated Base Pairs. It is natural to focus on the ionization potentials (IPs) and the HOMO-LUMO gaps of G3CuC and A2CuT in the respect that they are two important indexes associated with conductivity. Here, we investigate these two quantities of the natural and Cu-mediated pairs and try to achieve some interesting conclusions. The HOMO-LUMO gaps and the HOMO/LUMO distributions of the isolated G3CuC and A2CuT pairs are shown in Figure 4. The HOMO-LUMO gap of G3CuC is 2.49 eV, reduced by 1.27 eV relative to that (3.76 eV) of the natural GC pair as a result of up-shifting HOMO and down-shifting LUMO, as listed 2859

dx.doi.org/10.1021/jp107605k |J. Phys. Chem. C 2011, 115, 2855–2864

The Journal of Physical Chemistry C

ARTICLE

Figure 5. Molecular orbital energy levels of G3CuC and A2CuT. The HOMO and LUMO levels are marked with bold lines, and their gaps are marked by horizontal arrows with the corresponding values (2.49 versus 4.20 eV). Molecular orbitals stemming from or consisting of Cu atomic orbitals are red-marked. They are determined by the DFT/B3LYP method.

Figure 4. Comparison among the HOMO-LUMO gaps of GC, m-GC, G3CuC and AT, m-AT, A2CuT, and plots of HOMOs and LUMOs of G3CuC and A2CuT (isovalue = 0.04).

Table 4. Energies of the HOMO, LUMO, HOMO-LUMO (HL) Gaps (in eV), and Ionization Potentials (in kcal/mol) for G3CuC, A2CuT, GC, AT, and the Modified GC and AT (m-GC, m-AT) and Their Three-Layer Stacks Determined by the B3LYP Method systems

HOMO

LUMO

HL gap

IPv (IPa) 167.99 (159.68)

GC

-5.50

-1.74

3.76

m-GC G3CuC

-5.46 -4.69

-1.86 -2.20

3.60 2.49

144.84 (141.10)

repeat 3-GC

-4.75

-1.66

3.09

138.09

repeat 3-G3CuC

-3.79

-2.24

1.55

113.75

cross 3-GC

-5.11

-1.26

3.85

143.98

cross 3-G3CuC

-4.47

-1.62

2.85

127.69

AT

-6.18

-1.50

4.68

181.15 (177.70)

m-AT

-6.19

-1.57

4.62

A2CuT repeat 3-AT

-5.76 -5.62

-1.56 -1.48

4.20 4.14

168.80 (166.60) 158.57

repeat 3-A2CuT

-5.33

-1.47

3.86

149.46

cross 3-AT

-5.73

-1.19

4.54

158.44

cross 3-A2CuT

-5.39

-1.26

4.13

148.66

in Table 4. Like the natural GC pair, HOMO and LUMO in G3CuC are also contributed by HOMO of the guanine and LUMO of the cytosine moieties, respectively, and both are of π character and have only tiny components of Cu atoms with neglectable amounts. A similar but smaller effect takes place in A2CuT. Its HOMO-LUMO gap is 4.20 eV, smaller than that (4.68 eV) of the natural AT pair by a smaller extent, also due to the upshift of HOMO and down-shift of LUMO. Cu does not participate in the frontier orbitals either, and HOMO and LUMO in A2CuT directly come from A π-type HOMO and T π*-type LUMO, respectively, like that in the natural AT pair. Briefly, like the cases of other closed-shell metal-involved pairs,42 no Cu contributes to the frontier orbitals of G3CuC and A2CuT. As compared to the Cu(I)-imino GC pair, which is the only reference of single-Cu-mediated pair we found, incorporation

