Electromagnetic waves can be guided and manipulated on sub-wavelength scales by exploiting collective electron oscillations e.g. at the interfaces of metallic nanostructures (surface plasmon polaritons). Technologically relevant applications have thus become possible, like signal-processing plasmonic circuits or plasmonic sensors. While the excitation in these devices is typically induced by incident light, their integration into conventional electronic circuits requires direct interfaces between nanoelectronics and nanophotonics, and strategies to excite plasmons electrically. This can be achieved by tunnel junctions addressing the plasmonic structures. A powerful method to study the mechanisms of localized plasmon excitation at plain metal surfaces and metallic nanostructures is scanning tunneling microscopy (STM).
This experimental technique sketched in Fig. 1(a) employs the close proximity (≈1nm) between the analyzed surface and a metallic tip that can be scanned with atomic precision. The highly localized tunnel current excites tip-induced plasmons in the gap between surface and tip. These gap excitations can couple to propagating surface plasmons or radiate as photons. Detecting the intensity of the emitted photons as function of tip position leads to so-called photon maps that reflect the local excitation efficiency for plasmons (see e.g. Fig. 1(c)). Introducing individual molecules into this tunnel junction modifies the plasmonic emission with respect to spectral shape and intensity . The critical assessment of the underlying mechanisms and their control will be decisive for the application of single molecules as ultimate coupling elements for the electrical plasmon generation. In this study we simultaneously map individual molecular orbitals and the light emission with sub-molecular precision. This additional spatial information unequivocally proves that molecular orbitals can be exploited to confine plasmon excitation spatially and energetically.
The model system which we study is sketched in Fig. 1(a). It consists of single fac-tris(2-phenylpyridine)iridium(III) admolecules (Ir(ppy)3) deposited on a C60 monolayer, which acts as an organic decoupling layer to the Ag(111) substrate. Iridium(III) complexes are promising candidates for highly efficient organic light emitting devices of the next generation. The molecular structure is shown in Fig. 1(b). Figure 1(c) shows an STM topograph of a single Ir(ppy)3 admolecule adsorbed on the perfectly ordered C60 layer. The simultaneously recorded spectrally integrated luminescence intensity is detected by a single photon counting detector. While there is weak light emission on the C60 layer, on Ir(ppy)3 molecules a well-defined sub-molecular structured light pattern with four times increased intensity is observed (yellow color in Fig. 1(c)). In this experiment single Ir(ppy)3 molecules on top of the buffer layer locally enhance plasmon and light excitation. The sub-molecular structure in the photon map and the topographic appearance of the molecules are markedly different which we explore in more detail in Fig. 2 on a different Ir(ppy)3 molecule. Differential conductance spectra in Fig. 2(a) exhibit three distinct peaks at bias voltages of –1.8V, –0 .9V, and –0.2V , marked by colored arrows. This proves that occupied electronic states are located at energies of 1.8eV, 0.9eV, and 0.2eV below the Fermi level. Deviations from these values by a few tenths eV are observed for different molecules due to varying adsorption geometries. The state closest to the Fermi energy at –0.2V is the highest occupied molecular orbital (HOMO) of the adsorbed species. In the positive voltage range, no Ir(ppy)3-specific states have been observed up to +3.0V. Differential conductance spectra on the C60 buffer layer (Fig. 2(a), grey curve) show that the molecular states at –0.2V and –0.9V lie in the C60 band gap. Figure 2(c)–(e) map the spatial extension of the three distinct orbitals. A spatially confined light pattern appears for bias voltages beyond –1.6V. There is a significant difference between the photon map at –1.8V (Fig. 2(b)) and the simultaneously recorded molecular orbital (Fig. 2(d)). The photon map shows, however, a close similarity to the shape of the HOMO at –0.2V (Fig. 2(c)).
To explain this observation, we consider the energy level scheme of the studied system (Fig. 3) derived from the identified molecular levels. The figure also indicates the energy levels obtained from density functional theory (DFT) calculations. The calculated levels agree well with experiment. At a bias voltage of –1.8V, the main tunnel current passes through the C60 HOMO and reaches the tip without significant energy loss. However, the excitation of plasmons and thus the recorded luminescence stems from two main inelastic tunnel processes. In these processes part of the electron energy is used for the generation of light. Path (1) is due to electrons which tunnel inelastically from the substrate to the tip without involving Ir(ppy)3 states. On the Ir(ppy)3 molecules, the presence of its HOMO opens an energetically and spatially defined gate that significantly enhances the tunnel process (path (2)) at this energy. As a consequence, on Ir(ppy)3 a larger part of the electrons has sufficient energy to excite plasmons in the visible range which results in the locally enhanced luminescence. Since the spatial shape of the gate is defined by the HOMO, the photon map matches the differential conductance map at –0.2V which results from the HOMO. The lower-lying Ir(ppy)3 orbitals play a minor role in the luminescence excitation process. The comparison between the pattern of the photon map and the different molecular orbitals provides by itself a useful tool to identify the origin of the observed luminescence. If the luminescence were due to an electronic transition between two electronic states of the molecule (intrinsic luminescence) and not due to inelastic tunneling we would observe the pattern of the lower electronic state (Fig. 2(d)) in the photon map.
In conclusion, single molecules located in a tunnel junction act as sub-molecularly defined spatial gates for the electrical excitation of plasmons and the emission of light. It is moreover found in our study that the energy difference between HOMO and the Fermi level leads to a characteristic spectral shift in light emission . The intensity and spectral distribution of the generated plasmons are thus directly related to the shape and the energy of the orbital closest to the Fermi energy. This concept may be applied to other adsorbates with known orbitals. It suggests a way to tailor the electrical excitation of plasmons. This can be important for new plasmon and light sources, for example for the direct integration of plasmonics into electronic circuits. Furthermore, the observed relationship between the excited plasmons and the molecular orbitals may provide an efficient way to monitor fundamental molecular processes like charging or conformational changes by using the radiation of plasmons as an ultrafast read-out signal.