NANO LETTERS
Probing Molecule−Metal Bonding in Molecular Junctions by Inelastic Electron Tunneling Spectroscopy
2006 Vol. 6, No. 8 1693-1698
Mathias Kula, Jun Jiang, and Yi Luo* Theoretical Chemistry, Royal Institute of Technology, AlbaNoVa, S-106 91 Stockholm, Sweden Received April 27, 2006; Revised Manuscript Received June 26, 2006
ABSTRACT We present first-principles calculations for the inelastic electron tunneling spectra (IETS) of three molecules, 1-undecane thiol (C11), r,ωbis(thioacetyl)oligophenylenethynylene (OPE), and r,ω-bis(thioacetyl)oligophenylenevinylene (OPV), sandwiched between two gold electrodes. We have demonstrated that IETS is very sensitive to the bonding between the molecule and electrodes. In comparison with experiment of Kushmerick et al. (Nano Lett. 2004, 4, 639), it has been concluded that the C11 forms a strong chemical bond, while the bonding of the OPE and OPV systems are slightly weaker. All experimental spectral features have been correctly assigned.
Introduction. Inelastic electron tunneling spectroscopy (IETS) of molecular junctions has been introduced recently1,2 to the field of molecular electronics as a way of probing the molecule-metal interface as well as extracting information about the molecular conformation. Though IET spectra have proven to be highly sensitive and rich in detail, the exact nature of the experimental geometry is difficult to asses because of a lack of suitable references. Theory is needed to make accurate assignments and to interpret features of the spectra. Inelastic electron transport in molecular junctions has been studied by several groups3-10 using either model calculation or first-principles simulations. We have reported a study of the sensitivity of IET spectra previously3,4 and demonstrated that IETS is very sensitive to both intramolecular conformation and molecule-metal contact geometry changes. Our calculated spectra for octane dithiol are in perfect agreement with the experimental results of Wang et al.1 We have also reproduced the experimental spectrum of undecane thiol reported by Kushmerick et al.2 Our calculations show that the different profiles of two experimental spectra can be attributed to the changes in molecular conformation.3 A study by Troisi and Ratner9 focused on reproducing IETS for the molecules, 1-undecane thiol (C11), R,ω-bis(thioacetyl)oligophenylenethynylene (OPE), and R,ω-bis(thioacetyl)oligophenylenevinylene (OPV), reported by Kushmerick et al.2 They developed a computational model that is similar to ours but differs fundamentally in the way it treats the molecule-metal interface. They substitute the Au-S bond in the molecular junctions with a S-H bond, thereby * Corresponding author. E-mail:
[email protected]. 10.1021/nl060951w CCC: $33.50 Published on Web 07/22/2006
© 2006 American Chemical Society
producing a gas-phase-like system. The features that arise from the molecule-metal bonding are thus missing. Another study by Paulsson et al.10 also focus on calculating the IETS of these three molecules using periodic calculations at a density functional level. The bonding between metal and molecule is considered in the model. Their calculations have been able to reproduce the relative heights of the spectral peaks for the OPE and OPV systems, but not for the C11, for which the strong C-H peak around 2920 cm-1 shows too weak of an intensity in comparison with the experiment. The striking similarity between the results of Troisi and Ratner9 for gas-phase molecules and of Paulsson et al.10 for metal-molecule complexes seems to suggest that the IETS might not be sensitive to the actual metal-molecule bonding. If it should be a general rule, then the usefulness of IETS for the molecular devices could be reduced dramatically. Our previous calculations for octane dithiol, octane thiol, and undecane thiol systems have shown the effects of metalmolecule bonding on IETS clearly.3,4 Our results, in contrast to those of Troisi and Ratner,9 and Paulsson et al.,10 are in very good agreement with experiments for these systems. The intention of the present study is thus to examine whether our computational method can also work for the OPE and OPV systems. Furthermore, special attention will be paid to the dependence of IETS on the metal-molecule bonding. We have simulated the IETS of C11, OPE, and OPV in three different bonding situations. First we simulate the system in the gas phase, where the molecules couple weakly to artificially introduced electrodes. In the two other cases, the thiol or thioacetyl has been replaced by a Au-S bond. The sulfur is allowed to bind to two different types of
Figure 1. C11: (A) IETS for the gas-phase system, (B) IETS for C11 with chain contacts, (C) IETS for C11 with triangular contacts, and (D) experimental results.2 The geometries to the right indicate the used optimized geometries with an Au-S bond length of 2.85 Å.
contacts. In the first case, the sulfur binds over the middle of a triangular gold trimer, representing a bond over the hollow site of an Au(111) surface. In the second case, the outermost Au atom in a linear gold trimer, representing an ideal gold wire, is bonded with the sulfur atom. We use a Green’s function-based scattering theory to simulate the electron transport, where the vibronic coupling is introduced by expanding the electronic wave function along its normal modes within the harmonic approximation. Details of the method can be found in refs 3 and 4. The geometry optimizations have been done using the Gaussian0311 software package at a density functional theory (DFT) level with the B3LYP12 functional and the LanL2DZ13 basis set. The IETS have been simulated with the QCME program.14 In this work, we will focus on the spectral intensity rather than the specific spectral line width. We have thus adopted a uniform broadening factor, 1.24 meV or 10 cm-1, to all spectral features, which is only for the purpose of discussion. With this resolution, all important spectral features can be resolved. By increasing the broadening factor to 7.44 meV or 60 cm-1, we show that we can reproduce the experimental spectra very well, explaining most of the minor features not explained in previous theoretical studies.9,10 The experimental results together with the simulated IETS for the three situations considered are shown in Figure 1 for C11, Figure 2 for OPE, and Figure 5 for OPV. By considering these three ideal systems, a molecule in the gas phase, a perfect linear chain contact, and a perfect Au(111) surface represented by a triangular gold trimer, we gain an understanding of how the contact geometry perturbs the molecules and thereby influences the IET spectrum. In turn, when we compare the simulated IET spectra with the experiment, it helps us to choose which type of contact we should study more carefully with respect to bond distance. 1694
To be consistent with previous studies,3,4 we have used an Au-S bond distance of 2.85 Å, which will give the shortest junction widths in Figures 3 and Figure 6. The equilibrium geometries for OPE and OPV, however, will be slightly longer and have Au-S distances of about 3.0 Å. The experimental spectrum for C11, in Figure 1D, is dominated by a large C-H stretching peak at 2920 cm-1 and shows contributions from three modes corresponding to CH2 rocking, C-C stretching, and CH2 wagging. These modes are hinted in the gas-phase simulation as shown in Figure 1A; the corresponding IET spectrum is instead dominated by a δs(CH2) mode.15 Both bonded systems, whose IETS are shown in Figure 1B and C, are dominated by the same ν(C-H) peak as in the experiment. This is a strong indication that the supposed Au-S bond has actually been formed. Both the chain and triangular contacts give rise to the experimentally reported ν(C-C) and γw(CH2) modes, but the triangular contacts give better relative intensities, especially compared to the dominating ν(C-H) peak. Also, the peak positions agree better with experiment for the triangular contacts. In the case of chain contacts, the ν(C-H) peak is shifted up, while the ν(C-C) and γw(CH2) modes are shifted down slightly. However, only the chain contacts indicate contribution from the δr(CH2) mode, which suggests that the true bonding situation lies somewhere in between. The conclusion here is that the molecule is chemically bonded to one of the electrodes and the shape of the electrodes is probably somewhat flat as seen from the molecular end sites. The experimental results for OPE, displayed in Figure 2D, show three characteristic peaks corresponding to, using Wilson-Varsanyi terminology for aromatic rings,16 ν(18a), ν(8a), and ν(CtC) modes. The shapes of these peaks seem to suggest that they might be a sum of several overlapping Nano Lett., Vol. 6, No. 8, 2006
Figure 2. OPE: (A) IETS for the gas-phase system, (B) IETS for OPE with chain contacts, (C) IETS for OPE with triangular contacts, and (D) experimental results.2 The geometries to the right indicate the used optimized geometries with an Au-S bond length of 2.85 Å.
Figure 3. IETS for OPE due to changes in junction distance. The values in parentheses indicate the relative energy in kcal/mol (left) and junction width in Å (right). The red dashed lines represent the experimental IET spectrum. A broadening factor of 60 cm-1 has been used. Low frequencies, below 0.1 eV have been removed.
spectral features. Furthermore, the spectrum has intensity in the 100-300 meV region, which cannot be originated from the three dominating peaks mentioned above alone. The simulated spectrum for the gas-phase system in Figure 2A shows little resemblance to the experimental one. The characteristic peaks seen in the experiment are small compared to the dominating normal mode. Bonding to chain contacts, as shown in Figure 2B, does not reproduce the Nano Lett., Vol. 6, No. 8, 2006
experimental spectrum well. In particular, the two strong ν(8a) and ν(CtC) modes are too weak. The only peak that can be related to the experiment is the strong ν(18a) peak. The triangular contacts, however, (Figure 2C) reproduce the experimental spectrum very well, except for the apparent absence of the characteristic ν(CtC) peak. Interestingly, both bonded systems show significant contributions from ν(CS), which is also found in the theoretical predictions of 1695
Figure 4. IETS for OPE with a junction width of 25.66 Å. (blue, solid) Experimental (blue, solid) and calculated IETS with a broadening factor of 60 cm-1 (black, dashed) and a broadening factor of 10 cm-1 (red, dot-dashed). Regions and peaks marked with “bg” represent background contributions.
Figure 5. OPV: (A) IETS for the gas-phase system, (B) IETS for OPV with chain contacts, (C) IETS for OPV with triangular contacts, and (D) experimental results.2 The geometries to the right indicate the used optimized geometries with an Au-S bond length of 2.85 Å.
Paulsson et al.10 The absence of the ν(CtC) peak may be due to the fact that the simulated system is over-tightly bounded. As has been noted above, both Troisi and Ratner,9 as well as Paulsson et al.10 manage to find this peak for very loosely bound systems. To test this, we have gradually moved the planes spanned by the triangular contact atoms apart; that is, we have increased the junction width gradually. Because the simulated spectra in Figure 2 show many peaks, while the experiment only shows a broad spectrum with three distinct peaks and many other features, it becomes necessary to introduce a broadening that elucidates the effect of these simulated peaks 1696
on the overall spectra. We have therefore introduced a uniform broadening factor of 7.44 meV or 60 cm-1. Furthermore, we have removed the peaks for the lowfrequency region that give the ν(Au-S) modes because the small metal clusters used in the calculations cannot represent the electrodes very well. The resulting junction-width-dependent IETS can be found in Figure 3. When the junction width increases, so does the ν(CtC) peak relative to the other peaks. The reason for this is the change in coupling between the molecule and electrodes. If one just considers the ν(CtC) peak, it seems that increasing the total junction width by 0.6-0.8 Å gives Nano Lett., Vol. 6, No. 8, 2006
Figure 6. IETS for OPV at different junction distances. The values in parentheses indicate relative energy in kcal/mol (left) and junction width in Å (right). The red dashed lines represent the experimental IET spectrum. A broadening factor of 60 cm-1 has been used for the calculated spectrum. Low frequencies below 0.1 eV have been removed.
a reasonable agreement with experiment. This loosely bonded system would lie about 6.35 to 9.77 kcal/mol higher than the equilibrium geometry. However, a more careful analysis of the experimental spectrum might tell that the actual bonding situation is closer to the equilibrium geometry. Considering the situation when the junction width is 25.66 Å, the IETS for this system, in the interesting region of 0.10.3 eV, is shown in Figure 4. The red dot-dashed curve is the IET spectrum that one would get with a broadening factor of 10 cm-1, which shows resolved peaks corresponding to contributions from different normal modes. The same IET spectrum, but with a broadening factor of 60 cm-1, is given by the dashed black line, and the solid blue line gives the corresponding experimental IET spectrum. Contrary to the very simple IET spectrum reported by other groups,9,10 the experimental curve shows many bumps and shoulders that indicate that there might be many more modes than just the ν(18a), ν(8a), and ν(CtC) that make substantial contributions to the IET spectrum. As can be seen from Figure 4 when comparing the experimental curve with the 60 cm-1 broadened curve, almost all of the major features in the 0.10.22 eV region can be reproduced successfully. It is also possible to relate these features to the normal modes that are responsible for them by comparing with the IET spectrum with sharp peaks. In the energy region 0.23-0.26 eV, we see peaks that cannot be accounted for by theory because there are no normal modes in this energy region for the OPE molecule. We can conclude that they are most likely to be the background contribution. Around 0.295 eV, we also observe a small peak that cannot be attributed to vibrations of the molecular device. The actual intensity of the ν(CtC) peak should be much smaller than what it is shown in the experimental spectrum. Therefore, given the good agreement Nano Lett., Vol. 6, No. 8, 2006
in the 0.1-0.22 eV region and the possible background contributions, it is very reasonable to suggest that the true junction width lies around 25.66 Å, which is only slightly larger than the equilibrium geometry. Because the interaction energy typically displays an exponential dependence on distance, it is interesting to note that while the ν(CtC) increases the remaining parts of the spectrum remains relatively unchanged up to a 0.4 Å increase in the junction width. This can be compared to the theoretical results of Paulsson et al.,10 which find the IETS to be rather insensitive to a total displacement ranging from -0.3 Å to +0.4 Å. Our results thus largely agree with their results in the sense that small displacements of the contacts only induce small changes in the IET spectrum, but at the same time we demonstrate that the distance may still be an important factor that needs to be considered. Finally, the experimental IETS for OPV is shown in Figure 5, together with the theoretically predicted IETS for OPV in the gas phase, bonded between chain contacts and between triangular contacts. Here we also see that both a gas-phase system and a system bonded to triangular contacts can give rise to IET spectra that resemble the experimental situation. However, it should be possible to distinguish the gas-phase system from a bonded system because of the small contribution of ν(CdO), which can only come from the acetyl group. Another feature is that the gas-phase system shows very little contribution from the ν(18a) mode, which becomes a distinguishable feature next to the ν(C-S) peak in the case of triangle contacts. To determine whether the experimental situation has a strongly or weakly bonded system, we study the changes in the IETS with respect to junction width. The procedure is identical to that used for OPE, and the resulting junction-width dependence is shown in Figure 6. 1697
Although there is a mismatch in the exact position of the normal modes, there is a good agreement when considering relative intensities. Furthermore, many of the finer features of the experimental spectrum can be reproduced. The best agreement with experiment seems to come from the junction widths between 24.47 and 25.67 Å, though the 25.67 Å seems unlikely from an energetic point of view because it lies 9.18 kcal/mol above the equilibrium geometry. This study has demonstrated the power of the IETS for investigating the local bonding between molecule and electrodes in molecular junctions. Theoretical simulations have been able to provide detailed information that is not accessible in the experiments. It is shown that by comparing experiments with high-level theoretical calculations the determination of both local contact geometry and local metal-molecule bond length in molecular junctions becomes possible. The use of IETS can thus solve the long-standing problem in the molecular electronics, namely, the uncertainty in the contact geometries of molecular junctions made in different laboratories. Acknowledgment. This work was supported by the Swedish Research Council (VR). References (1) Wang, W.; Lee, T.; Kretschmar, I.; Reed, M. A. Nano Lett. 2004, 4, 643. (2) Kushmerick, J. G.; Lazorcik, J.; Patterson, C. H.; Shashidhar, R.; Seferos, D. S.; Bazan, G. C. Nano Lett. 2004, 4, 639. (3) Jiang, J.; Kula, M.; Lu, W.; Luo, Y. Nano Lett. 2005, 5, 1551. (4) Jiang, J.; Kula, M.; Luo, Y. J. Chem. Phys. 2006, 124, 034708. (5) Pecchia, A.; Di Carlo, A.; Gagliardi, A.; Sanna, S.; Frauenheim, T.; Gutierrez, R. Nano Lett. 2004, 4, 2109.
1698
(6) Chen, Y.-C.; Zwolak, M.; Di Ventra, M. Nano Lett. 2005, 5, 621. (7) Solomon, G. C.; Gagliardi, A.; Pecchia, A.; Frauenheim, T.; Di Carlo, A.; Reimers, J. R.; Hush, N. S. J. Chem. Phys. 2006, 124, 094704. (8) Sergueev, N.; Roubtsov, D.; Guo, H. Phys. ReV. Lett. 2005, 95, 146803. (9) Troisi, A.; Ratner, M. A. Phys. ReV. B 2005, 72, 033408. (10) Paulsson, M.; Frederiksen, T.; Brandbyge, M. Nano Lett. 2006, 6, 258. (11) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A., Jr.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; AlLaham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian 03, revision A.1; Gaussian, Inc.: Wallingford, CT, 2004. (12) Becke, A. D. J. Chem. Phys. 1993, 98, 5648. (13) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 270. (14) Jiang, J.; Luo, Y.; QCME-V1.0 (Quantum Chemistry for Molecular Electronics); Royal Institute of Technology: Sweden, 2004. (15) The notation used is δr for in-plane rocking, δs for in-plane scissoring, γw for out-of-plane wagging, γt out-of-plane twisting, and ν for stretching modes. (16) Varsanyi, G. Assignments for Vibrational Spectra of SeVen Hundred Benzene DeriVatiVes; John Wiley & Sons: New York, 1974.
NL060951W
Nano Lett., Vol. 6, No. 8, 2006