of three Cu(I) results in a significant reduction of the HOMO-LUMO gap from 3.79 eV of GC to a value of 2.49 eV for G3CuC, being much smaller than that (3.23 eV) of the Cu(I)-imino GC pair.41 To explore the origin of the HOMO-LUMO gap shrink for G3CuC and A2CuT, and to split the influences of Cu insertion and geometric alterations of the bases, two artificially perturbed base pairs, modified-GC (m-GC) and modified-AT (m-AT), were calculated. That is, fixing the Watson-Crick H-bond face distances (e.g., the N-H 3 3 3 N and N-H 3 3 3 O distances fixed) of GC and AT, the geometries of their purine and pyrimidine moieties are distorted to correspond to those in G3CuC and A2CuT. Comparison of the property parameters of them with the native GC and AT base pairs can reveal the effect from pure geometry distortions. The orbital energies were determined using the B3LYP method with a 6-311þþG**(H,C,N,O)/ LANDL2DZ(Cu) hybrid basis set. In detail, the HOMOLUMO gap of m-GC is 3.60 eV, slightly smaller than 3.76 eV of the natural GC by 0.16 eV and considerably larger than 2.49 eV of G3CuC by 1.11 eV. A similar situation also occurs in the AT series: the HOMO-LUMO gap of m-AT is 4.62 eV, which is slightly smaller than that of the natural AT by 0.06 eV but larger than that of A2CuT by 0.42 eV. In addition, although there is no weight of Cu in HOMO and LUMO of G3CuC and A2CuT, it is observed that Cu considerably contributes to the other occupied molecular orbitals immediately below HOMO both for G3CuC and for A2CuT, as shown in Figure 5. These observations suggest that geometric changes such as subtle bond alterations of the base moieties are not the main factors to affect the HOMO-LUMO gaps. Instead, the HOMO-LUMO gap shifts are essentially due to electrostatic effects rather than orbital hybridization, just like other closed-shell systems.42 Overall, it can be concluded that reductions of the HOMO-LUMO gaps of G3CuC and A2CuT as compared to those of the native GC and AT base pairs originate mainly from Cu-insertion-induced electrostatic effects, and slightly from geometric distortions of the bases. Theoretical adiabatic and vertical ionization potentials (IPa and IPv) for G3CuC and A2CuT are presented in Figure 6, and the spin densities of the ionized ones are also given. For G3CuC, its IPa and IPv values are 141.10 and 144.84 kcal/mol, respectively, both lower than the corresponding values of 159.68 and 167.99 kcal/mol of GC. The spin density of the ionized G3CuC is distributed on guanine with a π-character like the native GC base pair. The same situation occurs for A2CuT. Its IPa and IPv values are 166.60 and 168.80 kcal/mol, respectively, smaller than those of the natural 2860

dx.doi.org/10.1021/jp107605k |J. Phys. Chem. C 2011, 115, 2855–2864

The Journal of Physical Chemistry C

Figure 6. The adiabatic and vertical ionization potentials (IPa and IPv) of G3CuC and A2CuT and the natural GC and AT pairs, and spin densities of the ionized G3CuC and A2CuT (isovalue = 0.002).

AT pair, while the spin density of the ionized A2CuT resides on the A fragment with a π-character, the same as the natural AT base pair. Additionally, for either G3CuC or A2CuT, the small difference between IPa and IPv suggests that relaxation of the nuclear framework is not a crucial factor for the IP changes of G3CuC or A2CuT. Thus, we boldly believe that the IP changes are also mainly attributed to electrostatic effects of Cu insertion like the HOMO-LUMO gap shrink. In comparison with the natural GC and AT pairs, G3CuC and A2CuT are easier to oxidize due to their lower IPs. This is in agreement with a suggestion from some other related studies that a low IP is an important factor for design of the DNA-based hole-transport molecular wires.61,62 The above analyses indicate that equi-stoichiometic H-by-Cu replacements yield different impacts from viewpoints of reductions of the HOMO-LUMO gaps and IPs. Clearly, the effect of Cu on GC is more profound than that on AT. G3CuC has a HOMO-LUMO gap of 2.49 eV, with a reduction of 1.27 eV or 33.8% relative to that (3.76 eV) in the natural GC, while for A2CuT, the HOMO-LUMO gap is 4.20 eV, decreased by 0.48 or 10.3% with respect to that (4.68 eV) of the natural AT pair. As for IPs (either IPa or IPv), reduction extents are also larger for G3CuC versus GC than A2CuT versus AT. Interestingly, it is noteworthy that G3CuC and A2CuT share the same relationship as that between the natural GC and AT pairs. That is, G3CuC possesses a smaller HOMO-LUMO gap and IPs than A2CuT, and the G motif is the easiest to oxidize, implicating that G3CuC is more suitable than A2CuT for charge conduction along Cum-DNA, and the electron-loss center created in Cum-DNA is still localized at the G moieties. In summary, because the HOMOLUMO gap of a building block is closely related to the band gap of DNA, and IP can measure the ability of a pair to oxidize, it can be predicted that the Cum-DNA constructed by G3CuC and A2CuT could have smaller band gaps. This makes it much more possible to achieve wires with good conduction properties. UV Spectroscopic Evidence of Electronic Communication. As we all know, TD-DFT and CIS are two relatively cheaper methods in investigating the excited states. Although the TDDFT/B3LYP method can accurately predict energies and transition dipole moments for valence state, it seriously underestimates the energies of the charge-transfer states.63-65 Thus, prediction of low-lying charge-transfer states by the TDB3LYP method should be treated with extreme caution. We only present the CIS results here, with the help of the SWizard program,

ARTICLE

Figure 7. Ultraviolet absorption spectra of G3CuC and GC determined from CIS calculations. Charge-transfer transitions are marked out by red and black arrows for G3CuC and GC, respectively.

Figure 8. Ultraviolet absorption spectra of A2CuT and AT determined from CIS calculations. Charge-transfer transitions are marked out by red and black arrows for A2CuT and AT, respectively.

revision 4.4,66 using the Gaussian mode. It was found in early studies that the computed and scaled electronic transitions at the CIS level are in good agreement with the corresponding experimental data.67-71 The UV absorption spectrum of a system essentially connects with its electronic structures. Because the HOMO-LUMO gaps of G3CuC and A2CuT are small and Cu contributes to the molecular orbitals below HOMOs for both of them, they should exhibit different ultraviolet absorption characters from those of the natural GC and AT pairs. Here, we only analyze the charge-transfer transitions (e.g., excited from π(purine) to π*(pyrimidine), or from π(purine) to d*(Cu), or from π(pyrimidine) to d*(Cu), or those in the opposite directions) because the charge-transfer transitions directly reflect the electronic communications qualitatively, which may facilitate the conductivity of DNA.47,72,73 The absorption peaks assigned to the charge-transfer transitions are marked out by arrows as shown in Figures 7 and 8. Figure 7 shows the UV absorption spectra of G3CuC and the natural GC with charge-transfer transitions marked by red arrows for G3CuC, while the black ones are for the natural GC. From the 2861

dx.doi.org/10.1021/jp107605k |J. Phys. Chem. C 2011, 115, 2855–2864

The Journal of Physical Chemistry C viewpoint of absorption wavelengths, charge-transfer transitions for G3CuC are red-shifted with the onset absorption peak at 340.4 nm, while the delocalized transitions are higher in number (5 versus 2 for the natural GC pair). In Figure 8, the UV absorption spectra of A2CuT and the natural AT are also presented. Charge-transfer transitions of A2CuT are red-shifted relative to those of the natural AT pair; moreover, the number of charge-transfer transitions increase to 4, while there are only 2 for the natural AT pair. Considering the small HOMO-LUMO gaps and Cu’s participation only in the orbtials just below HOMOs for both G3CuC and A2CuT, these results are not unexpected. These increased charge-transfer transitions in G3CuC and A2CuT imply an enhancement of the transverse electronic communications between two base moieties, relative to the natural GC and AT base pairs. It is reasonably believed that improvement of the transverse charge transfer or delocalization in G3CuC and A2CuT could exhibit a positive effect on charge migration along duplex DNA. Stacking Effect. If modified appropriately, DNA is believed to be charge-conductive. With narrow HOMO-LUMO gaps and small IPs, G3CuC and A2CuT are believed to be able to improve charge conductivity of the DNA duplexes if they are incorporated in, which is exactly our target to design G3CuC and A2CuT. To further confirm our hypothesis, we constructed three-layer stacks of G3CuC and A2CuT with both repeat and cross sequences as models (Figure 9). The stacks of three-layer G3CuC and threelayer A2CuT were constructed by adopting the structural parameters of the natural B-DNA with stacking separation of 3.4 Å and twist angle of 36°. Although these stacked models consist of only three layers and do not effectively mimic a real Cum-DNA helix, some important information can be still extracted. Therefore, we are confident that the result obtained here is sufficient to qualitatively reflect the conducting ability of the duplex CumDNA. Like the isolated G3CuC and A2CuT, it is still a natural choice to focus on the HOMO-LUMO gaps and IPv in consideration because they are two important indexes of conductivity. The main information is collected in Table 4 and Figure 9. In the isolated G3CuC and A2CuT pairs, Cu’s incorporation narrows the HOMO-LUMO gaps. For the three-layer stacks of G3CuC and A2CuT with either repeat or cross sequences, H-by-Cu replacements yield the same effect: narrowing the HOMOLUMO gaps by up-shifting π-type HOMOs on the purine bases and down-shifting π*-LUMOs on the pyrimidine bases. Repeat 3-G3CuC has the smallest HOMO-LUMO gap of 1.55 eV, 48.84% smaller than 3.09 eV of the natural repeat 3-GC. Followed is the cross 3-G3CuC with a HOMO-LUMO gap of 2.85 eV, repeat 3-A2CuT of 3.86 eV, and cross 3-A2CuT of 4.13 eV. It is clear that Cu decoration has a stronger impact for repeat 3-G3CuC and cross 3-G3CuC than repeat 3-A2CuT and cross 3-A2CuT. This may relate to the differences between the fundamental natures of the GC and AT pairs, or the number of Cu that participate because G3CuC has three Cu atoms while A2CuT involves only two Cu atoms. Another point that should be mentioned is that, for the natural systems, the repeat 3-GC and 3-AT possess narrower HOMO-LUMO gaps than do their cross analogues, respectively. The results are consistent with the fact that the repeat-sequence DNA is better charge-conductive than the cross-sequence DNA. In the Cu-mediated systems, the same phenomena also appear: the repeat 3-G3CuC and 3-A2CuT have narrower HOMO-LUMO gaps than the cross 3-G3CuC and 3-A2CuT. This observation indicates that Cum-DNA of the repeat sequence conducts better than that with the cross sequence. It is a pity that H-by-Cu substitution does not eliminate the

