1: %%http://militera.lib.ru/memo/russian/rokossovsky/index.html
2:
3: %
4: \documentclass[prl,aps,twocolumn,showpacs]{revtex4}
5: % \documentclass[prl,aps,preprint]{revtex4}
6: \usepackage[dvips]{graphicx}
7: \usepackage{dcolumn}
8: \newcommand{\be}{\begin{equation}}
9: \newcommand{\ee}{\end{equation}}
10: \newcommand{\bea}{\begin{eqnarray}} \newcommand{\ba}{\begin{array}}
11: \newcommand{\eea}{\end{eqnarray}} \newcommand{\ea}{\end{array}}
12: \newcommand{\nn}{ \nonumber}
13: \newcommand{\ds}{\displaystyle}
14: \topmargin=-25mm
15: \mathsurround=2pt
16: \begin{document}
17: % \large
18: % \small
19:
20: \title{Negative differential resistance in molecular junctions: The effect
21: of the electrodes electronic structure}
22:
23: \author{ Natalya A. Zimbovskaya$^{1,2,3}$ and Mark R. Pederson$^3$}
24:
25: \affiliation{$^1$Department of Physics and
26: Electronics, University of Puerto Rico-Humacao, CUH Station, Humacao, PR 00791,}
27: \affiliation{$^2$Institute for Functional Nanomaterials, University of Puerto
28: Rico, San Juan, PR 00931,}
29: \affiliation{$^3$code 6390, Naval Research Laboratory, 4555 Overlook Ave SW,
30: Washington, DC 20375}
31:
32:
33: \begin{abstract}
34: We have carried out calculations of electron transport through a
35: metal-molecule-metal junction with metal nanoclusters taking the part of
36: electrodes. We show that negative differential resistance peaks
37: could appear in the current-voltage curves. The peaks arise due to narrow features
38: in the electron density of states of the metal clusters. The proposed analysis
39: is based on the ab initio computations of the relevant wave functions and
40: energies within the framework of the density functional theory using NRLMOL
41: software package.
42: \end{abstract}
43:
44: \pacs{73.63.Rt, 73.23.Ad, 31.15.A-}
45: \maketitle
46: \date{\today}
47:
48:
49: Electron transport through molecular-scale systems has been intensively studied
50: in the past two decades \cite{1}. Largely,
51: the unceasing efforts of the research community to further advance these studies
52: are due to important application potentials of single molecules as active
53: elements in various nanodevices intended to complement current silicon based
54: electronics \cite{2}.
55: Among various important properties of the electron transport
56: through metal-molecule junctions one may separate out the negative
57: differential resistance (NDR), that is the decrease of the current $I$ while
58: the bias voltage $ V$ across the molecule increases. The NDR efffect
59: was originally observed in tunneling semiconducting diodes \cite{3,4}.
60: Later, the NDR was viewed in quantum dots\cite{5,6} and metal-molecule-metal
61: junctions (see e.g. Refs. \cite{6,7,8,9,10,11,12,13,14}).
62:
63: Several possible scenarios are proposed to explain the NDR occurence in
64: the electron transport through molecules. The NDR could appear due
65: to alignment and subsequent misalignment of the Fermi levels of the leads
66: with molecular orbitals which happens as the bias voltage increases
67: \cite{11,15}. This
68: may noticeably modify the coupling of the molecule to the leads. The
69: variations in the coupling strengths could serve as an immediate
70: reason for the NDR peaks to appear.
71: Also, such peaks could occur
72: as a Coulomb blockade induced effect \cite{5,16}, and/or they could originate
73: from conformational changes in the molecule \cite{17} and electron interaction
74: with the molecule vibrational modes \cite{7,18}. It is likely that different
75: mechanisms could play a major part in the NDR appearance in different molecular
76: junctions where it was observed so far. However, reviewing the available
77: experimental data, one may conclude that the most distinguished NDR features
78: in the current-voltage curves (sharp and narrow peaks separated by intervals of
79: extremely low conductivity, like those reported in the Ref. \cite{8}) are usually
80: attributed to the matching-mismatching of the molecule energy levels with
81: those of the leads. Keeping in mind that precisely
82: such NDR characteristics are potentially valuable for molecular electronics
83: applications, we further concentrate on this mechanism.
84:
85:
86:
87: Commonly, while studying electron transport through molecules, one assumes that
88: electrodes are large enough to have a feartureless electron density of states
89: below the Fermi energy.
90: In the present work we analyze the NDR effect in the electron transport
91: through a molecular junction where the leads are small metal clusters whose
92: electron density of states reveals sharp and distinct features. This system
93: provides better opportunities to analyze the effects of matching-mismatching
94: of the molecule energy levels with those of the ``nanoleads" in the electron
95: transport characteristics. We
96: show that under certain conditions such junctions may show very
97: distinguished multiple NDR features in the $I-V$ characteristics. This
98: demonstrates their potential usefulness in nanoelectronics applications.
99:
100:
101: \begin{figure}[t]
102: \begin{center}
103: \includegraphics[width=1.7cm,height=6.5cm,angle=-90]{p0.eps}
104: \caption{ (Color online)
105: Schematic of the considered system including large electron reservoirs
106: labelled $S$ and $D,$ respectively; metal nano\-clusters $L $ and $R$;
107: and the molecule placed in between.
108: }
109: \label{rateI}
110: \end{center}\end{figure}
111:
112:
113: To simplify the computational procedure we omit from the present consideration
114: effects originating from Coulomb interactions of electrons and from molecule
115: vibrations. To maintain a steady supply of electrons tunneling through the
116: junction, we assume that the metal clusters (``nanoleads") keep contact with
117: large source and drain electron reservoirs as sketched in the
118: Fig. 1. The latter, however, are separated
119: from the molecule is such a way that electrons cannot directly tunnel from
120: these reservoirs to the molecule. In the absence of the applied bias voltage
121: the whole system is supposed to be in the equilibrium state characterized with
122: the equilibrium Fermi energy $E_F.$ We write the effective Hamiltonian for the
123: molecule $(H_{eff})$ in the usual form \cite{1,19,20,21}:
124:
125: \be %f3
126: H_{eff} = H_M + H_{L} + H_{R}
127: \label{1}
128: \ee
129: where the term $H_M$ corresponds to the molecule itself, and $H_{L,R}$
130: describe the coupling of the latter to the metal
131: nanoclusters.
132:
133:
134: Omitting for a while the molecule coupling
135: to the leads, one may introduce the retarded Green's function for the
136: molecule. The latter is defined by the matrix
137: equation:
138: \be % f2,5
139: \big[(E +i\eta) \hat S - \hat H_M \big] \hat G_0^R = \hat I
140: \label{2}
141: \ee
142: where $I$ is the identity matrix, $ \hat S $ is the overlap matrix,
143: \be %f3,5
144: S_{ij} = \int \psi_i^* {\bf(r)} \psi_j {\bf(r)} d^3 r,
145: \label{3} \ee
146: $ \eta $ is an infinitesimal positive parameter $(\eta \to 0^+)$,
147: $ E $ is the energy, and $ \psi_i, \psi_j $ are the orbitals included
148: into the basis set.
149:
150: Then we employ the Dyson equation. This equation relates the Green's function
151: of the molecule coupled to the leads $\hat G^R (E) $ to the Green's
152: function of the single molecule $ \hat G_0^R (E). $ It reads \cite{22}:
153: \be %f4,8
154: \hat G^R (E) = \big\{ \big[\hat G^R_0 (E)\big]^{-1} - \hat\Sigma
155: (E) \big\}^{-1} . \label{4}
156: \ee
157: Here, the self-energy correction $\hat\Sigma(E)$ consists of two terms
158: describing the effect of two clusters:
159: \be %f5,9
160: \hat\Sigma (E) = \hat\Sigma_L (E) + \hat \Sigma_R (E).\label{5}
161: \ee
162: For convenience, one may introduce a notation $[\hat G_0^R(E)]^{-1} = \hat
163: A(E),$ which allows to simplify the form of the Eq. \ref{4}, namely:
164: \be
165: \hat G^R (E) = \big[\hat A(E) - \hat\Sigma(E)\big]^{-1}. \label{6}
166: \ee
167: Now, one must calculate matrix elements $A_{ij} (E) $ and $\Sigma_{ij}(E) $
168: Assuming
169: that the wavefunctions are orthonormalized molecular orbitals,
170: the matrix $\hat A (E) $ is a diagonal matrix:
171: \be %f7,10
172: A_{ij} (E) = (E +i\eta - E_i) \delta_{ij} \label{7}
173: \ee
174: where $E_i $ are the energy eigenstates of the molecule.
175:
176: The matrix elements of self-energy corrections have the form \cite{1}:
177: \be
178: (\Sigma_\beta)_{i,j} = \sum_k
179: \frac{W_{ik,\beta}^*W_{kj,\beta}}{E-\epsilon_{k,\beta} - \sigma_{k,\beta}} .
180: \label{8} \ee
181: here, $\beta \in L,R, \ W_{ik,\beta}$ are, respectively, the coupling
182: strengths between $``i"$-th molecule state and $``k" $-th state on the
183: left/right metallic cluster (lead), $\epsilon_{k,\beta} $ are energy levels
184: of the corresponding leads, and the parameters $\sigma_{k,\beta}$ are the
185: self-energy corrections which originate from the coupling of the clusters to
186: the large electron reservoirs. Their imaginary parts characterize the width
187: of the clusters energy levels. The summation over $``k" $ in the Eq. \ref{8}
188: is carried out over the states of the left/right cluster.
189:
190:
191:
192: When the bias voltage $ V $ is applied across the system shown in the Fig. 1,
193: this causes charge redistribution, and subsequent changes in the energies $E_i$
194: and $\epsilon_{k,\beta}.$ In consequence, the matrix elements $ A_{ij} $ and
195: $\Sigma_{ij} $ values vary as $V $ changes. This affects the electron transmission
196: function $T $ given by the expression:
197: \be
198: T = Tr \big\{\hat\Gamma^L\hat G^R \hat\Gamma^R\hat G^A \big\} \label{9}
199: \ee
200: where $\hat\Gamma^{L,R} =- 2\mbox{Im}\hat\Sigma_{L,R},$ and $G^A $ is the
201: advanced Green's function of the molecule $\big(\hat G^A = (G^R)^\dag\big)$.
202: When the voltage $ V $ is applied,
203: the electron transmission which determines transport properties of the
204: molecular junction may significantly depend on its value and polarity.
205:
206: The electron tunnel current through the junction could be written in the form:
207: \be
208: I= \frac{e}{\pi\hbar}\int_{-\infty}^\infty dE T(E,V_{mol})
209: \left[f(E - \mu_S) - f(E - \mu_D)\right]. \label{10}
210: \ee
211: Here, $T(E,V_{mol}) $ is the electron transmission, given by the Eq. \ref{9},
212: $ f(E)$ is the Fermi distribution function for the energy $E.$ Chemical
213: potentials $\mu_{S,D}$ are attributed to the source and drain reservoirs,
214: respectively. They are shifted
215: with respect to the equilibrium Fermi energy due to the applied voltage:
216: \be %f11,20
217: \mu_S = E_F + (1- \nu) eV; \qquad \mu_D = E_F + \nu eV
218: \label{11}
219: \ee
220: where $ e $ is the electron charge and $ \nu $ is the division
221: parameter.
222:
223: We remark that the transmission $ T $ in the Eq. \ref{10} actually depends
224: on the voltage applied across the junction $(V_{mol} )$. The latter may
225: noticeably differ from the external voltage $V.$ The difference originates
226: from both charge redistribution inside the molecule and the electrostatic
227: potential drops between the source/drain reservoirs and the metal clusters.
228: The current-voltage characteristics shapes crucially depend on the
229: electrostatic potential profile in the considered system. If the voltage
230: mostly drops between the large electron reservoirs and the small metal
231: nanoparticles included in the junction $(V_{mol} \ll V)$ one may
232: approximate $T(E,V)$ as $T(E,0)$.
233: Then the applied bias
234: voltage does not change relative positions of the energy levels
235: $\epsilon_{k,\beta} $ and $E_i,$ and low-temperature
236: characteristics should display step-like shapes.
237: These are typical for electron tunneling through molecules (see e.g.
238: \cite{19,23}). Current increases as the voltage $ V $ increases, and the
239: NDR does not appear.
240: On the contrary, when no significant voltage drop occures between the
241: electron reservoirs and the metal nanoclusters $(V_{mol} \sim V),$ the
242: effect of the bias voltage on the energy levels of the clusters
243: $\epsilon_{k,\beta}$ could considerably differ from its effect on the
244: molecular energy $ E_i.$ Due to these differencies in the voltage
245: induced shifts, the relative positions of the energy levels on the
246: clusters and these on the molecule vary as the voltage changes. This
247: creates opportunities for alignment/disalignment of the molecule orbitals
248: with those associated with the metal clusters. Therefore, in such a case
249: one may expect the NDR to occur.
250:
251:
252: \begin{figure}[t]
253: \begin{center}
254: \includegraphics[width=4.3cm,height=9.4cm,angle=-90]{ped1-2n.eps}
255: %\includegraphics[width=5.5cm,height=6.5cm,angle=-90]{p2m2.eps}
256: \caption{ (Color online) Left panel:
257: The calculated electron transmission function through the molecule $T(E,0)$
258: within a certain range of energies $E.$ Right panel:
259: The calculated current voltage chara\-cteristics for the electron tunneling
260: through the junction. The curves are plotted asuming $ T=50K, $ and
261: $ V_{mol} = 0.1 V $ (solid line) and $ V_{mol} = 0.9 V $ (dashed line).
262: }
263: \label{rateI}
264: \end{center}\end{figure}
265:
266: To confirm the above suppositions we carried out calculations of the tunnel
267: electric current through a junction which consists of two copper
268: nanoclusters and an alaphatic-saturated hydrocarbon chain situated in
269: between them (see Fig. 2). The relevant eigenenergies and matrix elements
270: included in the Eqs. \ref{3}-\ref{9} are computed using the NRLMOL software
271: package \cite{24}.
272: First we performed full self-consistent
273: calculations on the considered nanosystem.
274: In addition to solving the Kohn-Sham equations self
275: consistently we have optimized the geometry as well. Then we use Lowden's
276: method \cite{25} of symmetrical orthogonormalization to construct atom centered
277: wannier-like functions \cite{26} from the nonorthogonal atom-centered gaussian
278: orbitals. We reconstructed the Hamiltonian matrix in this basis
279: and block diagonalized it separating out the blocks coprresponding to the
280: hydrocarbon chain and the copper clusters.
281: Small off-diagonal elements between
282: the L-M and M-R blocks of the Hamiltonian describe the coupling of the molecule
283: chain to the nanoleads.
284:
285: The final matrix was used to determine the various matrices needed in
286: Eq. \ref{1}-\ref{11}.
287: In the following calculations of the tunnel current we
288: assumed $ V_{mol} = 0.1 V$ and $ V_{mol} = 0.9 V,$ respectively. The
289: resulting $ I-V $ curves are presented in the Fig. 2. One can see that when
290: the electrostatic potential mostly drops between the sourse/drain
291: reservoirs and the nearby copper clusters $ (V_{mol} = 0.1 V),$ the
292: corresponding $ I-V $ curve reveals a stepwise profile without NDR features.
293: On the contrary, distinct NDR peaks emerge provided that there exists a
294: significant drop in the electrostatic potential between the copper clusters
295: and the molecule tips $( V_{mol} = 0.9 V).$ Also, one may notice that some
296: NDR peaks are rather sharp and narrow, and the current peak values are much
297: greater than in the valleys between the peaks. This is consistent
298: with the experimental data reported in the \cite{8}.
299:
300:
301:
302: In conclusion, we have considered a metal-molecule-metal
303: tunneling junction where metal leads are nanoparticles.
304: Due to the extremely small
305: size of these metal clusters their electron density of states reveals distinct
306: features which are washed out for larger electrodes. We did show that
307: NDR peaks could appear due to alignment and subsequent disalignment
308: of the energy levels of the metal nanoclusters with those of the molecule
309: when the voltage applied across the junction varies. The proposed analysis
310: is based on simple assumptions concerning the electrostatic potential
311: distribution inside the junction. It seems unlikely
312: that these simplifications could cause qualitative distortions
313: in the considered NDR manifistations. However, to quantitatively analyze the
314: effect one must properly compute the electrostatic potential profile
315: employing a self-consistent computational procedure. Also, one should take
316: into account the effects of molecular vibrations which could result in extra
317: NDR features superimposed upon those presently analyzed. Nevertheless, we do
318: believe that we showed that the tunneling junctions including a molecule
319: placed in between metallic nanoparticles could exhibit distinct NDR peaks
320: in the $ I-V$ curves. This makes such junctions useful
321: is designing nanoelectronic devices.
322:
323:
324:
325: {\it Acknowledgments:} We thank G. M. Zimbovsky for help
326: with the manuscript. NZ acknowledges support from the ASEE and ONR Summer Faculty Research Program.
327:
328:
329:
330:
331:
332:
333: \begin{thebibliography}{99}
334:
335: \bibitem{1} S. Datta, {\it Quantum Transport: Atom to Transistor}
336: (Cambridge University Press, 2006).
337:
338:
339: \bibitem{2} See e.g. G. Cuniberti et al (Ed), {\it Introduction to Molecular
340: Electronics} (Springer, Berlin, 2005).
341:
342: \bibitem{3} L. Ezaki and P. J. Stiles, Phys. Rev. Lett. {\bf 6}, 1108 (1966).
343:
344: \bibitem{4} L. L. Chang, E. E. Mendez, and C. Tejedor, {\it Resonant Tunneling
345: in Semiconductors} (Plenum, New York, 1991).
346:
347: \bibitem{5}M. A. Kastner, Rev. Mod. Phys. {\bf 64}, 849 (1992).
348:
349: \bibitem{6} L. W. Yu, K. J. Chen, J. Song. J. M. Wang, J. Xu. W. Li, and X. F.
350: Huang, Thin Solid Films {\bf 515}, 5466 (2007).
351:
352: \bibitem{7} J. Gandioso, L. J. Lauhon, and W. Ho. Phys. Rev. Lett. {\bf 85}, 1918,
353: (2000).
354:
355: \bibitem{8} F.-R. F. Fan, R. Y. Lai, J. Cornil, Y. Karzari, J-L. Bredas, L.-T.
356: Cai, L. Cheng, Y. Yao, D. W. Price, Jr., S. M. Dirk, J. M. Tour, and A. J. Bard,
357: J. Am. Chem. Soc. {\bf 126}, 2568 (2004).
358:
359: \bibitem{9} A. Salomon, R. Arad-Yellin, A. Shanzer, A. Karton, and D. Cahen,
360: J. Am. Chem. Soc. {\bf 126}, 11648 (2004).
361:
362: \bibitem{10} M. Grobis, A. Wachowiak, R. Yamachica, and M. F. Crommie, Appl.
363: Phys. Lett. {\bf 86}, 204102 (2005).
364:
365: \bibitem{11} J. Repp, G. Meyer, S. M. Stojkovic, A. Gourdon, and C. Joachim,
366: Phys. Rev. Lett. {\bf 94}, 026803 (2005).
367:
368: \bibitem{12} J. J. Davis, T. Wang, A. Morgan, G. Zhang, J. Zhao, Faraday
369: Discuss. {\bf 131}, 167 (2006).
370:
371: \bibitem{13} E. D. Mentovich, I. Kalifa, A. Tsukernik, A. Caster, N.
372: Rosenberg-Shraga, H. Marom, M. Gozin, S. Richter, Small {\bf 4}, 55 (2007).
373:
374: \bibitem{14} G. Maruccio, P. Marzo, R. Krahne, A. Passaseo, R. Cingolani, and
375: R. Rinaldi, Small {\bf 3}, 1184 (2007).
376:
377:
378:
379: \bibitem{15} Y. Xue, S. Datta, S. Hong, R. Reifenberger, J. I. Henderson, and
380: C. P. Kubiak, Phys. Rev. B {\bf 59}, R7852 (1999).
381:
382: \bibitem{16} N. Simonian, J. Li, and K. Likharev, Nanotechnology {\bf 18}, 424006
383: (2007).
384:
385: \bibitem{17} V. Mujica, A. Nitzan, S. Datta, M. A. Ratner, and C. P. Kubiak,
386: J. Phys. Chem. B {\bf 107}, 91 (2003).
387:
388:
389:
390: \bibitem{18} M. Yu. Galperin, M. A. Ratner, and A. Nitzan, Nanoletters {\bf 5},
391: 125 (2005).
392:
393: \bibitem{19} N. A. Zimbovskaya and G. Gumbs, Appl. Phys. Lett. {\bf 81}
394: 1518 (2002).
395:
396: \bibitem{20} Y. Xue, S. Datta, and M. A. Ratner, J. Chem. Phys.
397: {\bf 115}, 4292 (2001).
398:
399: \bibitem{21} M. Yu. Galperin, M. A. Ratner, and A. Nitzan, J. Chem. Phys.
400: {\bf 121}, 11965 (2004).
401:
402: \bibitem{22} M. Yu. Galperin, A. Nitzan, and M. A. Ratner, Phys. Rev. Lett.
403: {\bf 86}, 166803 (2006).
404:
405:
406: \bibitem{23} V. Mujica, A. E. Roitberg, and M. A. Ratner, J. Chem. Phys.
407: {\bf 112}, 6834 (2000).
408:
409: \bibitem{24} M. R. Pederson, D. V. Porezag, J. Kortis, and D. C. Patton,
410: Phys. Status Solidi B {\bf 217}, 197 (2000).
411:
412:
413: \bibitem{25} P. O. Lowden, J. Chem Phys. {\bf 18}, 365 (1950).
414:
415:
416: \bibitem{26} M. R. Pederson and C. C. Lin, Phys. Rev. B {\bf 35}, 2273 (1985).
417:
418:
419: \end{thebibliography}
420:
421:
422:
423:
424: \end{document}
425:
426:
427:
428:
429: Prof. N. A. Zimbovskaya, Dept. of Physics,
430:
431: University of Puerto Rico at Humacao,
432:
433: CUH Station, Humacao, Puerto Rico 00791
434: \vspace{20mm}
435:
436:
437: Prof. N. A. Zimbovskaya, Dept. of Physics,
438:
439: University of Puerto Rico at Humacao,
440:
441: CUH Station, Humacao, Puerto Rico 00791
442: \vspace{20mm}
443:
444:
445:
446: Prof. N. A. Zimbovskaya, Dept. of Physics,
447:
448: University of Puerto Rico at Humacao,
449:
450: CUH Station, Humacao, Puerto Rico 00791
451: \vspace{10mm}
452: