1: %\documentclass{article}
2: \documentclass[prb,aps,preprint]{revtex4}
3: \usepackage{graphicx,amsmath}
4:
5: \begin{document}
6:
7: \title{Electron transport through dipyrimidinyl-diphenyl diblock molecular wire: protonation effect}
8:
9: \author{Zhenyu Li\footnote{E-Mail: zhenyuli@umd.edu.}}
10:
11: \address{Department of Chemistry and Biochemistry,
12: University of Maryland, College Park, 20742}
13:
14: %\date{today}
15:
16: \begin{abstract}
17: Recently, rectifying direction inversion has been observed in
18: dipyrimidinyl-diphenyl (PMPH) diblock molecular wire [J. Am. Chem.
19: Soc. (2005) 127, 10456], and a protonation mechanism was suggested
20: to explain this interesting phenomena. In this paper, we study the
21: protonation effect on transport properties of PMPH molecule by first
22: principles calculations. No significant rectification is found for
23: the pristine diblock molecular wire. Protonation leads to
24: conductance enhancement and rectification. However, for all
25: considered junctions with rectifying effect, the preferential
26: current directions are samely from dipyrimidinyl side to diphenyl
27: side. Effect of molecule-electrode anchoring geometry is studied,
28: and it is not responsible for the discrepancy between experiment and
29: theory.
30: \end{abstract}
31:
32: \maketitle
33:
34: \section{Introduction}
35:
36: Molecular rectifier is the first example proposed for electronic
37: device at molecular scale, \cite{aviram7477} and it is an
38: important topic in molecular electronics. \cite{nitzan0384,
39: joachim0501} Considerable experimental and computational efforts
40: have been devoted in recent years to molecular rectification.
41: \cite{Metzger0303} In a recent experiment on rectification of
42: dipyrimidinyl-diphenyl (PMPH) diblock molecular wire,
43: \cite{Morales0556} the preferential current direction was found to
44: be able to change possibly by protonation of the dipyrimidinyl
45: moiety. In this experiment, PMPH molecule was first co-assembled
46: on the Au(111) substrate together with dodecane through thiol
47: group. Then the top end of PMPH molecule was connected to a
48: suspended gold nanoparticle, also via thiol group. By using
49: different protect group at the two ends, the direction of the PMPH
50: diblock molecule on the surface can be precisely controlled. The
51: current-voltage characteristics were measured via scanning
52: tunneling spectroscopy (STS). Without protonation, an average
53: rectification ratio of 7.4 was observed, with the preferential
54: current direction from the diphenyl block to the dipyrimidinyl
55: block. After using perchloric acid in a methanol/tetrahydrofuran
56: mixture and sodium ethoxide in methanol to protnate and
57: deprotonate the nitrogen atom in dipyrimidinyl group, a reversible
58: change in the rectifying direction was observed. The protonated
59: diblock molecular wire gave an average inverse of the
60: rectification ratio as 9.2, with current preferring to flow from
61: dipyrimidinyl side to diphenyl side.
62:
63: Unfortunately, the microscopic mechanism of this interesting
64: experimental observation of rectification inversion remains unclear.
65: Moreles \emph{et al.} \cite{Morales0556} tried to suggest a simple
66: model to rationalize the rectification inversion. In their model,
67: before protonation, the intrinsic dipole moment of the diblock
68: molecule will induce local vacuum level shift and make the highest
69: occupied molecular orbital (HOMO) closer to electrode Fermi energy.
70: After the molecule is protonated, the positive charge centered on
71: the nitrogen reverses the direction of the dipole moment, which
72: makes the lowest unoccupied molecular orbital (LUMO) becomes closer
73: to the electrode Fermi energy. But we notice that the relative
74: position of HOMO/LUMO comparing to the electrode Fermi level is not
75: directly related to rectifying effect. An orbital approaching Fermi
76: energy will typically give conductance enhancement, as we can see
77: later, but rectification should be determined by the bias voltage
78: response of the relevant orbitals.\cite{taylor0201, li0693, li0616}
79:
80: In this paper, we report a comprehensive first principles
81: transport study on PMPH diblock molecular wire. Non-equilibrium
82: Green's function (NEGF) technique combined with density functional
83: theory (DFT) is used, which has been widely used in molecular
84: electronics and successfully applied to molecular rectification
85: analysis. \cite{taylor0201, li0693, li0616} In the rest part of
86: this paper, computational details are given in section II. In
87: section III, we discuss the protonation effect on transport
88: properties of PMPH molecule wire, with standard hollow site S-Au
89: anchoring. Different anchoring models are discussed in section IV.
90: At last, we conclude in section IV.
91:
92: \section{Computational Methods}
93:
94: The electronic structures are described with the implementation of
95: DFT in SIESTA program, \cite{Soler0245} which solves the Kohn-Sham
96: equation with numerical atomic basis sets. Double-$\zeta$ with
97: polarization (DZP) basis set is chosen for all atoms except Au,
98: for which single-$\zeta$ with polarization (SZP) is used. Our test
99: calculations indicate that using SZP basis set for Au does not
100: effect the accuracy of our calculations. Core electrons are
101: modeled with Troullier-Martins pseudopotentials.\cite{Troullier91}
102: Perdew-Zunger local density approximation \cite{Perdew8148} is
103: used to describe the exchange-correlation potential.
104:
105: The electronic transport properties are calculated with NEGF
106: technique using ATK package, \cite{Brandbyge02, atk} in which the
107: molecular wire junction is divided into three regions, left
108: electrode, contact region, and right electrode. The contact region
109: typically includes parts of the physical electrodes where the
110: screening effects take place, to ensure that the charge
111: distributions in the left and right electrode region correspond to
112: the bulk phases of the same material. The semi-infinite electrodes
113: are calculated separately to obtain the bulk self-energy. The
114: stacking order of the electrode atomic layer is chosen with
115: inversion symmetry for left and right electrode, which makes the
116: supercell not periodic along the current direction. Therefore,
117: the electrostatic potential is calculated with multigrid method
118: instead of fast Fourier transformation. A (2 $\times$ 2) and (4
119: $\times$ 4) $k$-point grid on the $x$-$y$ plane is used for
120: self-consistent calculation and transmission coeffecients
121: evaluation respectively.
122:
123: \section{Protonation Effect on Transport Properties}
124:
125: \subsection{geometrical model of molecular junctions}
126:
127: The molecular junction is modeled by sandwiching the diblock
128: molecule between two Au(111)-(3$\times$3) surfaces with thiol
129: anchoring group. Before exploring anchoring geometry effects in
130: the next section, we limit ourselves to the most popular hollow
131: site adsorption first. Because the PMPH molecule is embeded in
132: dodecanethiol self-assembly monolayer (SAM) in experiment,
133: \cite{Morales0556} the lower pyrimidinyl ring should highly
134: unlikely be protonated. So, we only consider protonaton for one or
135: both of the two nitragon atoms within the top pyrimidinyl ring.
136: The resulted molecular junctions are called monoprotonated PMPH
137: (MP-PMPH) and diprotonated PMPH (DP-PMPH).
138:
139: Fig. \ref{fig:geo} shows the optimized geometries for the three
140: molecular junctions with and without protonation. Only one unit
141: cell of the semi-infinite left/right electrode, which contains
142: three Au layers, is plotted. The dangling bonds shown in the
143: figure indicate the periodic boundary condition (PBC) in $x$-$y$
144: plane and semi-PBC for electrodes in $z$ direction. In the contact
145: region, two Au layers at both left and right sides are included,
146: of which the most left/right layer is constrained to its
147: theoretical bulk geometry to match the structure of the Au(111)
148: surface. The rest of the contact region is fully optimized. The
149: distance between the left and right electrodes is determined by a
150: serial of optimization calculations with different fixed
151: electrode-electrode distances.
152:
153: \subsection{transmission and current-voltage characteristics}
154:
155: Based on the optimized geometries, the current-voltage ($I$-$V$)
156: curves for PMPH, MP-PMPH, and DP-PMPH junctions have been
157: calculated self-consistently, as shown in Fig. \ref{fig:iv}.
158: Almost symmetric currents through PMPH molecular wire are obtained
159: for positive and negative bias voltages, especially at low
160: magnitudes of bias voltage. However, obvious asymmetric currents
161: are observed for MP-PMPH and DP-PMPH, for which negative bias
162: leads to larger current. The asymmetry of the $I$-$V$ curves can
163: be measured by rectification ratio ($R$), which is define by
164: $R(V)=|I(V)/I(-V)|$. In the inset of Fig. \ref{fig:iv}, we plot
165: inverse of $R$ as a function of bias voltage magnitude for all
166: these three junctions. $R$ approaches to 1 at low bias voltage for
167: all these three junctions, which means that the slope of I-V curve
168: does not change abruptly at zero bias. Although the rectification
169: ratios for MP-PMPH and DP-PMPH are distinguished with that of PMPH
170: by their much larger values, the rectifying directions are same
171: for all these three junctions, with current preferring to flow
172: from dipyrimidinyl side to diphenyl side. Besides rectification
173: effect, protonation also leads to conductance enhancement. The
174: conductance of MP-PMPH and DP-PMPH is about a factor of 2 and 4
175: respectively larger than that of PMPH. We notice that the
176: calculated currents are an order of 5 larger than the experimental
177: values. It may be partly caused by the high vacuum barrier between
178: gold nanoparticle and STM tip electron should overcome in
179: experiment. \cite{Morales0556}
180:
181: To understand the different electron transport properties for PMPH
182: molecular wire before and after protonation, we plot the
183: transmission curves for PMPH, MP-PMPH, and DP-PMPH at zero bias
184: voltage in Fig. \ref{fig:trans}. The main common feature of the
185: three transmission curves is that there are four broad peaks above
186: Fermi level up to 4 eV. Below the fermi energy, above -2.5 eV,
187: there are two transmission peaks. The one higher in energy is
188: however very low in transmission. For PMPH, the four
189: above-Fermi-level peaks are almost perfect resonances with
190: transmission possibility close to one. However, for MP-PMPH and
191: DP-PMPH, the first of these four peaks is much low than one. As we
192: can see, the conductance at zero bias voltage is mainly determined
193: by the tail of this peak. We name it peak $A$ thereafter. Although
194: the hight of peak $A$ is much smaller for MP-PMPH and DP-PMPH than
195: that for PMPH, the Fermi energy is farer away from peak $A$ for
196: PMPH. As a result, the conductance of PMPH is smaller than those
197: of MP-PMPH and DP-PMPH. From the transmission curves, we can
198: expect that MP-PMPH and DP-PMPH have similar conductance at zero
199: bias voltage, which is exactly what we observed in Fig.
200: \ref{fig:iv}.
201:
202: To consider the finite bias voltage transport and rectification
203: behavior, the voltage response of transmission spectrum should be
204: studied. In Fig. \ref{fig:trans}, we also plot the transmission
205: curves at a bias voltage of $\pm$1.0 V. Because the heights and
206: positions of peak $A$ is more strongly affected by the bias
207: voltage in MP-PMPH and DP-PMPH junctions than that in PMPH
208: junction, the latter gives much smaller rectification effect. We
209: notice that for MP-PMPH, there is a sharp transmission peak
210: overlapping with peak $A$ at zero bias voltage, and closing to
211: peak $A$ at -1.0 bias voltage. At negative bias, this sharp peak
212: even give near perfect transmission, but since the peak is narrow
213: and relatively far away from the Fermi level, it does not really
214: contribute to the low bias conductance of MP-PMPH.
215:
216: \subsection{MPSH orbital analysis}
217:
218: The transmission peaks can be considered as resonant transmission
219: of the electron through renormalized molecule orbitals, which can
220: be obtained by diagonalizing molecular projected self-consistent
221: Hamiltonian (MPSH). Both the energies of orbitals of free molecule
222: (with two thiol groups at two ends) and MPSH orbitals at zero bias
223: voltage are marked by upside-down triangles in Fig.
224: \ref{fig:trans}. The first row corresponds to molecular orbitals,
225: and the second row is MPSH eigenvalues. By comparing the energies,
226: we notice that almost all transmission peaks are mainly
227: contributed by a MPSH orbital. At the same time, the MPSH orbitals
228: are closely related to molecular orbitals. Typically, we can find
229: a one-to-one map for molecular orbitals and MPSH orbitals by
230: comparing their real space distribution. \cite{li0693, li0616,
231: wu0512} As shown in Fig. \ref{fig:trans}, their energies are also
232: comparable after a rigid shift. The amount of the rigid energy
233: shift for molecular orbitals is chosen to align an arbitrarily
234: chosen molecular orbital with its corresponding MPSH orbital, as
235: indicated by solid triangles in Fig. \ref{fig:trans}. There are
236: several reasons why an energy shift is necessary. The real part of
237: the electrode self energy may shift the MPSH energy, and, as
238: suggested by Morales et al., \cite{Morales0556} the dipole moment
239: of the molecule may also shift local vacuum level.
240:
241: Voltage response of transmission can also be analyzed by MPSH
242: orbitals. Bias voltage may change both the position and the shape of
243: a transmission peak. The peak position generally will follow the
244: energy of corresponding MPSH orbital. In Fig. \ref{fig:eig}, we plot
245: MPSH energies versus bias voltage, which can also be roughly
246: considered as transmission peak positions versus bias voltage. The
247: line shape and height of transmission peak will be determined by the
248: MPSH orbital real space distribution (transmission channel). Well
249: delocalized transmission channel will give almost perfect
250: transmission (close to one), and strong electrode-molecule coupling
251: results broad transmission peak. For all three junctions studied
252: here, according to the position of the Fermi energy, the low bias
253: transport is mainly determined by peak $A$, the first
254: above-Fermi-level peak. The real space distribution of the
255: corresponding MPSH orbital of peak $A$ is plotted in Fig.
256: \ref{fig:mpsh}.
257:
258: For PMPH, peak $A$ mainly comes from MPSH LUMO. As shown in Fig.
259: \ref{fig:eig} and \ref{fig:mpsh}, the energy of MPSH LUMO is only
260: slightly affected by the bias voltage, and the real space
261: distribution is also not very sensitive to bias voltage. As a
262: result, the position of transmission peak $A$ is almost
263: independent with the bias voltage, and peak width and height are
264: only slightly changed. For MP-PMPH, the corresponding MPSH orbital
265: is relatively insensitive to negative bias voltage, but become
266: much more symmetric and delocalized at positive bias voltage.
267: Therefore, its peak $A$ is much higher at positive bias voltage.
268: However, the MPSH orbital also upshift much at positive bias
269: voltage, which makes peak $A$ far from Fermi energy. Finally, we
270: get smaller conductance at positive bias voltage. The sharp peak
271: overlapped with peak $A$ at zero bias comes from the next MPSH
272: orbital, which is mainly distributed within bipyrimidinyl group,
273: and thus gives only a low transmission possibility.
274:
275: It is interesting to note that the transmission response to bias
276: voltage for peak $A$ behaviors in the same way for all these three
277: junctions. At negative bias, the peak shift to lower energy, and
278: its magnitude decrease. At positive bias, we get both higher peak
279: position and height. At a first look, it is strange, since the
280: molecules are oppositely polarized. However, if we look at the
281: corresponding MPSH orbitals for peak $A$ in Fig. \ref{fig:mpsh},
282: we find that they are all polarized at the same direction. They
283: are all distributed in the dipyrimidinyl group more than in
284: diphenyl group. Of course, the magnitude of polarization is very
285: different. The MPSH orbital for PMPH junction is almost
286: unpolarized, so it is well delocalized within the whole molecule,
287: and gives very high transmission. For MP-PMPH and DP-PMPH, the
288: MPSH orbital is highly polarized even at zero bias, therefore peak
289: $A$ is much lower than one, and the peak position is more strongly
290: dependent on the bias voltages. As a result, MP-PMPH and DP-PMPH
291: gives much larger rectification coefficient at small bias voltage.
292: Therefore, to understand the transport properties, we should look
293: into the electronic structure in more details, not only the
294: polarization of the whole molecule, but also the polarization and
295: its voltage response of the relevant orbitals which are close to
296: Fermi level.
297:
298: \section{Effects of molecular anchoring model}
299:
300: Until now, the thiol anchoring group is connected to gold surface
301: through the fcc hollow site. This is a standard model, but it may
302: not be adequate to describe the experimental setup by Morales et
303: al., \cite{Morales0556} where a gold nanoparticle is suspended.
304: The surface of a nanoparticle may be far away from a clean (111)
305: surface. On the other hand, the nanoparticle may address some
306: stress to the diblock molecule, and the electrode-electrode
307: distance may be much different from its equilibrium value. Another
308: big issue is that the STM tip typically is not contact with the Au
309: nanopariticle in STS measurements. Based on these concerns, we
310: construct two more geometrical models for PMPH and DP-PMPH, namely
311: apex model and cluster model. In apex model, we add an apex Au
312: atom on Au (111) surface at the bipyrimidinyl side, and obtain
313: junctions PMPH-A and DP-PMPH-A. In these two junctions, the
314: distance of the two electrode is determined by an optimization
315: with electrode represented by a Au$_4$ cluster. The cluster model
316: is constructed with the diblock molecule connecting to a sphere
317: shaped Au$_{13}$ cluster, which is separated with the Au(111)
318: surface by 4 \AA. The resulted two junctions are called PMPH-C and
319: DP-PMPH-C. The geometrical structures of these four junctions are
320: plotted in Fig. \ref{fig:geo2}.
321:
322: Despite the difference of anchoring geometries, the calculated
323: current-voltage curves, as shown in Fig. \ref{fig:iv2}, however,
324: are qualitatively same for standard, apex, and cluster model.
325: Especially for apex and standard model, the difference of I-V
326: curves is very small. The cluster model gives relatively larger
327: difference, but still no rectification inversion observed. This
328: result indicates that the failure to reproduce of the experimental
329: rectification inversion is not a result of unrealistic anchoring
330: model. It is interesting to notice that, with a 4 \AA\, gap, the
331: current through PMPH-C junction is larger than that through PMPH
332: junction. For DP-PMPH-C, current increases comparing to that of
333: DP-PMPH at positive bias voltages, and decreases at negative bias
334: voltages.
335:
336: The conductance behavior can be understood from transmission
337: spectra. In Fig. \ref{fig:trans2}, we plot the transmission spectra
338: for these junctions. For apex model, as we expected, the
339: transmission spectra are more or less the same as those for hollow
340: model, except that there is a shift for PMPH-A comparing to PMPH.
341: For cluster model, transmission characters are much more complicated
342: than standard model, because the electronic structure of the gold
343: cluster manifest itself in, as it does in dithiocarboxylate
344: anchoring group.\cite{li0616} The interesting conductance
345: enhancement of cluster model comes from some new small peaks very
346: close to Fermi energy. Therefore, for off-resonance transport, it
347: not obvious to get small conductance with weaker electrode-molecule
348: coupling. \cite{comment} At the same time, the richer transmission
349: feature brought by Au cluster makes the I-V curve less smooth for
350: the cluster model junctions.
351:
352:
353: \section{Conclusions}
354:
355: Transport properties of PMPH diblock molecular wire with and without
356: protonation are studied theoretically by combining NEGF and DFT.
357: Protonation is found to be able to enhance conductance and
358: rectification, but no rectifying direction inversion is found in
359: this study. Anchoring geometry effect is carefully checked and it
360: turns out that it is not relevant in this issue. Our calculations
361: indicate that the rectification inversion observed in experiment may
362: not be an intrinsic molecular property related to protonation, and
363: more sophisticated theory should be developed to explain this
364: experiment. There may be two important things missed in this study.
365: One is the interaction between molecular wire and environmental
366: dodecane and solvent molecules, and the other is the electron
367: correlation beyond the NEGF+DFT level of theory.
368:
369: \section*{acknowledgments}
370: The author is grateful to D. S. Kosov for helpful discussion.
371:
372: \begin{thebibliography}{99}
373: \bibitem{aviram7477}A. Aviram and M. A. Ratner, Chem. Phys. Lett. \textbf{29}, 277 (1974).
374: \bibitem{nitzan0384}A. Nitzan and M. A. Ratner, Science \textbf{300}, 1384 (2003).
375: \bibitem{joachim0501}C. Joachim and M. A. Ratner, Proc. Natl. Acad. Sci. U.S.A. \textbf{102}, 8801 (2005).
376: \bibitem{Metzger0303}R. M. Metzger, Chem. Rev. \textbf{103}, 3803
377: (2003).
378: \bibitem{Morales0556}G. M. Morales, P. Jiang, S. Yuan, Y. Lee, A. Sanchez, W. You, and L. Yu, J. Am. Chem. Soc. \textbf{127}, 10456 (2005)
379: \bibitem{taylor0201}J. Taylor, M. Brandbyge, and K. Stokbro, Phys. Rev. Lett. \textbf{89}, 138301 (2002).
380: \bibitem{li0693}Z. Li and D. S. Kosov, J. Chem. Phys. B \textbf{110}, 9893 (2006).
381: \bibitem{li0616}Z. Li and D. S. Kosov, J. Chem. Phys. B
382: \textbf{110}, 19116 (2006).
383: \bibitem{Soler0245}J. M. Soler, E. Artacho, J. Gale, A. Garcia, J. Junquera, P. Ordejon, and D. Sanchez-Portal, J. Phys.: Condens. Matter \textbf{14}, 2745 (2002).
384: \bibitem{Troullier91}N. Troullier and J. L. Martins, Phys. Rev. B \textbf{43}, 1993 (1991).
385: \bibitem{Perdew8148}J. P. Perdew and A. Zunger, Phys. Rev. B \textbf{23}, 5048 (1981).
386: \bibitem{Brandbyge02}M. Brandbyge, J.-L. Mozos, P. Ordejon, J. Taylor, and K. Stokbro, Phys. Rev. B \textbf{65}, 165401 (2002).
387: \bibitem{atk} ATK version 2.0, Atomistix A/S (www.atomistix.com).
388: \bibitem{wu0512}X. Wu, Q. Li, J. Huang, and J. Yang, J. Chem. Phys. \textbf{123}, 184712 (2005)
389: \bibitem{comment}We also want to emphasize that this conductance
390: enhancement of cluster model comparing to standard model may be an
391: artificial result at the NEGF+DFT level of theory, which does not
392: consider electron correlation very well.
393: \end{thebibliography}
394:
395: \clearpage
396: \begin{figure}
397: \includegraphics[keepaspectratio,totalheight=12cm]{geo.eps}
398: \caption{The relaxed geometries of Au-molecule-Au junctions. (a)
399: PMPH, (b) MP-PMPH, and (c) DP-PMPH. } \label{fig:geo}
400: \end{figure}
401:
402: \clearpage
403: \begin{figure}
404: \includegraphics[keepaspectratio,totalheight=10cm]{iv.eps}
405: \caption{Current-voltage curves for PMPH, MP-PMPH, and DP-PMPH
406: molecular junctions. Inset: corresponding inverse of rectification
407: ratios.} \label{fig:iv}
408: \end{figure}
409:
410: \clearpage
411: \begin{figure}
412: \includegraphics[keepaspectratio,totalheight=11cm]{trans.eps}
413: \caption{Transmission curves at zero (black solid), -1.0 V (red
414: dash-dot), and 1.0 V (green dash) bias voltage for (a) PMPH, (b)
415: MP-PMPH, and (c) DP-PMPH molecular junctions. Energies of
416: molecular orbitals and MPSH orbitals at zero bias voltage are
417: indicated by upside down triangles. The averaged electrode Fermi
418: energy is set to zero, and there is a rigid energy shift for
419: molecular orbitals.} \label{fig:trans}
420: \end{figure}
421:
422: \clearpage
423: \begin{figure}
424: \includegraphics[keepaspectratio,totalheight=10cm]{eig.eps}
425: \caption{Voltage response of MPSH eigenvalues for (a) PMPH, (b)
426: MP-PMPH, and (c) DP-PMPH molecular junctions. There are two energy
427: band plotted bold for each junction. The blue one comes from HOMO
428: of PMPH molecule, and is chosen to align molecular orbital and
429: MPSH orbital in Fig. \ref{fig:trans}. The similar energy and bias
430: voltage response of this MPSH orbital for all these three
431: junctions indicate it is a good choice for energy alignment
432: purpose. The red one comes from LUMO of PMPH molecule, and it
433: corresponds to the transmission peak $A$, which determines the low
434: bias transport properties.} \label{fig:eig}
435: \end{figure}
436:
437: \clearpage
438: \begin{figure}
439: \includegraphics[keepaspectratio,totalheight=6cm]{mpsh.eps}
440: \caption{MPSH orbitals corresponding to the transmission peak $A$,
441: which determines the low bias transport properties, for (a) PMPH,
442: (b) MP-PMPH, and (c) DP-PMPH molecule. From left to right, MPSH
443: orbitals at -1.0, 0, and 1.0 bias voltage respectively. }
444: \label{fig:mpsh}
445: \end{figure}
446:
447: \clearpage
448: \begin{figure}
449: \includegraphics[keepaspectratio,totalheight=16cm]{geo2.eps}
450: \caption{Geometry of Au-molecule-Au junctions. (a) PMPH-A, (b)
451: DP-PMPH-A junction, (c) PMPH-C, and (d) DP-PMPH-C. }
452: \label{fig:geo2}
453: \end{figure}
454:
455: \clearpage
456: \begin{figure}
457: \includegraphics[keepaspectratio,totalheight=10cm]{iv2.eps}
458: \caption{Current-voltage curves for PMPH, PMPH-A, PMPH-C, DP-PMPH,
459: DP-PMPH-A, and DP-PMPH-C molecular junctions.} \label{fig:iv2}
460: \end{figure}
461:
462: \clearpage
463: \begin{figure}
464: \includegraphics[keepaspectratio,totalheight=10cm]{trans2.eps}
465: \caption{Transmission curves at zero bias voltage for PMPH,
466: PMPH-A, and PMPH-C, DP-PMPH, DP-PMPH-A, and DP-PMPH-C molecular
467: junctions. } \label{fig:trans2}
468: \end{figure}
469:
470: \end{document}
471: