1: \documentclass[twocolumn,prl,showpacs]{revtex4}
2: %\documentclass[preprint,prl,showpacs]{revtex4}
3: \usepackage{psfrag}
4: \usepackage{graphicx}
5: \usepackage{amsmath}
6: \usepackage{amssymb}
7:
8: \begin{document}
9: \title{Forces and conductances in a single-molecule bipyridine junction}
10: \author{R. Stadler}
11: \affiliation{Center for Atomic-scale Materials Physics, \\
12: Department of Physics, Technical University of Denmark, DK - 2800 Kgs. Lyngby, Denmark}
13: \author{K. S. Thygesen}
14: \affiliation{Center for Atomic-scale Materials Physics, \\
15: Department of Physics, Technical University of Denmark, DK - 2800 Kgs. Lyngby, Denmark}
16: \author{K. W. Jacobsen}
17: \affiliation{Center for Atomic-scale Materials Physics, \\
18: Department of Physics, Technical University of Denmark, DK - 2800 Kgs. Lyngby, Denmark}
19:
20: \date{\today}
21:
22: \begin{abstract}
23: Inspired by recent measurements of forces and conductances of
24: bipyridine nano-junctions, we have performed density functional
25: theory calculations of structure and electron transport in a
26: bipyridine molecule attached between gold electrodes for seven
27: different contact geometries. The calculations show that both the
28: bonding force and the conductance are sensitive to the surface
29: structure, and that both properties are in good agreement with
30: experiment for contact geometries characterized by intermediate coordination of the metal atoms
31: corresponding to a stepped surface. The conductance is mediated by the
32: lowest unoccupied molecular orbital, which can be illustrated by a
33: quantitative comparison with a one-level model. Implications for
34: the interpretation of the experimentally determined force and
35: conductance distributions are discussed.
36: \end{abstract}
37: \pacs{73.63.Rt, 73.20.Hb, 73.40.Gk}
38: \maketitle
39:
40: %Intro start
41: With the advance of new experimental techniques for the fabrication of
42: atomic-scale contacts, it has become possible in recent years to study
43: electron transport through a few or even a single molecule suspended
44: between metal electrodes.~\cite{joachim95,reed97,reichert02,smit02}
45: The atomistic details of the local contact geometry are crucial in
46: such experiments, as is evident from the low degree of reproducibility
47: of for example the measured I-V
48: curves~\cite{reichert02} or conductance traces~\cite{smit02}. Since the
49: exact contact geometry is difficult to control and characterize,
50: theoretical calculations are needed for interpreting and understanding
51: the results of the experiments.
52:
53: Recently, the conductance of bipyridine molecules in a toluene
54: solution trapped between two gold electrodes has been measured
55: simultaneously to the force required to break the contact.~\cite{xu}
56: The statistical data derived from
57: $\sim$1000 independent measurements show a quantization of both the
58: conductance and the rapture force, indicating that the contacts
59: consist of single molecules. Such bipyridine junctions are of
60: interest for two reasons: i) it is one of the few systems where fairly
61: reliable measurements exist of several properties (forces and conductances) for junctions of single aromatic
62: molecules, and ii) pyridine contacts represent an alternative to the
63: more commonly used thiol anchor groups~\cite{bilic}. Conductances for
64: a bipyridine molecule between defect-free Au (111) surfaces have
65: recently been calculated by Hou et al.~\cite{hou} for a single
66: structure using a cluster
67: model for the description of the electrodes. However, in the
68: experiment the surface structures are not known, and here we take up
69: the issues of stability and the extent to which the force and conductance
70: distributions can be related to local structure.
71:
72: We have performed plane wave-based density functional theory
73: (DFT)~\cite{dacapo} calculations for the conductance and breaking
74: force of a bipyridine-contact. The contact between the surface and
75: the molecule is modeled by a variety of different structures based on
76: the (111) orientation of the Au fcc-lattice. By varying the
77: substrate/adsorbent distance rigidly we obtain total energy curves,
78: from which rapture forces can be derived for the different contact
79: geometries. A recently developed scheme for coherent electron
80: transport~\cite{transport} allows us to calculate the conductance of
81: the molecule within the same theoretical framework as employed for the
82: total energy calculations.
83: Based on our data obtained for the different contact geometries we find that the force and
84: conductance distributions observed in the experiment can only be understood in terms
85: of a high structure-selectivity for the contact geometry.
86: %Intro end
87:
88: \begin{figure}
89: \begin{center}
90: \includegraphics[width=1.0\linewidth,bb=0 0 784 502, clip]{fig1.eps}
91: \end{center}
92: \caption[system]{\label{fig1}A bipyridine molecule suspended between
93: Au electrodes. The insets illustrate the variation of the surface
94: contact geometry, where clockwise from top-right to bottom-left a
95: single ad-layer with one to seven Au ad-atoms (white
96: spheres), where $N_{c}$ varies from three to nine (see text), on top of a flat Au (111) surface are shown. Smaller dark
97: spheres define the nitrogen positions.}
98: \end{figure}
99: %Method1 start
100: Fig.~\ref{fig1} shows an example of the Au/bipyridine/Au
101: nano-junctions and defines the different electrode geometries
102: investigated in this article. By using our DFT-plane wave
103: approach~\cite{dacapo} with the PW91 functional for exchange and
104: correlation~\cite{pw91} and an energy cutoff of 340 eV for the plane
105: wave expansion, we have established that bipyridine prefers to bond to
106: the Au (111) surface in an on-top position, and therefore all
107: calculations have been performed for an on-top contact. The supercells
108: consist of repeated slabs with seven Au layers in a (111) stacking
109: with 3x3 periodicity within the surface plane, where the slabs are
110: linked by one bipyridine molecule per unit cell. As depicted in
111: Fig.~\ref{fig1} the number of Au atoms in the surface plane has been
112: varied from one to seven in such a way that the coordination number $N_{c}$ of the Au
113: atom directly connected to the molecule varies from three to
114: nine. The two structures, where the remaining vacancies in the surface plane are also filled for $N_{c}$=9, are not shown in Fig.~\ref{fig1}, since our calculations showed that occupying these atomic positions has no effect on the forces or conductances. The geometry of the isolated molecule has been
115: relaxed by total energy minimization, but the Au atoms have been fixed
116: to their positions in the bulk crystal structure throughout our study.
117: The torsion angle, $\alpha$, which arises if the two pyridine segments
118: within the molecule are rotated relative to each other has been
119: optimized and found to be around 25$^{\circ}$ when the molecules are adsorbed in the junction. However, since the energy gain of this
120: rotation is only around 0.03 eV, we cannot predict the effect that a
121: solvent or finite temperatures might have on it.
122: We found that the influence of this rotation on the calculated
123: forces is negligible. For the conductance we have performed
124: calculations for both $\alpha$=0$^{\circ}$ and 25$^{\circ}$.
125: An optimization of the Au-N bond length gave values between 2.12 and
126: 2.42~\AA\ depending on the number of atoms in the surface plane but showed no dependency on $\alpha$.
127:
128: In the top panel of Fig.~\ref{fig2} the binding energies at the
129: optimal Au-N bond lengths for all the structures are shown. The
130: binding energies vary significantly with an almost linear dependence
131: on $N_{c}$, and their magnitude indicate intermediate-strength bonding
132: stronger than typical van der Waals bonding but weaker than covalent
133: bonding between open shell systems. Depending on the detailed choice
134: of exchange-correlation functional the binding energies may shift a
135: few tenths of an eV but the trend with metal coordination number is
136: unchanged (see Fig.~\ref{fig2}). The same linear dependence of the
137: binding energy on $N_{c}$ has been found for the chemisorption of
138: oxygen and CO on Au chains and surfaces \cite{oxygen}, indicating that
139: the nature of the bonding is also covalent in the present case.
140: A closer
141: inspection of the surface states shows, that the trend in the binding
142: energies correlates with an upward shift of the $d$-states towards $E_F$ which is expected to enhance the energy gain of a
143: hybridization with the highest occupied molecular orbital (HOMO, see
144: inset) located at $\sim$ -4.6 eV.\cite{catalysis} These findings are in contrast with a recent
145: article, in which the adsorption of pyridine on Au (111) surfaces was
146: interpreted in terms of Van der Waals interactions~\cite{bilic} based
147: on more qualitative arguments such as the amount of charge buildup in
148: the center of the bonding region.
149:
150: From the total energy curves obtained by varying the Au-N bond lengths, the
151: breaking forces can be calculated by taking the derivatives at the
152: inflexion point. As can be seen in Fig.~\ref{fig2} the
153: forces like the energies depend almost linearly on $N_c$. The rather large
154: variation of the calculated breaking
155: forces should be compared with the experimental situation where a
156: quite sharp peak at a value of $0.8\pm0.2$ nN is found.~\cite{xu} This
157: distribution is consistent with our results if the
158: experimental situation results in the preference for a subset of
159: structures in the range $N_c$=4-7. Such a preference could be either thermodynamic or kinetic
160: in origin. Energetically bipyridine clearly favors to bond to metal
161: sites with low metal-coordination numbers according to the
162: calculations. On the other hand such metal defects may be
163: energetically costly to produce and therefore limited in number. Also
164: the experimental process in which the nanocontact is created may bring
165: the system out of local equilibrium and the resulting structures will
166: then depend on the evolution of the process. Finallly it should be
167: noted that the calculations do not take potential effects of the
168: solvent into account.
169:
170: \begin{figure}
171: \begin{center}
172: \includegraphics[width=1.0\linewidth,bb=0 0 550 606,clip]{fig2.eps}
173: \end{center}
174: \caption{\label{fig2}Top: Binding energies for a
175: single N-Au bond as a function of $N_{c}$. The inset shows the HOMO.
176: The solid (dashed) line marks calculations using the PW91
177: [\protect\onlinecite{pw91}] (RPBE [\protect\onlinecite{RPBE}])
178: XC-functional. The position of the {\it d}-band center with respect to the Fermi level is shown
179: as a dotted line. Bottom: Rapture force as a function of $N_{c}$. The horizontal dotted
180: line indicates the experimental value and the grey shaded area marks the region defined by the experimental error bars.}
181: \end{figure}
182:
183: % Forces end
184:
185: %Method2 start
186: The conductance of the bipyridine junctions is calculated using a
187: numerical method for phase-coherent electron transport which combines
188: a Green's function formalism with a plane wave-based DFT description
189: of the atomic and electronic structure of the entire system. The
190: Green's function of the central region, comprising the molecule and
191: part of the gold electrodes, is evaluated in terms of a basis
192: consisting of maximally localized Wannier functions which are defined
193: by an appropriate transformation of the Kohn-Sham eigenstates.
194: Despite their optimal localization, the Wannier functions still have a
195: fairly long-ranged but small tail, which for the systems studied in this article
196: we truncate at a distance of 19.4~\AA\ from the center of each function.
197: The couplings to the semi-infinite gold electrodes are included via
198: self-energy matrices which are also represented in the Wannier
199: function basis. A detailed discussion of the transport scheme can be
200: found in Ref.~\onlinecite{transport}. The central regions used in the
201: transport calculation are the same as the supercells applied in the
202: total energy calculations. These supercells contain $3\times 3$ atoms
203: in the directions perpendicular to the transport direction, and we
204: have found that $\bold k$-point sampling in the transverse plane is
205: important in order to obtain well converged results for the
206: conductance. Thus all calculations presented in the following have
207: been performed using a $4\times 4$ $\bold k$-point grid in the
208: transverse Brillouin zone. For a systematic study of the role of $\bold k$-point sampling for transport calculations see Ref.~\onlinecite{kpoints}.
209:
210: \begin{figure}
211: \begin{center}
212: \includegraphics[width=1.0\linewidth,bb=0 0 550 706,clip]{fig3.eps}
213: \end{center}
214: \caption[channel]{\label{fig3}Top: Conductance in dependence on
215: $N_{c}$ for torsion angles $\alpha$=0$^{\circ}$ (black line) and
216: 25$^{\circ}$ (grey line) in units of the conductance quantum $G_0=2e^2/h$. The dashed line refers to a simple model explained in the
217: text. The horizontal dotted line shows the experimental value.
218: Bottom: Transmission functions for all nano-junctions
219: ($\alpha$=0$^{\circ}$). The peaks are due to the LUMO, shown in the inset,
220: which rises continuously in energy as $N_{c}$ increases.}
221: \end{figure}
222:
223: %Conductance start
224: In Fig.~\ref{fig3} we have plotted the calculated transmission near
225: the Fermi level (which is taken as the zero point) for all geometries
226: outlined in Fig.~\ref{fig1}. The top panel of Fig.~\ref{fig3} shows
227: the conductances for nano-junctions with $N_{c}$=3 to 9 for
228: $\alpha$=0$^{\circ}$ and $\alpha$=25$^{\circ}$ as black and grey solid
229: lines, respectively.
230: The conductance is seen to vary significantly with coordination
231: number. However, the conductance
232: for $\alpha$=25$^{\circ}$ is in good agreement with the experimentally observed value of
233: one percent of the conductance quantum $G_0=2e^2/h$.\cite{xu} It can be noted that this
234: is the same range of $N_c\sim$5, where also the forces in Fig.~\ref{fig2} match the measurements.
235: But as already
236: discussed in connection with the force distributions the actual
237: propensity of the different structures in the experiment is not known.
238:
239: We now proceed to analyze why the conductance varies the way it does
240: with coordination number. The analysis is based on the calculations with
241: $\alpha=0^{\circ}$.
242:
243: Within our calculational scheme it is possible to define molecular
244: orbitals even when the molecule is coupled to the metal. The molecular
245: orbitals are constructed from Wannier functions located on the
246: molecule \cite{transport}. It is therefore also possible to study the
247: transport through a single or a subset of the molecular orbitals by simply removing the remaining orbitals from the basis set.
248: When we focus on the contributions coming from individual
249: molecular orbitals, we find that the single peak in the transmission
250: function around $E_F$ (bottom panel in Fig.~\ref{fig3}) is exclusively
251: due to the LUMO, whose wavefunction is shown in the inset. Since the
252: peak is quite narrow for all geometries, correlation effects, leading
253: to Coulomb blockade behaviour in other weakly bonded molecular
254: junctions~\cite{kubatkin}, could be relevant for its accurate
255: description. However, even in the "coherent limit" there is good agreement between
256: our transport calculations and the experimental results.
257:
258: One way of analysing the shape of the $T(N_{c})$ curve further is by
259: applying a simple one-level model~\cite{newns} with only one molecular
260: state (the LUMO) coupled symmetrically to the leads . Within this model the
261: transmission through a level at energy $E_0$ can be expressed as
262: $T(E)=\Delta^2/((E_0-E_F)^2+\Delta^2)$, where the peak-width
263: is defined by $\Delta=\pi V^2 \rho^0_g$. $V$ is the coupling
264: constant of the LUMO to its group-orbital,
265: \cite{transport,group} and $\rho^0_g$ is the density of states
266: of group orbital in the absence of coupling to the LUMO evaluated at $E_0$. By taking the values
267: for $\rho^0_g, E_0$, and $V$ from the DFT calculation and using them to calculate $\Delta$, we
268: obtain the dashed line in Fig.~\ref{fig3}, which thus represents the
269: conductance when the electrons are allowed to travel between the leads via the LUMO only.
270: The good agreement of this one-level model when
271: compared with our data further illustrates that the variation of the conductance is entirely controlled by the position of
272: the transmission peak due to the LUMO and that the effect of the interference with other orbitals as well as through-vacuum tunneling are negligible.
273:
274: But what is
275: governing the LUMO position?
276: We investigated the variation of the energetic position of the lowest-lying
277: molecular orbital (denoted MO1) with respect to the metal's Fermi level, in dependence on
278: the distance between surface and molecule for all studied surface structures.
279: In this analysis it can be seen that at the bonding distance
280: MO1 and the LUMO depend on $N_{c}$ in quantitatively the
281: same way.
282: This pattern cannot be explained in terms of vacuum level alignment alone, since the variation
283: of the metal's workfunction with $N_{c}$ is parabolic with a minimum at $N_{c}$=5-6.
284: However, MO1 is $\sim$ 10 eV below the lowest-lying Au valence states,
285: and therefore the exchange or bonding interaction between the molecule and the surface do not
286: have a direct influence on the energy of this orbital.
287: From that we can conclude that the dependence of MO1 as well
288: as the LUMO's energy on $N_{c}$ must be exclusively guided by rigid potential shifts, where
289: we found that equilibrium charge transfer dominates at the optimal bonding distance and also Coulomb repulsion due
290: to the confinement of electrons at the interface plays a role.
291: We will discuss these electrostatic effects and their structure-dependent influence on the aligment of molecular levels in detail in a separate publication.
292:
293:
294: %Conductance end
295:
296: % Summary start
297: In summary, we presented a theoretical analysis of the forces and
298: conducances of a bipyridine molecular junction. Since important
299: parameters of the experiments, such as the atomic configuration of the
300: electrode's surfaces are not known, we scanned a wide range of surface
301: structures and found that both the forces and conductances depend
302: crucially on their geometries, but our calculations agreed with the experiments in orders of
303: magnitude for both quantities. The narrow peaks
304: found in the statistics of the experiments, however, could not be directly
305: reproduced from our calculations, and the statistics of the measurements
306: site-selectivity of the adsorption of the molecules has to play a major role.
307: Our study suggests that varying surface structures might be an important ingredient for a
308: systematic understanding of the bonding and conductance properties of
309: molecular junctions.
310: % Summary end
311:
312: The Center for Atomic-scale Materials Physics is sponsored by the
313: Danish National Research Foundation. We acknowledge support from the
314: Nano-Science Center at the University of Copenhagen and from the
315: Danish Center for Scientific Computing through Grant No. HDW-1101-05.
316:
317: %%%%%%% References
318: \begin{thebibliography}{18}
319:
320: \bibitem{joachim95}C.~Joachim, J.~K.~Gimzewski, R.~R.~Schlittler and
321: C.~Chavy, Phys. Rev. Lett. {\bf 74}, 2102 (1995).
322:
323: \bibitem{reed97}M.~A. Reed,
324: C.~Zhou, C.~J. Muller, T.~P. Burgin and J.~M. Tour, Science {\bf
325: 278}, 252 (1997).
326:
327: \bibitem{reichert02}J.~Reichert, R. Ochs, D. Beckmann, H.~B. Weber,
328: M. Mayor and H.~v. L{\"ohneysen}, Phys. Rev. Lett. {\bf 88}, 176804
329: (2002).
330:
331: \bibitem{smit02}R.~H.~M.~Smit, Y.~Noat, C.~Untiedt, N.~D.~Lang, M.~C.~van
332: Hemert, and J.~M.~van Ruitenbeek, Nature {\bf 419}, 906 (2002).
333:
334: \bibitem{xu} B.~Xu and N.~J.~Tao, Science {\bf 301}, 1221 (2003);
335: B.~Xu, X.~Xiao and N.~J.~Tao, J. Am. Chem. Soc. {\bf 125}, 16164
336: (2003).
337:
338: \bibitem{bilic} A.~Bili\'{c}, J.~R.~Reimers and N.~S.~Hush, J. Phys.
339: Chem. B {\bf 106}, 6740 (2002).
340:
341: \bibitem{hou}S.~Hou, J.~Zhang, R.~Li, J.~Ning, R.~Han, Z.~Shen,
342: X.~Zhao, Z.~Xue and Q.~Wu, Nanotechnology {\bf 16}, 239 (2005).
343:
344: \bibitem{dacapo} B. Hammer, L.B. Hansen, and J.K. N{\o}rskov,
345: Phys. Rev. B {\bf 59}, 7413 (1999);
346: S.R. Bahn and K.W. Jacobsen, Comp. Sci. Eng. {\bf 4}, 56 (2002);
347: The Dacapo code can be downloaded at http://www.camp.dtu.dk/campos.
348:
349: \bibitem{transport} K.~S.~Thygesen and K~ W.~Jacobsen, Chem. Phys., in
350: press, arXiv:cond-mat/0501238.
351:
352: \bibitem{pw91} J.~P.~Perdew, J.~A.~Chevary, S.~H.~Vosko,
353: K.~A.~Jackson, M.~R.~Pederson, D.~J.~Singh, and C.~Fiolhais, Phys.
354: Rev. B {\bf 46}, 6671 (1992).
355:
356: \bibitem{RPBE} B.~Hammer, L.~B.~Hansen and J.~K.~N{\o}rskov, Phys.
357: Rev. B {\bf 59}, 7413 (1999).
358:
359: \bibitem{oxygen} M.~Mavrikakis, P.~Stoltze and J.~K.~N{\o}rskov,
360: Catal. Lett. {\bf 64}, 101 (2000); B.~Hammer and J.~K.~N{\o}rskov,
361: Adv. Catal. {\bf 45}, 71 (2000); S.~R.~Bahn, N.~Lopex,
362: J.~K.~N{\o}rskov and K.~W.~Jacobsen, Phys. Rev. B {\bf 66}, 081405 (R)
363: (2002); Y.~Xu and M.~Mavrikakis, J. Phys. Chem. B {\bf 107}, 9298
364: (2003).
365:
366: \bibitem{catalysis}B.~Hammer and J.~K.~N{\o}rskov, Surf. Sci. {\bf
367: 343}, 211 (1995).
368:
369: \bibitem{newns} D.~M.~Newns, Phys. Rev. {\bf 178}, 1123 (1969).
370:
371: \bibitem{group} M.~J.~Kelly, Surf. Sci. {\bf 43}, 587 (1974).
372:
373: \bibitem{kubatkin} S.~Kubatkin, A.~Danilov, M.~Hjort, J.~Cornil,
374: J.~L.~Br\'{e}das, N.~Stuhr-Hansen, P. Hedeg\aa rd and
375: T.~Bj{\o}rnholm, Nature {\bf 425}, 698 (2003).
376:
377: \bibitem{kpoints} K.~S.~Thygesen and K~ W.~Jacobsen, Phys. Rev. B {\bf 72}, 033401 (2005).
378:
379:
380: \end{thebibliography}
381:
382:
383: \end{document}
384: