Combined Scanning Tunneling Microscopy and High-Resolution

Aug 21, 2012 - ... Sports, Science and Technology, Japan (MEXT), and World Premier International Research Center Initiative (WPI), MEXT. We are gratef...
0 downloads 0 Views 1MB Size
Download PDF
Article pubs.acs.org/Langmuir

Combined Scanning Tunneling Microscopy and High-Resolution Electron Energy Loss Spectroscopy Study on the Adsorption State of CO on Ag(001) Ryuichi Arafune,*,† Hyung-Joon Shin,‡ Jaehoon Jung,§ Emi Minamitani,§ Noriaki Takagi,§ Yousoo Kim,§ and Maki Kawai§ †

International Center for Materials Nanoarchitectonics, National Institute for Materials Science, 1-1 Namiki, Ibaraki 304-0044, Japan School of Mechanical and Advanced Materials Engineering and Low Dimensional Carbon Materials Center, UNIST (Ulsan National Institute of Science and Technology), 100 Banyeon-ri, Eonyang, Ulju-gun, Ulsan 689-798, Republic of Korea § RIKEN Advanced Science Institute, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan § Department of Advanced Materials Science, The University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa, Chiba 277-8561, Japan ‡

ABSTRACT: The adsorption site and vibrational energies of CO on a clean Ag(001) surface were determined using scanning tunneling microscopy, inelastic electron tunneling spectroscopy with a scanning tunneling microscope, and high-resolution electron energy loss spectroscopy. The CO molecules were found to adsorb on the atop site of the Ag(001) surface, which was similar to their adsorption on the Cu(001) surface. The vibrational energy of the CO internal stretching mode was found to be 263 meV, which is only 3 meV less than that of CO in the gas phase. This result indicates that the CO molecules chemisorb very weakly on the Ag(001) surface.

■

INTRODUCTION Vibrational analysis of adsorbates on metal surfaces has provided a great deal of valuable information concerning not only the chemical bonding and local environment but also the dynamics of molecules on solid surfaces.1 In particular, CO is an extremely important adsorbate molecule for surface vibrational spectroscopy because of the large oscillator strength of its internal stretching (IS) mode, as well as the simplicity of its molecular structure and its chemical reactivity. In general, the vibrational energy of the IS mode is softened by the adsorption on metal surfaces. The strong chemisorption of CO leads to a large overlap of the antibonding 2π* orbital with surface electronic states, resulting in a sizable back-donation of electrons from the metal into the 2π* orbital.2 This weakens the internal bond strength and thus the force constant of the C−O bonding. As a consequence, the degree of interaction between the metal and CO can be evaluated from the vibrational energy of the IS mode. The stronger interaction leads to the lower vibrational energy. It is interesting to consider that this trend is applicable to weakly chemisorbed CO on the surfaces of coinage metals, such as Cu and Ag. For example, the binding energies are 0.57, 0.52, and 0.22 eV on Cu(001),3 Cu(011),3 and Ag(011), 4 respectively. The corresponding vibrational energies of the IS mode are 259 meV,5−11 259 meV,12 and 261 meV,13 respectively. These show that the trend is valid for these weak chemisorption systems. In contrast, a paradoxical relationship between the bond strength and vibrational energy of the IS mode has been observed for CO on Ag(001). The © 2012 American Chemical Society

vibrational energy was found to be 256 meV for the IS mode and 51 meV for the metal−C stretching (MC) mode for CO on Ag(001).14 This value of 256 meV indicates that the C−O bond is softened so that the interaction of CO with Ag(001) should be comparable to that of Cu(001).5−10 However, considering that CO certainly adsorbs onto Cu(001) even above 100 K15 while no CO adsorption on Ag(001) is usually observed at 100 K,14 it is clear that the interaction of CO with Ag(001) is weaker than that with Cu(001). Recently, Ortigoza et al.16,17 tried to solve this puzzle using a theoretical method. They calculated the vibrational energies and binding energy of CO on Ag(001) and demonstrated that the vibrational energy of the IS mode was almost the same as or slightly higher than that on Cu(001). Thus, the reason for the discrepancy is still unknown. To investigate this further, we experimentally assessed the adsorption state of CO on Ag(001) using a combination of scanning tunneling microscopy (STM) and high-resolution electron energy loss spectroscopy (HREELS). The vibrational normal modes of CO on Ag(001) assigned by HREELS experiments were verified by a comparison with inelastic electron tunneling spectroscopy with STM (STM−IETS) experiments and periodic density functional theory (DFT) calculations. Note that the HREELS spectra of CO on Ag(001) have not actually been shown in ref 14. The authors just Received: June 14, 2012 Revised: August 16, 2012 Published: August 21, 2012 13249

dx.doi.org/10.1021/la3024088 | Langmuir 2012, 28, 13249−13252

Langmuir

Article

to the positions of individual surface Ag atoms. The center of the dip is located at the cross position, which indicates that the CO molecule adsorbs on the atop site. Figure 2 shows the CO exposure dependence of HREEL spectra in the specular geometry. The intensity was normalized

mentioned the appearance of the energy losses at 51 and 256 meV under particular conditions. Thus, to the best of our knowledge, there have been no reliable experimental data for the vibrational mode of CO on Ag(001). In this study, the vibrational energy of the IS mode was found to be 263 meV, which was higher than that on Cu(001) (259 meV). The paradoxical relationship between the binding energy and the vibrational energy of the IS mode did not appear in the CO adsorption on both Ag(001) and Cu(001).

■

EXPERIMENTAL SECTION

In this work, two independent ultra-high-vacuum (UHV) systems were used for STM (Omicron LT-STM) and HREELS (Omicron, model IB500) experiments. The Ag(001) surface was cleaned by repeated cycles of Ar ion sputtering (600 eV, 20 min) and annealing (750 K, 30 min). The cleanliness was verified using low-energy electron diffraction, STM, and HREELS measurements. The sample was maintained at 4.7 K during both STM imaging and STM−IETS measurements. CO molecules were dosed onto the surface through a dosing tube located near the substrate at a temperature below 50 K. In the HREELS experiments involving CO exposure, the substrate was cooled to 20 K using liquid He. The primary electron energy was set at 4.5 eV, and the resolution was kept at a level better than 4 meV, as measured from the full width at half-maximum of the elastic peak. Note that this energy resolution is associated with the line width, not the vibrational energy position. The precision of the vibrational energy that depends on the loss intensity is significantly better than 4 meV. The elastic peak intensities were 1.4 × 106 cps for the clean surface and 7 × 105 cps for the saturated CO-covered surface. In all HREELS experiments the incident angle was fixed at 60° measured from the surface normal.

Figure 2. HREEL spectra of CO on Ag(001) taken at 20 K in the specular geometry. The peak intensities are normalized to the height of the elastic peak.

to the height of the elastic peak. There is a clear peak at 263 meV and a shoulder at 33 meV in the spectra. From the magnitude of the vibrational energy, the peak at 263 meV was assigned to the IS mode and the shoulder at 33 meV to the MC mode. Since greater exposure did not significantly affect the loss intensity of the 263 meV peak, 0.51 langmuir (1 langmuir = 1 × 10−6 Torr s) of exposure was regarded as the saturation coverage. Both peak and shoulder positions were essentially independent of the exposure amount. This tendency has previously been observed for CO on Cu(001). 5−8 In combination, these results imply weak dynamic coupling among the neighboring CO molecules on Ag(001). It is conceivable that a coverage-dependent chemical shift could compensate a dipole coupling induced frequency shift, but we believe this is unlikely due to the weak interaction of CO with silver. Further confirmation of the peak assignments required HREELS experiments to be carried out with off-specular geometry. Figure 3 shows the emission angle dependence of the HREEL spectra for saturated CO on Ag(001). Since the intensities of both the shoulder and main peak rapidly decrease as the emission angle deviates from the specular geometry, it is judged that these are of dipole origin. Therefore, it was confirmed that the 263 meV peak and the 33 meV shoulder were indeed due to the IS and MC modes, respectively. It should be noted that the IS mode on Ag(001) was higher in energy than that on Cu(001) (the vibrational energy of the IS mode of CO on Cu(001) is 259 meV5−10). The vibrational energy is almost identical to that of the CO molecule in the gas phase (266 meV18), indicating that the CO molecules adsorb very weakly. In addition to the 263 meV peak and the 33 meV shoulder, two other peaks were observed particularly at large deflection angles. One is located at 16 meV, which appears as a shoulder at small deflection angles. The loss intensity weakly depended on the emission angle, indicating that the vibrational mode observed at 16 meV was excited via the impact scattering rather than the dipole one. The other loss peak is present at around 5

■

RESULTS AND DISCUSSION Figure 1a shows a topographic STM image of CO on Ag(001) taken with 0.5 nA at 50 mV. A CO molecule on Ag(001) appears as a dip (Figure 1b). Figure 1c is an STM image taken by the tip modified with CO for accurate determination of the CO adsorption site. The cross points in Figure 1c correspond

Figure 1. (a) STM image of CO adsorbed on Ag(001) (7.0 × 7.0 nm2, VS = 50 mV, IT = 0.5 nA). (b) Height profile of a CO molecule along the dashed line marked in (a). (c) High-resolution STM image of CO adsorbed on Ag(001) measured with a molecular tip (7.0 × 7.0 nm2, VS = 9 mV, IT = 0.8 nA). The cross points indicate atop sites of silver atoms. The scale bars in (a) and (c) represent 1 nm. 13250

dx.doi.org/10.1021/la3024088 | Langmuir 2012, 28, 13249−13252

Langmuir

Article

CO occurred whenever the bias voltage exceeded 100 meV, making it impossible to detect the IS mode of CO on Ag(001) with STM−IETS. Thus, we have measured the IET spectrum within this energy range shown in Figure 4. The MC mode did not appear in the IET spectrum due to the small inelastic tunneling cross section. Using a combination of HREELS and STM experiments, the adsorption site of CO on Ag(001) was determined along with the vibrational energies of four normal modes for CO at the atop site. The periodic DFT calculations for CO adsorbed on Ag(001) demonstrated that the atop site is the most preferable position for CO, being more stable than bridge and hollow sites by 0.04 and 0.18 eV, respectively.22 The calculated vibrational energies for CO adsorbed on the atop site of Ag(001) agree qualitatively with the experimental data: 253.6, 27.4, 17.1, and 3.7 meV for the IS, MC, FR, and FT modes, respectively. Ortigoza et al.16 argued that the vibrational energy of the IS mode alone could not be used invariably to evaluate the metal− CO interaction because dynamic couplings among the neighboring CO molecules and from the substrate23 also significantly affect the vibrational energy of the IS mode. They concluded that the apparent paradoxical relationship between the vibrational energy of the IS mode and the bond strength, which had been observed on Ag(001) and Cu(001), was due to the different contributions of the dynamic couplings to the internal force constant of CO. Qualitatively, the argument that the dynamic couplings affect the bonding of CO on the metal, even though the metal−CO interaction is weak, holds. However, this assertion was based on insufficient experimental evidence. Table 1 summarizes the experimentally determined

Figure 3. Off-specular HREEL spectra of saturated CO adsorption on Ag(001) loss energy ranging from −30 to +70 meV. The incident angle is 60° as measured from the surface normal. The inset shows the loss energy range of 250−280 meV.

meV for the 77° emission angle, which is completely hidden under the strong elastic peak for the other emission angle measurements. From the emission angle dependence, it can be deduced that the 5 meV peak is also of impact scattering origin. Hence, the polarization of both modes is not oriented normal to the surface. Comparing the vibrational modes of CO on Cu(001),8,19,20 we assigned the 5 and 16 meV peaks to the frustrated translation (FT) and the frustrated rotation (FR) modes, respectively. Note that the observed vibrational energy of the FR mode is close to that on Ag(110) (19 meV).21 In the current experimental setup, the uncertainty of the energy is comparable to the vibrational energy of the FT mode. It is known that these frustrated modes on Cu(001) can be detected by STM-IETS.9 Figure 4 shows the second derivative

Table 1. Vibrational Energies (meV) for CO on Ag[001] Observed in This Work and Those on Cu[001] from References 5−10, 19, and 20 for Comparisona on Ag(001) on Cu(001)

CO IS

MC

FR

FT

263 2595−10

33 42.85−8

16 (18) 35.68,9,19

5 (4) 3.920

a

The values in parentheses are the vibrational energies determined by STM−IETS. Uncertainties of the vibrational energies in this work are less than 1 meV for the CO internal stretching mode and ±2 meV for the other three modes.

vibrational energies of CO on Ag(001) and also those on Cu(001) for comparison. It is clear that the results are not as contradictory as claimed. The lower vibrational energies of the MC and the FR modes are consistent with weak coupling of CO with Ag(001). The IS mode on Ag(001) was higher in energy than that on Cu(001). Consequently, the vibrational energy of the IS mode is still a reliable guideline in evaluating the bonding strength between CO and Ag(001). Figures 2 and 3 show a tiny peak at around 55 meV that we attribute to the adsorption of water molecules. Although the purity of CO gas was maintained by cooling the gas line to avoid water contamination, a very small amount of water was still present. Though the mass spectroscopic data did not show any significant signals due to water contamination in the exposed CO gas, cooling the gas line to 273 K significantly inhibited peak swelling. This means that the sticking probability of the water molecule on Ag(001) is high at low temperature.24 We surmise that the peak at approximately 51 meV observed by Burghaus et al.14 may have arisen from the adsorption of water onto Ag(001).

Figure 4. Vibrational spectrum (d2I/dV2) obtained from CO on Ag(001) by STM−IETS at a fixed tunneling gap of 0.35 nA and 140 mV with 4.0 mVrms ac modulations at 797 Hz. The spectrum was averaged over 32 passes.

of the tunneling current taken as an IET spectrum of CO on Ag(001). A lock-in technique, where a modulation voltage (4 mVrms, frequency 797 Hz) was superimposed to the bias voltage sweep, was used to measure the second derivative. Two peaks at approximately 4 and 18 meV, and the corresponding dips at negative bias, were consistently observed. This indicates that these arise from the inelastic electron tunneling process. The peak (dip) positions are identical to those of the peaks observed in the HREEL spectra with off-specular geometry, within experimental uncertainty. Incidentally, the migration of 13251

dx.doi.org/10.1021/la3024088 | Langmuir 2012, 28, 13249−13252

Langmuir

■

Article

(14) Burghaus, U.; Vattuone, L.; Gambardella, P.; Rocca, M. HREELS study of CO oxidation on Ag(001) by O2 or O. Surf. Sci. 1997, 374, 1. (15) Peterson, L. D.; Kevan, S. D. Coverage dependent desorption measurements of CO/Cu(001). Surf. Sci. 1990, 235, L285. (16) Ortigoza, M. A.; Rahman, T. S.; Heid, R.; Bohnen, K. P. Ab initio calculations of the dispersion of surface phonons of a c(2×2) CO overlayer on Ag(001). J. Phys.: Condens. Matter 2010, 22, 395001. (17) Ortigoza, M. A.; Heid, R.; Bohnen, K.-P.; Rahman, T. S. Nature of the binding of a c(2×2)-CO overlayer on Ag(001) and surface mediated intermolecular coupling. J. Phys. Chem. A 2011, 115, 7291− 7299. (18) Weber, A. Raman Spectroscopy of Gases and Liquids; Springer: Berlin, 1979. (19) Arafune, R.; Hayashi, K.; Ueda, S.; Uehara, Y.; Ushioda, S. Detection of the frustrated rotation mode of CO on Cu(001) by very low energy photoelectron spectroscopy. Surf. Sci. 2006, 600, 3536. (20) Ellis, J.; Toennies, J. P.; Witte, G. Helium atom scattering study of the frustrated translation mode of CO adsorbed on the Cu(001) surface. J. Chem. Phys. 1995, 102, 5059. (21) Lee, H. J.; Ho, W. Single-bond formation and characterization with a scanning tunneling microscope. Science 1999, 286, 1722. (22) The periodic DFT calculations were performed using the Vienna Ab-initio Simulation Package (VASP) code,25,26 with a PBE exchange-correlation functional27 and PAW pseudopotential.28 According to the experimental CO coverage, we used (2 × 2) surface supercells consisting of seven Ag layers, where the bottom two layers are frozen at their bulk positions. The periodically replicated slabs were separated by a vacuum region of ∼15 Å. Γ-centered grids (12 × 12 × 1) were used for the k-point sampling of the Brillouin zone. (23) Mahan, G. D.; Lucas, A. A. Collective vibrational modes of adsorbed CO. J. Chem. Phys. 1978, 68, 1344. (24) Ding, X.; Garfunken, E.; Dong, G.; Yang, S.; Hou, X.; Wang, X. The adsorption of water on clean and oxygen-covered Ag(100) studied by high resolution electron energy loss spectroscopy. J. Vac. Sci. Technol., A 1986, 4, 1468. (25) Kresse, G.; Hafner, J. Ab initio molecular dynamics for liquid metals. Phys. Rev. B 1993, 47, 558. (26) Kresse, G. Efficient iterative schemes for ab initio total-energy calculations using a planewave basis set. Phys. Rev. B 1996, 54, 11169. (27) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized gradient approximation made simple. Phys. Rev. Lett. 1996, 77, 3865−3868. (28) Kresse, G.; Joubert, D. From ultrasoft pseudopotentials to the projector augmented-wave method. Phys. Rev. B 1999, 59, 1758−1775.

SUMMARY We have investigated CO adsorbed onto Ag(001) by combining STM and HREELS and succeeded in determining the energies of four vibrational modes, the IS, MC, FR, and FT modes, as well as the CO adsorption site. From STM experiments, the CO molecule was found to adsorb on the atop site of Ag(001). The vibrational energies of the IS and MC modes were found to be 263 and 33 meV, respectively. Compared to CO on Cu(001), the high vibrational energy of the IS mode and low energy of the MC mode indicate that the CO adsorbed very weakly onto the Ag(001) surface. Thus, it is concluded that the vibrational energy of the interal stretching mode is a reliable gauge of the binding energy.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was partially supported by Grants-in-Aid for Scientific Research (24540332 and 21225001) from the Ministry of Education, Culture, Sports, Science and Technology, Japan (MEXT), and World Premier International Research Center Initiative (WPI), MEXT. We are grateful for the computational resources of the RIKEN Integrated Cluster of Clusters (RICC) supercomputer system.

■

REFERENCES

(1) Ibach, H.; Mills, D. L. Electron Energy Loss Spectroscopy and Surface Vibrations; Academic: New York, 1982. (2) Blyholder, G. Molecular orbital view of chemisorbed carbon monoxide. J. Phys. Chem. 1964, 68, 2772. (3) Peterson, L. D.; Kevan, S. D. Desorption and molecular interactions on surfaces: CO/Cu(OO1) and Cu(O11). J. Chem. Phys. 1991, 94, 2281. (4) Peterson, L. D.; Kevan, S. D. Coverage dependent desorption measurements for CO/Ag(011). J. Chem. Phys. 1991, 95, 8592. (5) Anderson, S. Vibrational excitations and structure of CO chemisorbed on Cu(100). Surf. Sci. 1979, 89, 477. (6) Ryberg, R. Carbon monoxide adsorbed on Cu(100) studied by infrared spectroscopy. Surf. Sci. 1982, 114, 627. (7) Andersson, S.; Persson, B. N. J. Inelastic electron scattering by a collective vibrational mode of adsorbed CO. Phys. Rev. Lett. 1980, 45, 1421. (8) Hirschmugl, C.; Williams, G.; Hoffmann, F.; Chabal, Y. Adsorbate-substrate resonant interactions observed for CO on Cu(100) in the far infrared. Phys. Rev. Lett. 1990, 65, 480. (9) Lauhon, L. J.; Ho, W. Single-molecule vibrational spectroscopy and microscopy: CO on Cu(001) and Cu(110). Phys. Rev. B 1999, 60, R8525. (10) Arafune, R.; Hayashi, K.; Ueda, S.; Uehara, Y.; Ushioda, S. Inelastic photoemission due to scattering by surface adsorbate vibrations. Phys. Rev. Lett. 2005, 95, 207601. (11) Strictly, there is an uncertainty in the reported values of the vibrational energy of the IS mode. A value of 259 meV is measured by infrared absorption spectroscopy,6,8 which is the highest energy resolution among the spectroscopic techniques that are used in refs 5−10. (12) Woodruff, D.; Hayden, B.; Prince, K.; Bradshaw, A. Dipole coupling and chemical shifts in IRAS of CO adsorbed on Cu(110). Surf. Sci. 1982, 123, 397. (13) Pelak, R.; Ho, W. Low temperature surface photochemistry: O2 and CO and Ag(110) at 30K. Surf. Sci. 1994, 321, L233. 13252

dx.doi.org/10.1021/la3024088 | Langmuir 2012, 28, 13249−13252