ARTICLE

Figure 9. The HOMO and LUMO plots (isovalue = 0.04) and spin density distributions (isovalue = 0.001) of the repeat and cross threelayer stacks of G3CuC and A2CuT, and also of the natural GC and AT pairs obtained using the B3LYP method.

sequence selectivity when designing Cum-DNA wires. Except for the repeat 3-A2CuT, distributions of HOMOs and LUMOs of other three Cu-decorated stacks are different from those of the corresponding natural stacks, as shown in Figure 9. However, on the whole, there are no essential alterations. That is, all HOMOs are located on purine with a π character, and all LUMOs are localized on pyrimidine with a π* character. As for the vertical IPs, 3-G3CuC and 3-A2CuT with either repeat or cross sequences possess smaller vertical IPv values than do the corresponding natural ones. Of all Cu-mediated stacks, the repeat 3-G3CuC has the smallest IPv of 113.75 kcal/mol, while the cross 3-G3CuC has IPv of 127.69 kcal/mol. The approximate IPv values of repeat and cross 3-A2CuT are 149.46 and 148.66 kcal/mol, respectively. These data show that Cu decoration has a stronger impact for the repeat 3-G3CuC and the cross 3-G3CuC than for the repeat 3-A2CuT and the cross 3-A2CuT. This difference may be attributed to the fundamental nature of GC and AT, or the number of Cu participating in the mediation of the base pairs in which G3CuC has three Cu atoms while A2CuT only possesses 2862

dx.doi.org/10.1021/jp107605k |J. Phys. Chem. C 2011, 115, 2855–2864

The Journal of Physical Chemistry C two Cu atoms. To sum, for the GC series, Cu’s effect on the repeat stack is larger than that on the cross stack. However, for the AT series, Cu’s effects for the repeat and the cross stacks are almost same. That is, for G3CuC, caution is still needed for sequence selectivity in designing conductive wires, but for A2CuT, Cu eliminates the selectivity for sequence by some extent. The spin densities of all stacks reside on the purine moieties, which possess lower IPs than the pairing pyrimidine moieties, indicating positions for hole trapping or electron loss. The spin densities for all three layer stacks are located on purines with π characters, which are the same as those of the corresponding natural stacks, as shown in Figure 9.

4. CONCLUSION In summary, we theoretically designed two multi-Cu-mediated GC and AT base pairs and explored their relevant electronic properties. G3CuC and A2CuT preserve the geometries of the natural Watson-Crick base pairs, such as planarity, but with about 1.0 Å size expansion, providing the structural basis for DNA helix construction. Moreover, they have some improved electronic properties, such as small HOMO-LUMO gaps, low IPs, and enhanced transverse base-to-base electronic communication, making them excellent candidates for the building blocks of nanowires with good conductivity. Interestingly, G3CuC and A2CuT share the relationship between the natural GC and AT base pairs in many aspects. G3CuC designed here possesses the smallest HOMO-LUMO gap among all similar metalGC base pairs in the closed-shell systems we have known so far. This design by equi-stoichiometric H-by-Cu replacements, inspired by previous works about metal-mediated pairs, represents a new pair-decoration by multimetal incorporation that is directed toward the rational design of appropriate building blocks of the conducting DNA nanowires. Certainly, further theoretical works on the duplex Cum-DNA and hybrid duplex DNA of G3CuC and A2CuT with the natural base pairs are still needed to examine their conductivity. In addition, because transition metals are good carriers of many functions of molecular devices in nanofields, the conjugated organic molecules or base pairs decorated by other transition metals are also hoped to be suitable candidates for constructing electronic nanodevices with excellent properties, such as good electric conductivity, magnetism, and optical properties, etc. On the other hand, this work also suggests a way to design the shortest metal-cored molecular wire (e.g., Cu-Cu-Cu) with tunable metal-metal linking properties. Examination of all these predictions is still being undertaken in our lab. ’ ASSOCIATED CONTENT

bS

Supporting Information. Vertical transition energies, oscillator strengths, and state assignments corresponding to the low-lying excited states of GC, G3CuC, AT, and A2CuT pairs; electronic properties of isolated G3CuT and A2CuT; and the reason for utilizing B3LYP but not M06-2X to examine the electronic properties of three-layer stacks. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected].

ARTICLE

’ ACKNOWLEDGMENT This work was supported by the NSFC (20633060, 20973101), NCET, and Independent Innovation Foundation (2009JC020) of Shandong University. ’ REFERENCES (1) Arkin, M. R.; Stemp, E. D. A.; Holmlin, R. E.; Barton, J. K.; Hormann, A.; Olson, E. J. C.; Barbara, P. F. Science 1996, 273, 475. (2) Dandliker, P. J.; Holmlin, R. E.; Barton, J. K. Science 1997, 275, 1465. (3) Cai, Z. L.; Li, X. F.; Sevilla, M. D. J. Phys. Chem. B 2002, 106, 2755. (4) Saito, I.; Nakamura, T.; Nakatani, K.; Yoshioka, Y.; Yamaguchi, K.; Sugiyama, H. J. Am. Chem. Soc. 1998, 120, 12686. (5) Giese, B.; Amaudrut, J.; Kohler, A. K.; Spormann, M.; Wessely, S. Nature 2001, 412, 318. (6) Shimazaki, T.; Asai, Y.; Yarnashita, K. J. Phys. Chem. B 2005, 109, 1295. (7) Tsukamoto, T.; Ishikawa, Y.; Natsume, T.; Dedachi, K.; Kurita, N. Chem. Phys. Lett. 2007, 441, 136. (8) Basko, D. M.; Conwell, E. M. Phys. Rev. Lett. 2002, 88, 098102. (9) Gomez-Navarro, C.; Moreno-Herrero, F.; de Pablo, P. J.; Colchero, J.; Gomez-Herrero, J.; Baro, A. M. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 8484. (10) de Pablo, P. J.; Moreno-Herrero, F.; Colchero, J.; GomezHerrero, J.; Herrero, P.; Baro, A. M.; Ordejon, P.; Soler, J. M.; Artacho, E. Phys. Rev. Lett. 2000, 85, 4992. (11) Braun, E.; Eichen, Y.; Sivan, U.; Ben-Yoseph, G. Nature 1998, 391, 775. (12) Alexandre, S. S.; Soler, J. M.; Seijo, L.; Zamora, F. Phys. Rev. B 2006, 73, 5. (13) Fink, H. W.; Schonenberger, C. Nature 1999, 398, 407. (14) Tran, P.; Alavi, B.; Gruner, G. Phys. Rev. Lett. 2000, 85, 1564. (15) Kasumov, A. Y.; Kociak, M.; Gueron, S.; Reulet, B.; Volkov, V. T.; Klinov, D. V.; Bouchiat, H. Science 2001, 291, 280. (16) Henry, A. A.; Romesberg, F. E. Curr. Opin. Chem. Biol. 2003, 7, 727. (17) Han, L.; Li, H. F.; Cukier, R. I.; Bu, Y. X. J. Phys. Chem. B 2009, 113, 4407. (18) Zhang, J. M.; Cukier, R. I.; Bu, Y. X. J. Phys. Chem. B 2007, 111, 8335. (19) Loakes, D. Nucleic Acids Res. 2001, 29, 2437. (20) Krueger, A. T.; Lu, H. G.; Lee, A. H. F.; Kool, E. T. Acc. Chem. Res. 2007, 40, 141. (21) Kool, E. T. Acc. Chem. Res. 2002, 35, 936. (22) Mitsuhiko, S. Macromol. Symp. 2004, 209, 41. (23) Clever, G. H.; Kaul, C.; Carell, T. Angew. Chem., Int. Ed. 2007, 46, 6226. (24) Keren, K.; Krueger, M.; Gilad, R.; Ben-Yoseph, G.; Sivan, U.; Braun, E. Science 2002, 297, 72. (25) Yan, H.; Park, S.; Finkelstein, G.; Reif, J.; LaBean, T. Science 2003, 301, 1882. (26) Mertig, M.; Colombi Ciacchi, L.; Seidel, R.; Pompe, W.; De Vita, A. Nano Lett. 2002, 2, 841. (27) Giese, B. Acc. Chem. Res. 2000, 33, 631. (28) Delaney, S.; Barton, J. K. J. Org. Chem. 2003, 68, 6475. (29) Aich, P.; Labiuk, S. L.; Tari, L. W.; Delbaere, L. J. T.; Roesler, W. J.; Falk, K. J.; Steer, R. P.; Lee, J. S. J. Mol. Biol. 1999, 294, 477. (30) Rakitin, A.; Aich, P.; Papadopoulos, C.; Kobzar, Y.; Vedeneev, A. S.; Lee, J. S.; Xu, J. M. Phys. Rev. Lett. 2001, 86, 3670. (31) Zamora, F.; Sabat, M. Inorg. Chem. 2002, 41, 4976. (32) Navarro, J. A. R.; Lippert, B. Coord. Chem. Rev. 2001, 222, 219. (33) Miyake, Y.; Togashi, H.; Tashiro, M.; Yamaguchi, H.; Oda, S.; Kudo, M.; Tanaka, Y.; Kondo, Y.; Sawa, R.; Fujimoto, T.; Machinami, T.; Ono, A. J. Am. Chem. Soc. 2006, 128, 2172. (34) Meggers, E.; Holland, P. L.; Tolman, W. B.; Romesberg, F. E.; Schultz, P. G. J. Am. Chem. Soc. 2000, 122, 10714. 2863

dx.doi.org/10.1021/jp107605k |J. Phys. Chem. C 2011, 115, 2855–2864

The Journal of Physical Chemistry C (35) Rulisek, L.; Sponer, J. J. Phys. Chem. B 2003, 107, 1913. (36) Di Felice, R.; Calzolari, A.; Zhang, H. Nanotechnology 2004, 15, 1256. (37) Noguera, M.; Bertran, J.; Sodupe, M. J. Phys. Chem. A 2004, 108, 333. (38) Fuentes-Cabrera, M.; Sumpter, B. G.; Sponer, J. E.; Sponer, J.; Petit, L.; Wells, J. C. J. Phys. Chem. B 2007, 111, 870. (39) Noguera, M.; Branchadell, V.; Constantino, E.; Rios-Font, R.; Sodupe, M.; Rodriguez-Santiago, L. J. Phys. Chem. A 2007, 111, 9823. (40) Bagno, A.; Saielli, G. J. Am. Chem. Soc. 2007, 129, 11360. (41) 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.; Bakken, V.; 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; Gaussian, Inc.: Wallingford, CT, 2004. (42) Brancolini, G.; Di Felice, R. J. Phys. Chem. B 2008, 112, 14281. (43) Foresman, J. B.; Head-Gordon, M.; Pople, J. A.; Frisch, M. J. J. Phys. Chem. 1992, 96, 135. (44) It is well-known that B3LYP does not properly treat the dispersion energy; therefore, M06-2X was also utilized for electronic property calculations of three-layer stacks. However, with respect to the frontier orbitals with which we were concerned, it is likely that B3LYP is more appropriate. The detailed discussion is given in part 3 of the Supporting Information. (45) Liu, H. B.; Gao, J. M.; Lynch, S. R.; Saito, Y. D.; Maynard, L.; Kool, E. T. Science 2003, 302, 868. (46) Pavelka, M.; Simanek, M.; Sponer, J.; Burda, J. V. J. Phys. Chem. A 2006, 110, 4795. (47) Lu, X.; Wu, C. M. L.; Wei, S.; Guo, W. J. Phys. Chem. A 2009, 737. (48) Humblot, V.; Bingham, C. J. A.; Le Roux, D.; Marti, E. M.; McNutt, A.; Nunney, T. S.; Lorenzo, M. O.; Roberts, A. J.; Williams, J.; Surman, M.; Raval, R. Surf. Sci. 2003, 537, 253. (49) Batsanov, S. S. Inorg. Mater. 2001, 37, 871. (50) Carvajal, M. A.; Santiago, A.; Juan, J. N. Chem.-Eur. J. 2004, 10, 2117. (51) Cotton, F. A.; Feng, X. J.; Timmons, D. J. Inorg. Chem. 1998, 37, 4066. (52) Clerac, R.; Cotton, F. A.; Daniels, L. M.; Gu, J. D.; Murillo, C. A.; Zhou, H. C. Inorg. Chem. 2000, 39, 4488. (53) Fu, W. F.; Gan, X.; Che, C. M.; Cao, Q. Y.; Zhou, Z. Y.; Zhu, N. N. Y. Chem.-Eur. J. 2004, 10, 2228. (54) Che, C. M.; Mao, Z.; Miskowski, V. M.; Tse, M. C.; Chan, C. K.; Cheung, K. K.; Phillips, D. L.; Leung, K. H. Angew. Chem., Int. Ed. 2000, 39, 4084. (55) Chi-Ming, C.; Zhong, M.; Vincen, M. M.; Man-Chung, T.; ChiKeung, C.; Kung-Kai, C.; David Lee, P.; King-Hung, L. Angew. Chem. 2000, 39, 4084. (56) Pyykk€o, P.; Runeberg, N.; Mendizabal, F. Chem.-Eur. J. 1997, 3, 1451. (57) Pyykk€o, P. Chem. Rev. 1997, 97, 597. (58) Singh, K.; Long, J. R.; Stavropoulos, P. J. Am. Chem. Soc. 1997, 119, 2942. (59) Hermann, H. L.; Boche, G.; Schwerdtfeger, P. Chem.-Eur. J. 2001, 7, 5333. (60) Wang, H. Y.; Li, X. B.; Tang, Y. J.; Chen, X. H.; Wang, C. Y.; Zhu, Z. H. Acta Phys. Sin. 2005, 54, 3565.

ARTICLE

(61) Baik, M. H.; Silverman, J. S.; Yang, I. V.; Ropp, P. A.; Szalai, V. A.; Yang, W. T.; Thorp, H. H. J. Phys. Chem. B 2001, 105, 6437. (62) Okamoto, A.; Tanaka, K.; Saito, I. J. Am. Chem. Soc. 2003, 125, 5066. (63) Dreuw, A.; Head-Gordon, M. Chem. Rev. 2005, 105, 4009. (64) Dreuw, A.; Weisman, J. L.; Head-Gordon, M. J. Chem. Phys. 2003, 119, 2943. (65) Jensen, L.; Govind, N. J. Phys. Chem. A 2009, 113, 9761. (66) Gorelsky, S. I. SWizard Program; CCRI, University Of Ottawa: Ottawa, Canada, 2009; http://www.sg-chem.net/. (67) Holmen, A.; Broo, A.; Norden, B. J. Am. Chem. Soc. 1997, 119, 12240. (68) Broo, A.; Holmen, A. J. Phys. Chem. A 1997, 101, 3589. (69) Hardman, S. J. O.; Thompson, K. C. Biochemistry 2006, 45, 9145. (70) Thompson, K. C.; Miyake, N. J. Phys. Chem. B 2005, 109, 6012. (71) Shukla, M. K.; Mishra, S. K.; Kumar, A.; Mishra, P. C. J. Comput. Chem. 2000, 21, 826. (72) Cotton, F. A.; Liu, C. Y.; Murillo, C. A.; Villagran, D.; Wang, X. J. Am. Chem. Soc. 2003, 125, 13564. (73) Mayr, A.; Yu, M. P. Y.; Yam, V. W. W. J. Am. Chem. Soc. 1999, 121, 1760.

2864

dx.doi.org/10.1021/jp107605k |J. Phys. Chem. C 2011, 115, 2855–2864