Tunneling electrons from a scanning tunneling microscope (STM) induce luminescence from C60 and C70 molecules forming fullerene nanocrystals grown on ultrathin NaCl films on Au(111). Intramolecular fluorescence and phosphorescence associated with the transitions between the lowest electronic excited state and ground state of C70 molecules are identified, leading to unambiguous chemical recognition on the nanoscale. Moreover we demonstrate that the molecular luminescence is selectively enhanced by localized surface plasmons in the STM tip-sample gap.
©2009 Optical Society of America
Light emission from pristine metal surfaces stimulated by scanning tunneling microscopy (STM) has been investigated in the past 20 years [1, 2]. It is now clearly established that the spectral characteristic of the light depends on the substrate as well as on the tip, and that its origin is the radiative decay of a localized surface plasmon (LSP) that is excited in the tip-sample gap region by inelastic electron tunneling.
Recently, STM-induced light emission (STM-LE) from adsorbed molecules which were electronically decoupled from the metal substrate by a thin spacer [3, 4] or several molecular layers [5, 6] has been observed. However, the exact role of the LSP in the molecular luminescence mechanism was not clear. Ho et al.  reported on an influence of the plasmon modes on the molecular light emission spectra, but it is only very recently that a plasmon enhancement phenomenon was proposed [6, 7]. On the other hand, the decisive role of plasmons is well known in surface- and tip-enhanced Raman spectroscopy [8,9], as well as in surface- or nanoparticle-enhanced fluorescence [10,11]. In particular, the tip-sample junction or the nanoparticle act as a nanoantenna which amplifies the local electromagnetic field .
Here we demonstrate, on the basis of measurements performed on C60 molecules, that the molecular light emission is enhanced in the STM tip-sample gap by efficient coupling to localized surface plasmon modes and report the first observation of fluorescence and phosphorescence from C70 molecules excited by tunneling electrons.
2. Experimental details
C60 and C70 nanocrystals were grown on thin insulating NaCl layers deposited onto an atomically flat Au(111) substrate. NaCl was deposited from a resistively heated evaporator onto a clean Au(111) surface at room temperature. Subsequently, the fullerenes were sublimated onto the NaCl covered substrate. Experiments have been performed with a homebuilt ultrahigh vacuum (UHV) STM operating at a temperature of 50 K, using etched W tips. Photons emitted from the tunnel junction were collected by a plano-convex lens (NA= 0.34) near the tip-sample gap along the direction 60° with respect to the surface normal. The collected beam was then transmitted through a view port outside the UHV chamber and guided simultaneously to (i) a grating spectrometer (50 l/mm) coupled to a liquid-nitrogen-cooled CCD camera for spectral analysis (90% of the signal) and to (ii) an avalanche photodiode to record the total light intensity and to optimize the alignment of the lens with the tunnel junction (10% of the signal). For the LE measurements, the tip was positioned over a target location with a fixed tunnel resistance. Spectra were not corrected for the wavelength dependent sensitivity of the detection system.
3. Results and discussion
Figure 1 shows STM topographic images of the investigated two types of samples. The fullerene molecules, C60 in (a) and C70 in (b), aggregate into extended monolayer islands on the bare metal surface (A) and form truncated triangular nanocrystals with a height of several molecular layers on NaCl (B). NaCl forms large (100)-terminated ultrathin islands on Au(111) of thickness between 2 and 3 monolayers (C).
The next step in the investigation of these complex samples is the characterization of their plasmonic properties related to the gold substrate. Figure 2(a) shows the plasmon-mediated light emission spectrum acquired on the bare Au surface (curve 1). This localized plasmon is excited through an inelastic electron tunneling mechanism and then decays radiatively [1,2]. Similar spectra are obtained with a reduced intensity on the ultrathin NaCl films (curve 2) and with very low intensity on the fullerene monolayer islands grown on the metal surface, as shown for C70 (curve 3). No intrisic luminescence is observed for the fullerene monolayers on Au(111), as radiationless relaxation takes place owing to the direct contact with the metal substrate, a process known in both, photoluminescence  and STM-LE experiments [5, 14].
In Fig. 2(b) the total plasmon-mediated light intensity measured on Au (curve 1) and on a NaCl layer (curve 2) as a function of the applied bias voltage is displayed. As expected, the LSP is excited for both polarities . However, the excitation parameters for bare Au and for the NaCl layer are not the same in the positive voltage range, in contrast to the negative one, owing to different image potentials . Considering the plasmon enhancement effect, the bias-dependent emission indicates that, if the NaCl layer is used as spacer to decouple the molecules from the metal substrate, a detection of molecular luminescence is only expected for negative voltages larger than -1.8 V and for positive ones between +1.8 V and +3.2 V.
In order to verify the effect of the plasmon amplification on the intrinsic molecular luminescence, different series of STM-LE spectra have been acquired on the fullerene nanocrystals. Figure 3 shows STM-induced light emission spectra acquired over two C60 nanocrystals (a) and over their respective NaCl spacer (b), for two different tip conditions (runs 1 and 2). The origin and nature of the molecular emission will be discussed later. The wavelength of the LSP resonance was changed between the two experimental runs by modifying the tip shape with a series of high-voltage pulses across the tunnel junction. Even though the two main peaks are found at similar positions in both molecular spectra, the spectral characteristics of the LSP influence their relative intensity. To substantiate this statement, we divide each molecular spectrum by its corresponding plasmon-emission spectrum. The resulting two normalized curves, shown in Fig. 3(c), identical in both spectral features and relative intensities, demonstrate an enhancement of the molecular luminescence in the STM tip-sample gap by coupling with the plasmon modes. This finding indicates that the localized surface plasmons are essential for the detection of molecular light emission.
Figure 4 shows STM-induced light emission spectra acquired over a C60 (a) and a C70 (b) nanocrystal, as well as over the NaCl spacer layers. Luminescence from the supported fullerenes, clearly distinguishable from the LSP emission, is observed for negative excitation voltages larger than -2.3 V for C60 and -2.5 V for C70. For positive voltages up to +5 V, no photon emission was detected for both fullerenes.
The observed bias dependence of the STM-induced light emission from the fullerene molecules is characteristic of a hot electron injection mechanism [3,5] followed by a radiative decay associated with the highest occupied molecular orbital-lowest unoccupied molecular orbital (HOMO-LUMO) gap of the molecules. This interpretation implies that the energy levels of the emitting molecules are not pinned to the gold substrate, but shift with the applied bias, as illustrated in Fig. 5(a). When the substrate is negatively biased with a voltage above the observed threshold value, the LUMO falls at an energy level lower than the Fermi energy (EF) of the sample and the HOMO at an energy level higher than the Fermi energy of the tip. Electrons tunnel elastically from the gold substrate into the LUMO through the NaCl barrier (hot electron injection) and simultaneously from the HOMO into the tip through the vacuum barrier to create electronically and vibrationally excited states of molecules. The transitions to the ground state, via the relaxation pathways illustrated in Fig. 5(b), give rise to the observed molecular luminescence. The fact that the latter is not observed for positive excitation voltages, in spite of the presence of the plasmon resonance between +1.8 V and +3.2 V, see Fig. 2(b), may be due to an asymmetry of the HOMO and LUMO position with respect to EF and to different characteristics for the tip-molecule and molecule-substrate double barrier [4, 16]. As a consequence, an energy level alignment leading to luminescence can not be realized in the range of positive bias corresponding to the plasmon enhancement.
The luminescence spectrum from a C60 nanocrystal shown in Fig. 4(a) (curve 2) is characterized by a peak at 738 nm followed by two shoulders, and a second peak at 828 nm. On the other hand, the excitation of a C70 molecular nanocrystal gives rise to two substantial different types of optical spectra shown in (b): (i) one characterized by an intense onset at 690 nm and small broad peaks up to 800 nm (type I) and (ii) a second one covering the spectral range between 700 and 950 nm with an intense line at 800 nm (type II). It is important to mention that the plasmon resonance, centered at 620 nm for C60 measurements (curve 1 in a) and at 650 nm (curve 1 in b) in the case of C70, was unchanged during each set of measurements. Thus, the existence of the two types of spectra observed for C70 can not be attributed to a different plasmon enhancement. For measurements on both fullerenes, a rigid shift of the entire spectrum has been observed, independently of the excitation voltages, covering to the following wavelength range for the main peaks: between 718 and 745 nm for C60; between 682 and 709 nm for C70 spectra of type I; between 800 and 813 nm for C70 spectra of type II. These shifts are independent of the corresponding plasmon-mediated emission.
In order to identify the electronic transitions giving rise to the observed molecular light emission spectra shown in Fig. 4, we compare our results with laser-induced photoluminescence data from fullerene molecules in different media.
Photoluminescence spectra acquired for C60 in rare gas matrices [17, 18], thin films [19,20], and single crystals [21, 22] have been considered. Although extensively investigated, the importance of solid state effects in the description of the luminescence process for C60 solids has not been completely clarified. Nevertheless, this comparison permits to attribute the observed spectra to fluorescence associated with the transition from the first excited singlet state to the ground state (S 1 → S 0). In the light of the present results on the plasmon-mediated amplification, the previously reported STM-induced luminescence from C60 molecules  has to be reinterpreted. Two different types of molecular spectra, assigned to fluorescence and phosphorescence, were claimed to be observed. The dissimilarity between both spectra is now understood as arising mainly from differently structured LSP spectra, originating from cut PtIr tips with unstable and ill-defined shape used in those measurements. We interpret the observed luminescence spectra as fluorescence from C60, although a contribution from a phosphorescence channel, corresponding to the transition from the first excited triplet state (T 1 → S 0), cannot be excluded in the low energy part of the spectra.
Intriguingly, the observation of both, fluorescence and phosphorescence is realized in the case of C70, as deduced from the comparison of our results with laser-induced photoluminescence data from dispersed C70 molecules in different media [23–25] and from C70 solids [26–28]. The observed spectra of type I are attributed to fluorescence (S 1 → S 0 transition) and the spectra of type II to phosphorescence (T 1 → S 0 transition). Because of a relatively low spectral resolution in the present experiment (8 nm), only the pure electronic origin of the triplet-to-singlet ground state transition is identified at about 800 nm. The other spectral features correspond to unresolved multiplet vibronic structures. Note that the broad peak at 800 nm in the spectrum of type I shown in Fig. 4(b) is assigned to the phosphorescence channel rather than to the fluorescence one. We indeed observed several times emission spectra consisting of a superposition of components from both radiative transitions.
The spectral shifts discussed above for both C60 and C70 which have been also observed in laser-induced photoluminescence spectra are attributed to a site dependence [21, 25]. The local environment or the presence of defects may also be at the origin of the two types of relaxation channels, fluorescence and phosphorescence, observed for C70.
In summary, light emission induced by tunneling electrons was observed from C60 and C70 molecules that were electronically decoupled from the metal substrate by a thin NaCl film. The molecular transitions giving rise to light emission were identified by comparison with laser-induced photoluminescence data from dispersed molecules or molecular crystals reported in the literature. The key role of localized surface plasmons in the enhancement of molecular emission is unambiguously demonstrated. The present observation of local fluorescence and phosphorescence represents a crucial step towards chemical recognition on the single molecule scale.
Financial support of the Swiss National Science Foundation is acknowledged.
References and links
4. E. Ćavar, M.-C. Blüm, M. Pivetta, F. Patthey, M. Chergui, and W.-D. Schneider, “Fluorescence and phosphorescence from individual C60 molecules excited by local electron tunneling,” Phys. Rev. Lett. 95, 196102 (2005). [CrossRef] [PubMed]
5. Z.-C. Dong, X.-L. Guo, A. S. Trifonov, P. S. Dorozhkin, K. Miki, K. Kimura, S. Yokoyama, and S. Mashiko, “Vibrationally resolved fluorescence from organic molecules near metal surfaces in a scanning tunneling microscope,” Phys. Rev. Lett. 1992, 086801 (2004). [CrossRef]
6. H. W. Liu, Y. Ie, R. Nishitani, Y. Aso, and H. Iwasaki, “Bias dependence of tunneling-electron-induced molecular fluorescence from porphyrin films on noble-metal substrates,” Phys. Rev. B 75, 115429 (2007). [CrossRef]
7. T. Uemura, M. Furumoto, T. Nakano, M. Akai-Kasaya, A. Salto, M. Aono, and Y. Kuwahara, “Local-plasmon-enhanced up-conversion fluorescence from copper phthalocyanine,” Chem. Phys. Lett. 448, 232–236 (2007). [CrossRef]
8. A. Campion and P. Kambhampati, “Surface-enhanced Raman scattering,” Chem. Soc. Rev. 27, 241–250 (1998). [CrossRef]
9. B. Pettinger, “Tip-Enhanced Raman Spectroscopy (TERS),” in Surface-Enhanced Raman Scattering - Physics and Applications, Top. Appl. Phys. 103, K. Kneipp, M. Moskovits, and H. Kneipp, eds. (Springer, Berlin / Heidelberg, 2006), pp. 217–242. [CrossRef]
10. E. Fort and S. Gresillon, “Surface enhanced fluorescence,” J. Phys. D 41, 013001 (2008). [CrossRef]
11. O. Stranik, R. Nooney, C. McDonagh, and B. D. MacCraith, “Optimization of nanoparticle size for plasmonic enhancement of fluorescence,” Plasmonics 2, 15–22 (2007). [CrossRef]
13. W. L. Barnes, “Fluorescence near interfaces: the role of photonic mode density,” J. Mod. Opt. 45, 661–699 (1998). [CrossRef]
14. R. Berndt, R. Gaisch, J. K. Gimzewski, B. Reihl, R. R. Schlittler, W.-D. Schneider, and M. Tschudy, “Photon-emission at molecular resolution induced by a scanning tunneling microscope,” Science 262, 1425–1427 (1993). [CrossRef] [PubMed]
15. R Berndt and J. K. Gimzewski, “Isochromat spectroscopy of photons emitted from metal surfaces in an STM,” Ann. Physik 2, 133–140 (1993). [CrossRef]
17. W. C. Hung, C. D. Ho, C. P. Liu, and Y. P. Lee, “Laser-induced fluorescence and phosphorescence of C60 isolated in solid Ne,” J. Phys. Chem. 100, 3927–3932 (1996). [CrossRef]
18. A. Sassara, G. Zerza, M. Chergui, F. Negri, and G. Orlandi, “The visible emission and absorption spectrum of C60,” J. Chem. Phys. 107, 8731–8741 (1997). [CrossRef]
19. Y. Wang, J. M. Holden, A. M. Rao, P. C. Eklund, U. D. Venkateswaran, D. L. Eastwood, R. L. Lideberg, G. Dresselhaus, and M. S. Dresselhaus, “Optical-absorption and photoluminescence in pristine and photopolymerized C60 solid films,” Phys. Rev. B 51, 4547–4556 (1995). [CrossRef]
20. V. Capozzi, G. Casamassima, G. F. Lorusso, A. Minafra, R. Piccolo, T. Trovato, and A. Valentini, “Optical spectra and photoluminescence of C60 thin films,” Solid State Comm. 98, 853–858 (1996). [CrossRef]
21. W. Guss, J. Feldmann, E. O. Göbel, C. Taliani, H. Mohn, W. Müller, P. Häussler, and H.-U. ter Meer, “Fluorescence from X traps in C60 single-crystals,” Phys. Rev. Lett. 72, 2644–2647 (1994). [CrossRef] [PubMed]
22. D. J. van den Heuvel, I. Y. Chan, E. J. J. Groenen, M. Matsushita, J. Schmidt, and G. Meijer, “On the fluorescence of crystalline C60 at 1.2 K,” Chem. Phys. Lett. 233, 284–290 (1995). [CrossRef]
23. S. M. Argentine, K. T. Kotz, and A. H. Francis, “Temperature and solvent effects on the luminescence spectrum of C70 - assignment of the lowest singlet and triplet-states,” J. Am. Chem. Soc. 117, 11762–11767 (1995). [CrossRef]
24. A. Sassara, G. Zerza, and M. Chergui,“Assignment of the lowest excited states of C70 and evidence for fluorescence from the S-2 state,” J. Phys. Chem. 102, 3072–3077 (1998). [CrossRef]
25. J. H. Rice, J. P. Galaup, and S. Leach, “Fluorescence and phosphorescence spectroscopy of C70 in toluene at 5 K: site dependent low lying excited states,” Chem. Phys. 279, 23–41 (2002). [CrossRef]
26. S. Kazaoui and N. Minami, “Intermolecular charge transfer excitons in C70 as compared with C60 films,” Synth. Met. 86, 2345–2346 (1997). [CrossRef]
27. M. Ichida, M. Sakai, T. Yajima, A. Nakamura, and H. Shinohara, “Luminescence due to intra-and inter-molecular transition in C70 single crystals,” Chem. Phys. Lett. 271, 27–32 (1997). [CrossRef]
28. V. Capozzi and G. Perna, “Role of the deposition parameters and aging on the optical and photoluminescence properties of C70 films,” Thin Solid Films 515, 7247–7252 (2007). [CrossRef]