11075
2008, 112, 11075–11077 Published on Web 07/09/2008
SERS of CO2 on Cold-Deposited Cu: An Electronic Effect at a Minority of Surface Sites⊥ M. Lust,† A. Pucci,† W. Akemann,‡ and A. Otto*,‡ Kirchhoff Institut fu¨r Physik, UniVersita¨t Heidelberg, Im Neuenheimer Feld 227, 69120 Heidelberg, Germany, and Institut fu¨r Physik der kondensierten Materie, Heinrich-Heine-UniVersita¨t, 40225 Du¨sseldorf, Germany ReceiVed: April 15, 2008; ReVised Manuscript ReceiVed: June 18, 2008
We compare the results of surface enhanced Raman scattering (SERS) and infrared reflection absorption spectroscopy (IRRAS) of carbon dioxide (CO2) on cold-deposited copper films. The SERS spectra of CO2 on copper films deposited at 40 K display neutral species at SERS active sites with bands not observed by Raman spectroscopy of CO2 gas, but identical to the loss bands of gaseous CO2 in electron energy loss spectroscopy. The absence of one component of the Fermi doublet of CO2 in SERS proves that the local electromagnetic field enhancement at SERS active sites cannot deliver signals above the noise level. The activated anionic CO2- is observed by transient electron transfer from the anionic molecule to the copper metal at a subgroup of SERS active sites, which are annealed below 200 K. The IRRAS spectra show only the expected infrared (IR) active modes of neutral CO2 representing the “majority species” of adsorbed CO2. I. Introduction SERS is a surface optical effect known since more than 25 years; see the geminating bibliography by Chang and Furtak.1 The discussion on the electronic contribution by transient charge transfer is still controversial,2,3 whereas the enhancement by increased local electromagnetic (EM) field in plasmonic resonances is generally accepted.4 At metal surfaces in ultrahigh vacuum, it is hard to prove the involvement of transient charge transfer effects. This is more easily achieved at electrode surfaces5–8 because the electrode potential can be varied, thus shifting the electronic levels of an adsorbed molecule with respect to the Fermi level in the electrode. The usual method of Raman excitation profiles cannot differentiate with certainty between plasmonic resonances and the electronic contribution to SERS.9 For CO2 however, the vibrational selectivity of SERS provides clear evidence for an electronic mechanism. Here we discuss previous SERS results10,11 in a new way using recent electron energy loss12,13 and infrared spectroscopic results. II. Discussion At first glance, one would expect nothing exciting for the neutral linear molecule O-C-O. For gas phase CO2, infrared spectroscopy yields the double generate O-C-O bending mode and Raman spectroscopy yields only the Fermi resonance doublet14 composed by the symmetric C-O stretch vibration and the second harmonic O-C-O bending mode.15 The lower frequency component (in gas phase at 1286 cm-1) will be denoted FRI and the higher frequency component (in gas phase at 1388 cm-1) FRII. The IRRAS spectrum of CO2 adsorbed on cold-deposited Cu films in the upper part of Figure 1 displays * To whom correspondence should be addressed. † Universita ¨ t Heidelberg. ‡ Heinrich-Heine-Universita ¨ t. ⊥ Dedicated to R. K. Chang on the occasion of his retirement.
10.1021/jp8032403 CCC: $40.75
Figure 1. Comparison of IRRAS and SERS of linear CO2 and activated CO2- on cold-deposited Cu films (40 K). The SERS spectrum at 8 L exposure is from ref.11 For further explanations, see the text.
a splitting of the double degenerate bending mode and the antisymetric stretch mode with nonsymmetric line shape depending on exposure, probably reflecting the disordered Cu surface. The broad structure near 2100 cm-1 is caused by CO contamination. The SERS results and the assignment of the bands to neutral and anionic CO2- on the substrates sapphire, potassium, colddeposited copper, and cold-deposited copper annealed at 200 K have been described extensively in ref 10; see also the Supporting Information. Figure 2 shows spectra and integrated band intensities as a function of exposure11,10 in the range of the Fermi resonance and the bending mode. The anionic CO2- species have a higher bonding enthalpy to the copper surface than the neutral species;10 hence, the bands of the anionic species have higher intensity at low exposures, see Figure 2b. All bands including FRII are broad and asymmetric, probably caused by strong electronic damping of the internal vibrations of CO2 at SERS active sites. Possibly the weak broad structure at about 1267 cm-1 (see Figures 1 and 2a) is the FRI band. Only at an exposure of 20 2008 American Chemical Society
11076 J. Phys. Chem. C, Vol. 112, No. 30, 2008
Letters
Figure 3. Overview of the elastic (no loss) and vibrational inelastic electron scattering cross sections in gaseous CO2 at a scattering angle of 135°. (100) signifies the symmetric stretch, (010) the bending mode, and (001) the antisymmetric stretch. The energy scale refers to the primary energy. Reproduced from ref 13. Figure 2. (a) SERS spectra at 20 L exposure in the range of the symmetric stretch and bending vibrations. Note the Fermi doublet of CO2 not adsorbed at SERS active sites with narrow line width at 1275 and 1382 cm-1. (b) Integrated intensities of the bending band δ of CO2 and δ(CO2-) versus exposure. (c) Integrated bands of FRII and the symmetric stretch mode νs(CO2-).
L also narrow bands of FRI at 1275 cm-1 and FRII at 1382 cm-1 appear, in close agreement with the Fermi doublet of CO2 condensed on sapphire. At the time of publication,10,11 it was not clear why only the higher component FII of the Fermi resonance appeared strongly and why the Raman forbidden bending mode was observed. Now a consistent electronic enhancement mechanism is indicated by comparison with the inelastic cross sections of electron scattering by gaseous CO2 in the energy range of 0-5 eV, including the 2Πu shape resonance at 2-5 eV12,13,16 and the so-called virtual state 2Σ of anionic CO2- below 1 eV17 in Figure 3. The cross sections of the Fermi doublet components below 1 eV display at decreasing electron energy an increase for FRII with and a negligible contribution of FRI. This experimental result in Figure 3 has been corroborated by theory.17 The selection rules in the 2Πu shape resonance and the virtual state 2Σ forbid the excitation of the antisymmetric stretch mode. The relatively strong cross section of the antisymmetric stretch mode (see Figure 3) at low electron energies is probably caused by a comparatively long-range electron interaction with the dynamic electric dipole of the antisymmetric stretch mode. This interaction will be shielded at the surface. The SERS spectra can easily be explained with the electronic energy scheme in Figure 4. The value of about 3.5 eV of the Coulomb relaxation in Figure 4 has been inferred from two-photon-emission spectroscopy18 of CO on Cu(111) and the resonant low energy electron scattering in CO gas.19 Within the coherent Raman process the “hot electron” created by laser photon annihilation will temporarily occupy the virtual state 2Σ of the directly adsorbed CO2 species and excite the various vibrational bands according to the cross sections in Figure 3 at energies below 0.5 eV. These are the bending mode
Figure 4. Electron energy diagram above the Fermi level of Cu. The work function of Cu(111) is 5.0 eV, fixing the vacuum level, e.g., the level of zero kinetic electron energy in the CO2 gas phase. The anionic electronic levels of the adsorbed CO2 are relaxed by the attraction of the shielding hole at the metal surface (see text). The most likely energy of the intermediate electron involved in temporary transfer to CO2 (indicated as “hot electron”) is the Fermi energy plus laser photon energy of 1.92 eV. All energies are up to scale.
and the higher frequency mode FRII. This explains the SERS spectrum of CO2 on Cu in Figure 1, especially the observation of the bending mode which is not Raman active for free CO2 and of only the higher frequency band FII of the Fermi dyad. If the transient formation of anionic CO2 did not exist, strong EM enhancement should show both FRI and FRII. But the near absence of the FRI component in SERS (see Figures 1 and 2) proves that the local EM field enhancement at SERS active sites cannot deliver signals above the noise level. The exclusive contribution of EM enhancement may cause the weak appearance of the narrow line Fermi doublet in Figure 2b. But even this is not granted, as demonstrated by a special electronic effect for the modes of C2H4 adsorbed in first layer on Cu(111) facets.20 The SERS spectrum of CO2 on Cu films cold-condensed at 40 K (Figure 1) contains a second couple of lines, which coincide with the positions of the CO2- bands observed on
Letters potassium, see the Supporting Information. These bands indicate the existence of the bent form of anionic CO2-, now permanently charged with one electron.21 The bending and symmetric stretching mode of the anionic form are observed by the transient transfer of the extra electron from CO2- to the metal, because these two nuclear motions are involved in the configurational change toward the linear molecular form of neutral CO2. This direction of photoinduced charge transfer is similar to the mechanism found in SERS of CN- on silver electrodes.7,8 When a cold-deposited Cu film, cryocondensed at 40 K, annealed at 200 K, and recooled to 40 K, is exposed to CO2, then the anionic CO2- is no more observed, only the bending modes and FRII of neutral CO2 persist. Apparently, the adsorption sites on Cu where CO2 is activated to its anionic shape are annealed below 200 K. Apparently, these “chemically active sites” form a subgroup of the “SERS active sites”.22 The fact, that the infrared active bending and symmetric stretch modes of anionic CO2- is not observed in IRRAS (see Figure.1) corroborates the assignment of the bonding of these species to a minority of surface sites. This resembles the case of NO exposure of cold-deposited Cu:23 SERS provides high spectroscopic sensitivity and is focusing on the “chemical active sites” covered with N2; N2-δ, N2O, Oads., but shows no NO, whereas with IRRAS only adsorbed NO was observed. III. Conclusions All spectral SERS features of neutral CO2 on SERS active sites of cold-deposited Cu involve transient electron transfer from Cu to CO2, leading to a temporary virtual 2Σ state of anionic CO2. This can be proven by a comparison of the SERS spectra with low electron energy loss spectra of gaseous CO2. The EM enhancement alone may deliver very weak Raman signals from species not adsorbed at SERS active sites, but this is not certain. On Cu films deposited at 40 K also a stable form of anionic CO2- is observed by SERS, involving transient electron transfer from anionic CO2- to Cu. The adsorption sites of the stable anionic CO2 are lost by annealing the film to 200
J. Phys. Chem. C, Vol. 112, No. 30, 2008 11077 K, whereas SERS of neutral CO2 is still observed after recooling to 40 K (see the supporting material). IRRAS observes the infrared active modes of the majority species of neutral CO2 adsorbed on cold-deposited Cu, whereas SERS is focusing on a minority of surface sites on the atomically rough Cu surface. Acknowledgment. We thank M. Allan for a discussion. Supporting Information Available: SERS spectra of CO2 on potassium, nonannealed and annealed cold-deposited Cu with assignments. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Surface Enhanced Raman Scattering; Chang, R. K., Furtak, T. E., Eds.; Plenum Press, New York, 1982. (2) Moskovits, M. J. Raman Spectrosc. 2005, 36, 485. (3) Otto, A. J. Raman Spectrosc. 2005, 36, 497. (4) Otto, A. Top. Appl. Phys. 1984, 54, 289. (5) Billmann, J.; Otto, A. Sol.State. Comm. 1982, 44, 105. (6) Otto, A. J. Electron Spectrosc. Relat. Phenom. 1983, 29, 329. (7) Furtak, T. E.; Macomber, S. H. Chem. Phys. Lett. 1983, 95, 328. (8) Billman, J.; Otto, A. Surf. Sci. 1984, 138, 1. (9) Pockrand, I.; Billmann, J.; Otto, A. J. Chem. Phys. 1983, 78, 6384. (10) Akemann, W.; Otto, A. Surf. Sci. 1992, 272, 211. (11) Akemann, W.; Otto, A. Surf. Sci. 1993, 287/288, 104. (12) Allan, M. Phys. ReV. Lett. 2001, 87, 33201. (13) Allan, M. J. Phys. B 2002, 35, L387. (14) Fermi, E. Z. Physik 1931, 71, 250. (15) Colbert, D.; Sibert, E., III J. Chem. Phys. 1989, 91, 350. (16) McCurdy, C.; Isaacs, W.; Meyer, H.; Rescigno, T. Phys.ReV. A 2003, 67, 042708. (17) Vanroose, W.; Zhang, Z. Y.; McCurdy, C. W.; Rescigno, T. N. Phys. ReV. Lett. 2004, 92, 053201. (18) Wolf, M.; Hotzel, A.; Knoesel, E.; Velic, D. Phys. ReV. B 1999, 59, 5926. (19) Ehrhardt, H.; Langhans, L.; Linder, F.; Taylor, H. S. Phys. ReV. 1968, 173, n/a. (20) Otto, A.; Akemann, W.; Pucci, A. Israel J. Chem. 2006, 46, 307. (21) Freund, H. J.; Roberts, M. W. Surf. Sci. Rep. 1996, 25, 225. (22) Otto, A.; Lust, M.; Pucci, A.; Meyer, G. Can. J. Anal. Sci. Spectrosc. 2007, 52, 150. (23) Lust, M.; Pucci, A.; Otto, A. J. Raman Spectrosc. 2006, 37, 166.
JP8032403
