1: %\documentclass{article}
2: \documentclass[prb,aps,preprint]{revtex4}
3: \usepackage{graphicx,amsmath}
4:
5: \begin{document}
6:
7: \title{Dithiocarbamate Anchoring in Molecular Wire Junctions: A First Principles Study}
8:
9: \author{Zhenyu Li}
10: \author{Daniel S. Kosov}
11:
12: \address{Department of Chemistry and Biochemistry,
13: University of Maryland, College Park, 20742}
14:
15: %\date{\today}
16:
17: \begin{abstract}
18: Recent experimental realization [J. Am. Chem. Soc., 127 (2005) 7328] of
19: various dithiocarbamate self assembly on gold surface opens the possibility for
20: use of dithiocarbamate linkers to anchor molecular wires to gold electrodes.
21: In this paper, we explore this hypothesis computationally. We computed the
22: electron transport properties of
23: 4,4$^\prime$-bipyridine (BP),
24: 4,4$^\prime$-bipyridinium-1,1$^\prime$-bis(carbodithioate) (BPBC),
25: 4-(4$^\prime$-pyridyl)-peridium-1-carbodithioate (BPC)
26: molecule junctions based on the density functional theory and non-equilibrium
27: Green's functions. We demonstrated that the stronger molecule-electrode coupling
28: associated with the conjugated dithiocarbamate linker broadens transmission resonances near the Fermi energy.
29: The broadening effect along with the extension of
30: the $\pi$ conjugation from the molecule to the gold electrodes lead to enhanced electrical conductance for BPBC molecule.
31: The conductance enhancement factor is as large as 25 at applied voltage bias 1.0 V.
32: Rectification behavior is predicted for BPC molecular wire junction, which has the asymmetric anchoring groups.
33: \end{abstract}
34:
35: \maketitle
36:
37: \section{Introduction}
38: Considerable experimental and computational efforts have been devoted in recent years
39: to the problem of molecular electronics. \cite{Nitzan03,Joachim05}
40: These efforts have been spurred on by the anticipation for
41: breakthrough technological applications.
42: When an electronic devices shrink to molecular level,
43: the molecule-electrode contact becomes a major
44: part of the device and chemical details of the bonding
45: becomes pivotal
46: for the device electron transport properties.\cite{Xue03, ke0497, Basch05, Ke05}
47: To date, the most widely used molecular wire
48: junctions have been formed by thiolated molecules assembled between
49: gold electrodes.
50: There is growing experimental awareness of the fact that one of the
51: most important challenges in molecular-scale electronics is
52: the ill-defined bonding between molecular wires and gold electrodes.
53: The aim of the reproducible experimental measurements has
54: become the search for better molecule-electrode bonding.\cite{Siaj0588, Tulevski0591, Tivanski0598, Guo0656, Venkataraman0600,He06}
55: Most of recent experiments has attempted to circumvent
56: shortcomings of thiol linkage by using carbenes on transition metals \cite{Tulevski0591}
57: and on metal carbides\cite{Siaj0588}, by inserting molecular wires into nanogaps in single-walled carbon nanotubes \cite{Guo0656}, or by fabricating metal-free silicon-molecule-nanotube molecular devices.\cite{He06}
58: But being an excellent conductor with mature manipulation technics, gold remains one of the
59: most attractive electrode material for molecular electronics.
60: %In this paper, we demonstrate
61: %that if an anchoring group is properly chosen, molecule-gold
62: %contact can still be a very promise type of linkage for molecular
63: %devices.
64:
65: It has been recently shown by Zhao \textit{et al.}\cite{Zhao05}
66: that dithiocarbamate formation provides a reliable technology for
67: conjugating secondary amines onto metal surfaces to form strongly
68: absorbed molecular ligands which are stable under various types
69: of the environmental stress. This result suggests a
70: possibility of connection of molecular wires to gold electrodes
71: via dithiocarbamate group instead of thiol group. In fact, the distance between the
72: sulphur atoms in the dithiocarbamate group is nearly ideal for
73: epitaxial adsorption onto Au surface, and a variety of secondary
74: amines can condense with dithiocarbamate linker onto Au surfaces
75: at room temperature,\cite{Zhao05} which makes the dithiocarbamate
76: anchoring structure to be easily fabricated and
77: characterized. On the other hand, the dithiocarbamate anchoring
78: group provides strong molecule-electrode coupling, and thus the
79: possibility of more stable junctions with enhanced conductance.
80:
81: Theoretically, stronger molecule-electrode coupling leads to more
82: significant manifestation of the effects associated with the
83: molecule-electrode interface. The substantial charge transfer
84: may occur between the molecule and the metal electrodes even in the
85: absence of the applied field upon initial chemisorption and it
86: will determine the transport properties of the molecular wire
87: junctions. The stronger is the molecule-electrode coupling, the
88: larger is the mixing between the discrete molecular levels and
89: the continuum of the metal electronic states and the larger is the
90: broadening of the resonances in the electron transmission
91: probability. The dithiocarbamate anchoring group thus provides an
92: ideal testing ground for computational and experimental
93: explorations of the role of such pivotal effects in the transport
94: properties of molecular wires.
95:
96: Dithiocarbamate assembly on the gold surface also opens
97: interesting possibility for creating rectifying molecular wire
98: junctions where electrons flow along one preferential
99: direction.\cite{Metzger03} A number of donor-acceptor type
100: molecular rectifiers has been proposed, \cite{Aviram7477,
101: Metzger03, Morales05, Elbing05, Derosa03} in which
102: the electron density asymmetry (like
103: in semiconductor p-n junctions) causes the current to flow in one
104: direction. Rectification effect could also be achieved by
105: adjusting the molecule-electrode bond length as it was suggested
106: theoretically,\cite{Taylor02} but it is difficult to realize such
107: kind of the rectifying molecular wire junctions in a controllable
108: way. We propose that molecular rectification can be realized in a more
109: practical way by choosing different anchoring group for the left
110: and the right sides.
111:
112: The aim of this paper is to explore computationally the role of
113: dithiocarbamate anchoring group on transport properties of
114: molecular wire junctions. In this paper, we consider three systems
115: as prototype molecular wire junctions: two symmetric molecular
116: wires 4,4$^\prime$-bipyridine (BP) and
117: 4,4$^\prime$-bipyridinium-1,1$^\prime$-bis(carbodithioate) (BPBC),
118: and one asymmetric junction
119: 4-(4$^\prime$-pyridyl)-peridium-1-carbodithioate (BPC). We choose
120: the bypyridine based molecular wire junctions because its transport properties
121: have been thoroughly characterized, both experimentally
122: \cite{Xu03a, Xu03} and theoretically \cite{Tada04,Wu05,Perez05}.
123: With two nitrogen atoms at the two opposite ends, BP
124: is able to connect to gold electrode directly. By repeatedly
125: forming thousands of molecular junctions between gold scanning
126: tunneling microscope tip and a gold substrate in solution, Xu
127: \textit{et al.} \cite{Xu03} found the conductance of BP molecule
128: near zero bias was about 0.01 $G_0$, where $G_0=e^2/\pi
129: \hbar=77.48 \mu S$ is the quantum of conductance. Theoretically,
130: the conductance of directly linked BP molecular wire has been
131: calculated non-self-consistently by both cluster \cite{Tada04} and
132: bulk electrode model.\cite{Perez05} Very recently, a more precise
133: fully self-consistent non-equilibrium Green's functions (NEGF)
134: calculation has also been performed on this system.\cite{Wu05} In
135: this paper, we perform combined density functional theory (DFT)
136: and NEGF calculations on the electron transport properties of
137: BP,BPBC, and BPC. The electric current at
138: finite bias voltage is determined self-consistently in our calculations. We find that
139: dithiocarbamate linkers yield significant broadening of peaks in
140: the transmission spectra and improve significantly electron
141: transport properties of molecular wire junctions. We suggest and
142: confirm computationally that BPC molecular wire junction
143: rectifies (with factor close to 2) the electric current which
144: flows through.
145:
146:
147: \section{Computational Methods}
148:
149: The electronic transport properties are calculated with the
150: NEGF technique using TranSIESTA-C
151: package (version 1.3.0.4). \cite{Brandbyge02} In this package, the molecular wire junction
152: is divided into three regions, left
153: electrode (L), contact region (C), and right electrode (R). The
154: contact region typically includes parts of the physical electrodes where
155: the screening effects take place, to ensure that the charge
156: distributions in L and R region correspond to the bulk phases
157: of the same material. The semi-infinite electrodes
158: (Au(111)-(3$\times$3) surfaces in our case) are calculated
159: separately to obtain the bulk self-energy.
160:
161: %%%%%%%%%%%%%%%%
162: The main loop for self-consistent NEGF/DFT calculations is described below.
163: For detailed description and for the technical
164: details suppressed here we refer to the paper of Brandbyde
165: {\it et al.}.\cite{Brandbyge02}
166: The Green's function is computed by inverting the finite matrix
167: \begin{eqnarray}
168: \mathbf{G}(E,V) =
169: \left[ E \mathbf{S} -
170: \left(
171: \begin{array}{ccc}
172: H_L +\Sigma_L & V_L & 0 \\
173: V_L & H_M & V_R \\
174: 0 & V_R & H_R +\Sigma_R
175: \end{array}
176: \right)
177: \right]^{-1},
178: \end{eqnarray}
179: with $V$ being the applied voltage bias.
180: The matrix product of the Green's function and the imaginary part of the left/right
181: electrodes self-energy yields the spectral densities $\mathbf{\rho}^{L/R}(E)$.
182: The spectral densities of the left and right electrodes are combined
183: together to compute the non-equilibrium, voltage-dependent
184: density matrix
185: \begin{equation}
186: \mathbf{D} = \int dE \left[
187: \mathbf{\rho}^L(E) f(E-\mu_L)
188: + \mathbf{\rho}^R(E) f(E - \mu_R)
189: \right],
190: \end{equation}
191: where $f$ is Fermi-Dirac occupation numbers.
192: The density matrix is converted into non-equilibrium electron density
193: \begin{equation}
194: n(\mathbf{r}) = \sum_{\mu \nu} \phi_{\mu} (\mathbf{r}) D_{\mu \nu}
195: \phi_{\nu}(\mathbf{r}),
196: \end{equation}
197: where $\phi_{\nu}(\mathbf{r})$ atomically localized basis functions.
198: The nonequilibrium electron density enables us to compute matrix elements
199: of the Green's function.
200: The Hartree potential is determined through the solution of the Poisson equation with appropriate voltage dependent boundary conditions.
201: Then this loop of calculations is repeated untill
202: self-consistency is achieved.
203: Once the self-consistent convergence is achieved,
204: the transmission
205: spectrum, which gives the probability for electron with incident energy $E$ to be transfered from the left electrode to the right,
206: is calculated by
207: \begin{equation}
208: \label{eq:trans}
209: T(E,V)=Tr[\mathbf{\Gamma}_L(E,V)\mathbf{G}(E,V)
210: \mathbf{\Gamma}_R(E,V)\mathbf{G}^\dagger(E,V)],
211: \end{equation}
212: where
213: $\mathbf{\Gamma}_{L/R}$ is the coupling matrix.
214: The integration of the transmission spectrum yields the electric current:
215: \begin{equation}
216: \label{eq:iv} I(V)=\int_{\mu_L}^{\mu_R} T(E,V) \left(f(E-\mu_L) -f(E-\mu_R) \right) dE,
217: \end{equation}
218: where
219: $\mu_L =-V/2$ ($\mu_R=V/2$) is the chemical potential of the left(right) electrode.
220:
221: %%%%%%%%%%%%%%%%%%%
222: If we project
223: the self-consistent hamiltonian onto the Hilbert space spanned by
224: the basis functions of the central molecule, we obtain the molecular projected
225: self-consistent hamiltonian (MPSH). The eigenstates of MPSH
226: can be associated with poles of the Green's function and thus roughly correspond to the positions of the peaks in the transmission spectrum (\ref{eq:trans}).
227:
228: The electronic structure is described within the implementation
229: of DFT in SIESTA computer program, \cite{Soler0245} which
230: solves the Kohn-Sham equations with numerical atomic basis sets.
231: Double zeta with polarization (DZP) basis set is chosen for all
232: atoms except Au, for which single zeta with polarization (SZP) is
233: used. Our test calculation indicates that using SZP basis set for
234: Au does not effect the accuracy of our calculations. Core
235: electrons are modeled with Troullier-Martins nonlocal
236: pseudopotentials.\cite{Troullier91} Perdew-Zunger local density
237: approximation (LDA) is used to describe the exchange-correlation
238: potential.\cite{Perdew8148} By comparing the results from two recent
239: calculations with hybrid functional and LDA
240: respectively, \cite{Perez05, Wu05} we conclude that
241: LDA is adequate in our case.
242:
243: \section{Results and Discussion}
244:
245: \subsection{Structural models of molecular wire junctions}
246:
247: Transport systems considered in this paper are formed by BP, BPC
248: or BPBC molecules sandwiched between two Au(111)-(3$\times$3)
249: surfaces. Fig. \ref{fig:geo} shows the structural models for these
250: molecular wire junctions. Only one unit cell for the semi-infinite
251: left or right electrode is plotted, which contains three
252: Au layers. In the contact region, two Au layers at both left and
253: right side are included, of which the most left and right layers
254: are constrained to their theoretical bulk geometry to match the
255: structure of the Au(111) surface.
256: We assume that all three junctions have coplanar geometry. This narrows
257: down the problem of difference between BP,BPBC, and BPC transport properties
258: to the role of the dithiocarbamate linkers.
259: The rest of the contact region
260: is fully optimized. We also optimized the length of the junctions
261: by computing the potential energy surface (PES, i.e. the energy of
262: the system as a function of the distance between the left and the
263: right electrodes). Every point on the energy surface is calculated
264: by performing a geometry optimization with constrained
265: electrode-electrode distance. Then the PES curve is fitted to
266: determine the correct electrode-electrode distance.
267: %%%%%%%%%%%%%%%%
268: We also performed test calculations for non-coplanar BP and BPBC.
269: Twisting of pyridine rings by $10^\circ$, which corresponds to equilibrium geometry for the
270: bipyridine junction,\cite{Wu05} has negligible influence on the transmission spectra.
271: %%%%%%%%%%%%%%
272: \begin{figure}
273: \includegraphics[keepaspectratio,totalheight=12cm]{geo.eps}
274: \caption{The relaxed geometry of Au-BP-Au junctions with different
275: anchoring groups. a) BP,
276: b) BPBC,
277: c) BPC.
278: }
279: \label{fig:geo}
280: \end{figure}
281:
282: Fig. \ref{fig:geo}a shows the optimized geometry of the BP
283: molecular wire, which is directly connected to two gold
284: electrodes. When BP molecules are adsorbed on gold surface they
285: adopt either atop, hollow, or bridge adsorption sites. Depending upon experimental realization BP molecule
286: may also be connected to the gold surface via apex Au atom. Wu
287: \textit{et al.} \cite{Wu05} have studied effects of different
288: interface geometry on the transport properties of BP molecular
289: wire junction, but focused on the hollow site adsorption. Here we
290: study the interface geometry with apex Au atom, which is more
291: likely realized on the typical experiments where
292: gold-molecule-gold junctions were formed by repeated pulling of
293: STM tip from Au surface. \cite{Xu03} This geometrical
294: configuration also gives the largest value of the conductance.\cite{Ke05, Wu05}
295: Therefore our calculations predict the lower limit for the
296: enhancement of the conductance by means of dithiocarbamate
297: anchoring. The structural model for BPBC molecular wire junction
298: is shown on Fig.\ref{fig:geo}(b); here each of the four S is
299: connected to the hollow sites of Au(111)-(3$\times$3) electrode.
300: This structure resembles the experimental scheme \textbf{7} of
301: dithiocarbamate ligands formations on Au surface in Fig. 1 of the
302: paper by Zhao \textit{et al.}.\cite{Zhao05} Fig.\ref{fig:geo}(c)
303: shows asymmetric BPC molecule sandwiched between two Au
304: electrodes, in which the left side of the molecule wire is
305: connected via dithiocarbamate linker and the right side is
306: connected via N-Au bond to the apex Au atom.
307:
308:
309: \subsection{Effect of dithiocarbamate anchoring}
310:
311: To study the effect of dithiocarbamate anchoring, we begin with
312: calculations of the transport properties of BP molecular wire
313: junction. First, we compute the transmission coefficient $T(E)$ (\ref{eq:trans}). The
314: zero bias transmission spectrum of BP junction is shown on
315: Fig.\ref{fig:trans012}. The distinctive feature of this
316: transmission is the existence of the narrow resonance which
317: appears very close to the Fermi level (0.09 eV above $E_F$). This
318: resonance determines the transport properties of BP molecular
319: wire junction under small bias voltages. Two other narrow peaks
320: are at 2.08 eV above and 1.11 eV below the electrode Fermi energy.
321: The transmission coefficient at the Fermi level (i.e. the
322: conductance at small voltage) is 0.195 $G_0$. This computed value
323: of the conductance is by order of magnitude larger than the
324: experimental value of 0.01 $G_0$.\cite{Xu03} But, we notice that the
325: transmission curve from our calculations is consistent with the
326: result of the recent first principles calculations on BP
327: junction,\cite{Wu05,Ke05,Perez05} if the same interface geometry
328: is considered. In fact, experimental conductance value can be
329: obtained theoretically by choosing a suitable model for the
330: interface geometry.\cite{Wu05} Transmission eigenchannel
331: decomposition indicates that the zero voltage conductance is
332: almost entirely determined by a single channel.
333:
334:
335: \begin{figure}
336: \includegraphics[keepaspectratio,totalheight=8cm]{trans.eps}
337: \caption{The transmission spectrum of molecular wire junctions. The
338: average chemical potential of the left and right electrodes is set
339: to zero.}
340: \label{fig:trans012}
341: \end{figure}
342:
343: The transmission spectrum can be further interpreted in terms of MPSH.
344: The MPSH
345: orbitals near the average Fermi level are plotted on Fig.
346: \ref{fig:MPSH}. MPSH orbitals of BP molecular wire junction
347: are very similar to the Kohn-Sham orbitals of an isolated BP
348: molecule except for some additional states which mainly come from
349: apex Au atoms. It indicates that the coupling between BP
350: molecule and apex Au atom is weak, which explains why the
351: resonances in the transmission spectrum are very narrow. Since the
352: resonant peaks in the transmission are narrow and well separated, it
353: is easy to put them into correspondence with the MPSH orbitals by
354: directly comparing the positions of the resonances with the MPSH
355: eigenenergies. The MPSH eigenenergies are given in Table
356: \ref{tbl:energy}. The previously mentioned three peaks correspond
357: to the four MPSH orbitals: they are orbitals 39, 40, 41, and 44. Orbital
358: 39 and 40 are almost degenerated. We can see from Fig.
359: \ref{fig:MPSH}a that all these four orbitals have notable density
360: distribution on N atoms.
361:
362:
363: \begin{figure}
364: \includegraphics[keepaspectratio,totalheight=12cm]{mpsh.eps}
365: \caption{Eigenstates of MPSH for (a) BP, (b) BPBC, and (c)
366: BPC. Anchoring groups and apex Au atom are included for molecular
367: projection. Refer Fig. \ref{fig:geo} for atom types, the atoms
368: are plotted smaller in this figure to present MPSH orbitals better. } \label{fig:MPSH}
369: \end{figure}
370:
371:
372: Since the transport properties of BP molecular wire junction
373: are determined by the narrow resonances in the transmission,
374: experimental measurements could be significantly affected by
375: stochastic switching and could be very sensitive to variations
376: of the electrode-molecule interface. The stochastic switching
377: arises from the structural evolution of the interface
378: geometry. This is undesirable phenomenon, which leads to the
379: changes in the conductance by factors of 2-10 depending upon the
380: scale of the fluctuations. \cite{Ramachandran03, Hu05, Basch05,
381: Donhauser01} The "robust" transmission spectrum should possess the
382: following features: a) very broad resonances to reduce the
383: influence of the current-induced fluctuations of the interface
384: geometry; b) large transmission probability in the vicinity of the
385: electrode Fermi energy to make a good conductor from a molecular
386: wire junction.
387:
388: \begin{table}
389: \caption{Eigenvalues of the molecular projected self-consistent Hamiltonian (MPSH) orbitals.
390: The energy of the average electrode Fermi energy is set to zero. }
391: \label{tbl:energy}
392: \begin{tabular}{cccccccc}
393: \hline\hline
394: \multicolumn{2}{c}{BP} && \multicolumn{2}{c}{BPBC} && \multicolumn{2}{c}{BPC} \\
395: \cline{1-2}\cline{4-5} \cline{7-8}
396: $n$ & Energy(eV) && $n$ & Energy(eV) && $n$ & Energy(eV) \\
397: \hline
398: 35 & -2.421 && 41 & -2.759 && 38 & -2.408 \\
399: 36 & -2.414 && 42 & -2.746 && 39 & -2.172 \\
400: 37 & -2.354 && 43 & -2.542 && 40 & -2.170 \\
401: 38 & -2.346 && 44 & -2.510 && 41 & -2.121 \\
402: 39 & -1.182 && 45 & -0.851 && 42 & -0.999 \\
403: 40 & -1.134 && 46 & 0.246 && 43 & -0.340 \\
404: 41 & 0.044 && 47 & 1.474 && 44 & 1.234 \\
405: 42 & 1.285 && 48 & 1.786 && 45 & 1.890 \\
406: 43 & 1.806 && 49 & 2.276 && 46 & 2.377 \\
407: 44 & 2.050 && 50 & 2.643 && 47 & 2.677 \\
408: \hline\hline
409: \end{tabular}
410: \end{table}
411:
412:
413: Dithiocarbamate linker provides strong coupling between the molecular
414: wire and the gold electrode and it continues $\pi$-conjugation from the
415: molecule into the gold electrode. Therefore it is very interesting
416: to see if this anchoring group leads to the "robust" molecular device. From scheme \textbf{7} of Fig. 1 in paper
417: \cite{Zhao05}, we can see that BPBC system is readily
418: accessible experimentally. The transmission spectrum for BPBC is
419: plotted in Fig. \ref{fig:trans012}. The transmission spectrum is
420: dominated by the several peaks around the average Fermi energy.
421: But these overlapping peaks are very broad comparing to those of
422: BP transmission spectrum and they form a conductance plateau with
423: high transmission probability within a broad range of the incident
424: electron energy. These transmission peaks can be fitted by Lorenz lineshape with
425: the broadening parameters $\sigma$ varying from
426: 0.3 to 0.8 eV, which are significantly larger than that of BP junction
427: (typically 0.04 eV). There are two sharp transmission peaks at 1.81 and
428: 2.29 eV above the average Fermi energy on the top of the plateau.
429: These two sharp peaks are associated with MPSH orbital 48 and 49,
430: and the four broad peaks correspond to the highest occupied MPSH
431: orbital (HOMO, orbital 45), the lowest unoccupied MPSH orbital (LUMO,
432: orbital 46), LUMO+1 (orbital 47) and orbital 50. HOMO and
433: LUMO+1 come from the mixture of atomic orbitals from
434: dithiocarbamate groups and BP molecule. HOMO is the bonding state, while
435: LUMO+1 is the anti-bonding state.
436:
437: According to the two rules of "robust" transmission spectrum, we
438: can clearly see that the significant performance improvement is achieved for
439: BPBC molecular wire junction comparing to BP junction system.
440: This improvement is due to the stronger molecule-electrode coupling
441: which results in the transmission peak broadening. The shift of peak
442: position is the manifestation of the electronic structure of the
443: dithiocarbamate group. In fact, in a
444: model system with the apex Au atom of BP junction substituted by S
445: atom, we also obtained significantly broadened transmission peaks,
446: however the position of these peaks were exactly the same as those
447: of BP junction. This result is consistent with the fact that S-Au bond
448: is stronger than the N-Au bond, and the $p$ orbital of S lays too deep
449: below the electrode Fermi energy to manifest itself to the transmission spectrum. We also
450: notice that the conjugation between the S$_2$C group and the end N
451: atom of BP molecule plays a key role in the conductance
452: improvement. As indicated in Fig. \ref{fig:MPSH}b, it is the $\pi$
453: conjugation of the anchoring group which continues the conjugation
454: from BP molecule directly to the metal.
455:
456:
457: Fig. \ref{fig:iv} shows current-voltage (I-V) curve and
458: conductance-voltage (dI/dV-V) characteristics of BP and BPBC
459: molecular wire junctions. The conductance-voltage characteristics
460: is computed via numerical differentiation of the I-V curve. The
461: current dependence on the applied voltage is very different for
462: BP and BPBC molecules reflecting the significant difference in
463: the transmission spectra. The narrow resonance at the Fermi
464: energy for BP molecular wire junctions leads to the steep
465: increase of the electric current at the small applied voltage
466: ($\le$0.5 V). The transmission for BP does not have any additional
467: resonances in the energy range between -1 eV and 1 eV, therefore
468: once the contribution of the first resonance has been included the
469: current remains almost constant. The conductance-voltage
470: characteristics reflects this behavior of the current-voltage
471: curve. Conductance reaches its maximum at near zero bias voltage
472: and then it sharply declines to very small values ($<$ 2.5
473: $\mu$S). The overlap of the peaks in the transmission spectrum for
474: BPBC molecular wire junction results in the transmission
475: coefficient, which is sufficiently large and does not change
476: significantly with the variation of electron incident energies
477: from -1 eV to 3 eV (relative to the average electrode Fermi
478: energy). It leads to the structureless I-V and dI/dV-V
479: characteristics for BPBC: current
480: increases linearly with the voltage bias and the conductance
481: remains constant within the broad range of bias voltage
482: variations.
483:
484: Although DFT plus NEGF represents the state-of-the-art transport
485: theoretical model, the aim of the direct comparison of experimental and theoretical I-V
486: characteristics remains illusive. The conductance enhancement factor (ration between BPBC and BP conductances)
487: could be less sensitive to particular uncertainties of molecular-electrode interface geometry due the error cancelation.
488: We thus suggest to use the conductance enhancement factor for comparing the results of theoretical calculations
489: and of experimental measurements.
490: Specifically, our calculations shows that the conductance enhancement factor at 1.0 V bias, where the conductance of both junctions are very stable (linear regime), is as large as 25.
491: We notice that this conductance enhancement factor is consistent with the recent experiment,\cite{Tivanski0598}
492: in which conductance of biphenyl with thiolate and dithiocarboxylate anchoring group was measured.
493: The experimentally measured conductance of biphenyl dithiolate is about 40 times smaller than that of bipyridine.\cite{Tivanski0598,Xu03}
494: On the other hand, the conductance of the dithiocarboxylate linked molecular wire is enhanced by factor 1.4 with respect to the conductance of
495: the thiolate anchored wire.\cite{Tivanski0598} If we assume that the conductance of BPBC and 4,4$^{\prime}$-biphenyl bis(dithiocarboxylate)
496: are the same within the order of magnitude, then our predicted enhancement factor qualitatively agrees with the existing experimental data.
497: \begin{figure}
498: \includegraphics[keepaspectratio,totalheight=16cm]{iv.eps}
499: \caption{Current-voltage (solid squares) and conductance-voltage
500: (open squares) curves for (a) BP and (b) BPBC molecular wires}
501: \label{fig:iv}
502: \end{figure}
503:
504: \subsection{Asymmetric anchoring}
505:
506: \begin{figure}
507: \includegraphics[keepaspectratio,totalheight=16cm]{asymm.eps}
508: \caption{
509: (a) The transmission curves of BPC system at different
510: bias voltages. (b) The
511: current-voltage (solid squares) and conductance-voltage (open squares) curves of BPC system.
512: Inset: rectification-voltage curve (abscissa -- voltage;
513: ordinate --rectification coefficient). } \label{fig:asymm}
514: \end{figure}
515:
516: We use BPC molecular wire junction as a prototype system to
517: study rectification of electron transport caused by the
518: asymmetric anchoring groups, in which the left end of the bipyridine
519: molecule is connected
520: to the gold electrode via N-CS$_2$ linker whereas the right is
521: attached to the electrode by N-Au bond. It is known experimentally
522: that the force needed to break the N-Au bond is 0.8 nN, while the
523: corresponding value for the S-Au bond is larger than 1.5 nN.
524: \cite{Xu03a} It indicates that the bonding between N and the gold
525: surface is at least twice as weak as Au-S bond. With two S-Au
526: chemical bonds on the left and with single N-Au bond on the right
527: such hypothetical molecular wire junction would interact
528: approximately 4 times stronger with the left electrode than
529: with the right one.
530:
531: The voltage dependent transmission
532: spectrum is plotted in Fig. \ref{fig:asymm}a. There are two main
533: peaks located in the vicinity of the average electrode Fermi
534: energy. These peaks are broad but their heights are much smaller
535: than those of BPBC and BP molecules. The first peak crosses the
536: average Fermi level energy and the second peak moves toward the
537: Fermi energy as the voltage bias increases. The MPSH orbitals for
538: BPC molecular wire junction are plotted in Fig.
539: \ref{fig:MPSH}c. Table \ref{tbl:energy} gives the corresponding
540: eigenenergies. The electron transport through HOMO (orbital
541: 43 in Fig. \ref{fig:MPSH}c) contributes to the first peak. The
542: other two transmission peaks above the Fermi energy are
543: associated with orbitals 44 and 47. Due to the strong coupling
544: between molecule and left electrode, the transmission
545: peaks are broad. On the other hand, because of the asymmetric
546: electrode-wire coupling, the conjugation of the molecular orbitals
547: does not continue into the right electrode, which leads to the low
548: transmission probability. In fact, the MPSH orbitals are
549: asymmetric even if we focus on the region of BP
550: molecule.
551:
552: The current-voltage characteristic of BPC is plotted
553: in Fig. \ref{fig:asymm}b. As is expected from the transmission
554: spectrum analysis, the I-V curve is asymmetric. Rectification
555: coefficient $R(V)=|I(V)/I(-V)|$ is used to quantify the asymmetry
556: of the molecular wire response on positive and negative applied
557: voltage bias. $R=1$ means that there
558: is no rectification of the electric current. As shown in the inset
559: of Fig. \ref{fig:asymm}b, for BPC junction, $R<1 $ when the
560: $V< 1.15$ V and $R>1$ for the higher voltages. This means that
561: the preferential direction for the current swaps with bias
562: voltage: electrons flow easier from dithiocarbamate linked end
563: toward nitrogen linked end at low applied voltage bias. When the
564: voltage becomes larger than 1.15 V the preferential direction
565: changes to the opposite.
566:
567: The conductance of BPC molecular wire junction is small at large
568: negative voltage bias, it grows up as the voltage
569: increases. It reaches its first maximum ($\sim$ 9 S) at -0.3 V
570: then decreases to its local minimum ($\sim$ 4 S)
571: at 0.6 V. The conductance curve shows five fold increase
572: within the voltage range of 0.5-1.5 V. This behavior of dI/dV-V
573: can be explained from the variation of the transmission spectrum
574: as a function of the bias voltage. We notice from Fig.
575: \ref{fig:asymm}a that the dependence of the transmission peaks
576: upon the applied voltage bias follows the chemical potential of
577: the electrode with stronger linkage (left electrode in our case).
578: As was pointed out by Taylor \textit{et al.}, \cite{Taylor02} this
579: is the main reason for asymmetry in I-V characteristics for
580: symmetric molecules. Electric current is obtained by the
581: integration of the transmission spectrum from the left chemical
582: potential $\mu_L$ to the right chemical potential $\mu_R$
583: (eq.(\ref{eq:iv})). When the bias voltage varies from 0.0 to -0.25 V, the
584: transmission peak enter the integration range (from $\mu_L =$-V/2
585: to $\mu_R$ =V/2). No new transmission peaks contribute to the integral for
586: the negative voltage bias from -0.25 to -1.5 V. That is the reason
587: why that maximum appears just below the zero bias voltage in the
588: dI/dV-V curve. On the other hand, at the positive bias voltage
589: side, at about 1.5 V, two transmission peaks enter the range of
590: the integration and contribute to the integral, which generates
591: the inversion of the rectification direction.
592:
593:
594: \section{Conclusions}
595: Based upon the combination of DFT and NEGF, we have calculated
596: the conductance for BP, BPBC and BPC molecular wire junctions. Our
597: calculations show that dithiocarbamate linking to the Au electrode
598: leads to the strong molecule-electrode coupling, which is able to continue
599: the $\pi$ conjugation from the molecule to the metal. The overlap
600: of the peaks in the transmission spectrum for BPBC junction
601: results in the transmission coefficient, which is sufficiently
602: large and does not change significantly with the variation of
603: electron incident energies from -1 eV to 3 eV (relative to the
604: average electrode Fermi energy). The electric current increases
605: linearly with the voltage bias and the conductance remains
606: constant within the broad range of bias voltage variations for
607: BPBC molecular wire junction. Our calculations thus demonstrate
608: that the conductance of BP molecular wire can be significantly
609: improved by using dithiocarbamate anchoring group. A conductance
610: enhancement by factor of 25 in the linear regime is predicted, which is qualitatively
611: consistent with the experimental data.
612: We propose to use
613: dithiocarbamate anchoring group to make asymmetric molecular wire
614: junction. We studied the rectification of the electron transport
615: caused by the asymmetric anchoring groups on the example of
616: BPC wire. We show that BPC molecular wire junction rectifies
617: the electric current by factor close to 2. The preferential direction for the current
618: flow undergoes the inversion at critical voltage 1.15 V.
619:
620: \section*{ACKNOWLEDGMENTS}
621: The authors are grateful to M. Gelin, L. Sita, A. Vedernikov and
622: J. Yang for helpful discussions. Part of the calculations were
623: performed on computational facilities at the Laboratory of Bond
624: Selective Chemistry, USTC. This work was partially supported by
625: the Mitsubishi Chemical Corporation.
626:
627:
628: \begin{thebibliography}{99}
629: \bibitem{Nitzan03}Nitzan, A.; Ratner, M. A. \textit{Science} \textbf{2003}, 300, 1384.
630: \bibitem{Joachim05} Joachim, C.; Ratner, M. A.
631: \textit{Proc. Natl. Acad. Sci. U.S.A.} \textbf{2005}, 102, 8801.
632: \bibitem{Xue03}Xue, Y.; Ratner, M. A. \textit{Phys. Rev. B} \textbf{2003}, 68, 115406.
633: \bibitem{ke0497}Ke, S.-H.; Baranger, H. U.; Yang, W. \textit{J. Am. Chem. Soc.} \textbf{2004}, 126, 15897.
634: \bibitem{Basch05}Basch, H.; Cohen, R.; Ratner, M. A. \textit{Nano Lett.} \textbf{2005}, 5, 1668-1675.
635: \bibitem{Ke05}Ke, S.-H.; Baranger, H. U.; Yang, W. \textit{J. Chem. Phys.} \textbf{2005}, 123, 114701.
636: \bibitem{Tulevski0591}Tulevski, G. S.; Myers, M. B.; Hybertsen, M. S.; Steigerwald, M. L., Nuckolls, C. \textit{Science} \textbf{2005}, 309, 591.
637: \bibitem{Siaj0588}Siaj, M.; McBreen, P. H. \textit{Science} \textbf{2005}, 309, 588.
638: \bibitem{Tivanski0598}Tivanski, A. V.; He, Y.; Borguet, E.; Liu, H.; Walker, G. C.; Waldeck, D. H. \textit{J. Phys. Chem. B} \textbf{2005}, 109, 5398.
639: \bibitem{Guo0656}Guo, X.; Small, J. P.; Klare, J. E.; Wang, Y.; Purewal, M. S.; Tam, I. W.; Hong, B. H.; Caldwell, R.; Huang, L.; O'Brien, S.; Yan, J.; Breslow, R.; Wind, S. J.; Hone, J.; Kim, P.; Nuckolls, C. \textit{Science} \textbf{2006}, 311, 356.
640: \bibitem{Venkataraman0600}Venkataraman, L.; Klare, J. E.; Tam, I. W.; Nuckolls, C.; Hybertsen, M. S.; Steigerwald, M. L. \textit{Nano Lett.} \textbf{2006}, ASAP Article.
641: \bibitem{He06} He, J.; Chen, B.; Flatt, A. K.; Stephenson, J.J.;
642: Doyle, C.D.; Tour, J.M. \textit{Nature Materials} \textbf{2006}, 5, 63.
643: \bibitem{Zhao05}Zhao, Y.; Perez-Segarra, W.; Shi, Q.; Wei, A. \textit{J. Am. Chem. Soc.} \textbf{2005}, 127, 7328.
644: \bibitem{Metzger03}Metzger, R. M. \textit{Chem. Rev.} \textbf{2003}, 103, 3803.
645: \bibitem{Aviram7477}Aviram, A; Ratner, M. A. \textit{Chem. Phys. Lett.} \textbf{1974}, 29, 277.
646: \bibitem{Morales05}Morales, G.; Jiang, P.; Yuan, S.; Lee, Y.; Scanchez, A.; You, W.; Yu, L. \textit{J. Am. Chem. Soc.} \textbf{2005}, 127, 10456.
647: \bibitem{Elbing05}Elbing, M.; Ochs, R.; Koentopp, M.; Fischer, M.; von Hanisch, G.; Weigend, F.; Evers, F.; Weber, H. B.; Mayor, M. \textit{Proc. Nat. Acad. Sci.} \textbf{2005}, 102, 8815.
648: \bibitem{Derosa03}Derosa, P. A.; Guda, S.; Seminario J. M. \textit{J. Am. Chem. Soc.} \textbf{2003}, 125, 14240.
649: \bibitem{Taylor02}Taylor, J.; Brandbyge, M.; Stokbro, K. \textit{Phys. Rev. Lett.} \textbf{2002}, 89, 138301.
650: \bibitem{Xu03a}Xu, B.; Xiao, X.; Tao, N. J. \textit{J. Am. Chem. Soc.} \textbf{2003}, 125, 16164.
651: \bibitem{Xu03}Xu, B.; Tao, N. J. \textit{Science} \textbf{2003}, 301, 1221.
652: \bibitem{Tada04}Tada, T.; Kondo, M.; Yoshizawa, K. \textit{J. Chem. Phys.} \textbf{2004}, 121, 8050.
653: \bibitem{Perez05}Perez-Jimenez, A. J. \textit{J. Phys. Chem. B} \textbf{2005}, 109, 10052.
654: \bibitem{Wu05}Wu, X.; Li, Q.; Huang, J.; Yang, J. \textit{J. Chem. Phys.} \textbf{2005}, 123, 184712.
655: \bibitem{Brandbyge02}Brandbyge, M.; Mozos, J.-L.; Ordejon, P.; Taylor, J.; Stokbro, K. \textit{Phys. Rev. B} \textbf{2002}, 65, 165401.
656: \bibitem{Soler0245}Soler, J. M.; Artacho, E., Gale, J.; Garcia, A., Junquera, J.; Ordejon, P.; Sanchez-Portal, D. \textit{J. Phys.: Condens. Matter} \textbf{2002}, 14, 2745.
657: \bibitem{Troullier91}Troullier, N; Martins, J. L. \textit{Phys. Rev. B} \textbf{2001}, 43, 1993.
658: \bibitem{Perdew8148}Perdew, J. P.; Zunger, A. \textit{Phys. Rev. B} \textbf{1981}, 23, 5048.
659: \bibitem{Ramachandran03}Ramachandran, G. K.; Hopson, T. J.; Rawlett, A. M.; Nagahara, L. A.; Primak, A.; Lindsay, S. M. \textit{Science} \textbf{2003}, 300, 1413.
660: \bibitem{Hu05}Hu, Y.; Zhu, Y.; Gao, H. \textit{Phys. Rev. Lett.} \textbf{2005}, 95, 156803.
661: \bibitem{Donhauser01}Donhauser, Z.; Mantooth, B.; Kelly, K.; Bumm, L.; Monnell, J.; Stapleton, J.; Price, D.; Rawlett, A.; Allara, D.; Tour, J. \textit{Science} \textbf{2001}, 292, 2303.
662:
663: \end{thebibliography}
664:
665: \end{document}
666: