Superluminescence from an optically pumped molecular tunneling junction by injection of plasmon induced hot electrons

Here, we demonstrate a bias-driven superluminescent point light-source based on an optically pumped molecular junction (gold substrate/self-assembled molecular monolayer/gold tip) of a scanning tunneling microscope, operating at ambient conditions and providing almost three orders of magnitude higher electron-to-photon conversion efficiency than electroluminescence induced by inelastic tunneling without optical pumping. A positive, steadily increasing bias voltage induces a step-like rise of the Stokes shifted optical signal emitted from the junction. This emission is strongly attenuated by reversing the applied bias voltage. At high bias voltage, the emission intensity depends non-linearly on the optical pump power. The enhanced emission can be modelled by rate equations taking into account hole injection from the tip (anode) into the highest occupied orbital of the closest substrate-bound molecule (lower level) and radiative recombination with an electron from above the Fermi level (upper level), hence feeding photons back by stimulated emission resonant with the gap mode. The system reflects many essential features of a superluminescent light emitting diode.

power up to 300 μW on the sample). Integration time per individual spectrum was between 1 s and 3 s. All spectra were acquired by an energy calibrated spectrometer (Acton SpectralPro SP2300) with an attached intensity calibrated CCD camera (Princeton Instruments Spec-10). For recording a sequence of spectra the tip was positioned statically above the sample surface at a fixed position and the tunnelling current was kept constant at 1 nA. A 200 nm thick gold layer was evaporated on a silicon wafer as the substrate.
These substrates were then dipped for 10 min in a 10 −2 M solution of Cl-MBT (Sigma Aldrich, 90% technical grade) molecules in Uvasol methanol. Afterwards, the samples were carefully rinsed with methanol and dried in a clean-air box. With this proven procedure we produced self assembled molecular monolayer's where the Cl-MBT molecules bind via their sulphur atoms to Au substrate and the molecular plane oriented away from the surface plane [1].

Tip-sample separation
The tunnelling rate increases with the applied voltage and decays exponentially with the gap distance [2]. For a bare Au-Au-junction, we find for a constant current of 1 nA and a bias voltage of |200 mV| a tip-sample distance of one nanometer. Hence, since we keep the tunnelling current fixed at 1 nA the tip sample distance increases (see

Quantum efficiency [QE] per tunnelling electron
In the following we calculate first the number electroluminescence photons per tunnelling electron for the pure Au-Au junction and for the Au-Cl-MBT-Au junction.
Since the luminescence intensity from the pure Au-Au junction increases by applying a bias voltage of 2500 mV, we can similarly calculate the quantum efficiency per The optical detection efficiency of the whole system was approximately 1% (estimated by the efficiency of the detectors, spectrograph and the several mirrors and filters in the detection path). Therefore all measured intensities were multiplied by a factor of 100 to be consistent with the theoretical calculations.

Arrangement of the involved energy levels
Ultraviolet photoelectron spectroscopy (UPS) is well suitable for the determination of the energetic position of molecular orbitals at particular interfaces. UPS measurements were performed at the third-generation synchrotron radiation source BESSY II at the optics end station. Figure S6

Resonance frequency of the tip/sample junction
The resonance of the molecule-free tunneling junction is red-shifted by about 50 nm with respect to the broad luminescence from the Au substrate. hour. The spectra shown in Figure S8c show clearly the typical Raman bands of the MBT which are excited along their long molecular axis in excellent agreement with SERS-spectra known in the literature [5]. By comparing successive spectra recorded in one row with a separation of only 3.9 nm we clearly observe differences from one to the next spectrum such as intensity variations or splitting of individual Raman peaks. This behavior is reproducible and reflects local disorder in the film and the substrate morphology. Hence we can assume that our tip enhanced optical resolution is on the order of 4 nm or better. S11 Figure S8: Acquisition time is 300 ms per spectrum. b) and c), taken from the red square marked area. Clear differences in successive TERS-spectra of MBT can be observed reflecting a local disorder in the molecular film and the substrate morphology.