† Corresponding author. E-mail:
Project supported by the National Key R&D Program of China (Grant Nos. 2016YFA0300901 and 2017YFA0205003), the National Natural Science Foundation of China (Grant Nos. 11634001 and 21725302), and the Key Research Program of the Chinese Academy of Sciences (Grant No. XDPB08-1).
The plasmon-enhanced light emission of rutile TiO2(110) surface has been investigated by a low-temperature scanning tunneling microscope (STM). We found that the photon emission arises from the inelastic electron tunneling between the STM tip and the conduction band or defect states of TiO2(110). In contrast to the Au(111) surface, the maximum photon energy as a function of the bias voltage clearly deviates from the linear scaling behavior, suggesting the non-negligible effect of the STM tip on the band structure of TiO2. By performing differential conductance (dI/dV) measurements, it was revealed that such a deviation is not related to the tip-induced band bending, but is attributed to the image charge effect of the metal tip, which significantly shifts the band edges of the TiO2(110) towards the Femi level (EF) during the tunneling process. This work not only sheds new lights onto the understanding of plasmon-enhanced light emission of semiconductor surfaces, but also opens up a new avenue for engineering the plasmon-mediated interfacial charge transfer in molecular and semiconducting materials.
The light emission induced by the tunneling current of scanning tunneling microscope (STM) has received increasing attention recently, since it provides the unprecedented possibility of studying the electron-photon conversion at the atomic scale,[1–6] which is critical for understanding and controlling the properties of future nanoscale optoelectronic devices. Inelastic electron tunneling (IET) process can excite the localized surface plasmon (LSP) modes confined within the cavity between STM tip and the metal substrate. Those LSP modes decay into propagating photons in free space, which are detectable by far-field optical setups (Figs.
There are several difficulties for studying the plasmon-enhanced light emission on semiconducting materials by STM. First, the free-carrier density in semiconductors is much less than that in metals, leading to a much weaker plasmonic field.[15] Therefore, the plasmon enhancement of the tip-surface cavity is not as efficient as metal surfaces and the photon emission should be rather weak on semiconducting surfaces. Second, the band structure of semiconductors is more complicated than that of metals. The dopants and defects often induce prominent in-gap states,[16–18] which greatly modulate the inelastic electronic transition and thus the light emission process. Third, the STM tip can induce significant changes in the electronic structure of semiconductors, such as tip-induced band bending[19–21] and image charge effect.[22–25] In this work, we choose to investigate the rutile TiO2(110), which is one of the most widely studied materials in photoconversion and photocatalysis. We are able to obtain high-quality spectroscopy of light emission using our homemade optical STM and the effect of STM tip on the light emission is unambiguously revealed.
The experiments were performed with a homemade ultrahigh vacuum (UHV) STM system, with a base pressure better than 2 × 10−10 mbar. A rutile TiO2(110) single crystal (MaTech) was cleaned with cycles of sputtering by Neon ions and annealing at about 1000 K. In order to increase the conductivity at low temperature, the TiO2 sample was prepared with a high concentration of surface and subsurface defects. After the surface preparation, the sample was transferred immediately to the cryogenic STM stage to avoid the contamination of the TiO2(110) surface.
The STM characterization and light emission experiments of the TiO2(110) surface were performed with electrochemically etched silver tips. The scanning tunneling spectroscopy (STS) was acquired using lock-in detection of the tunneling current by adding a 13 mVrms modulation at 287 Hz to the sample bias at 5 K. Bias voltage refers to the sample voltage with respect to the tip.
Photons emitted from the tunneling junction were collected by a UHV compatible aspheric lens (Edmund Optics, EFL = 20 mm, N.A. = 0.38), which was fixed to a high-precision nanopositioner (Attocube) and then guided through a mirror system into a grating spectrometer coupled to a liquid-nitrogen-cooled charge-coupled device (CCD) (Princeton Instrument, SP2300). All light emission experiments were carried out at a sample temperature of 77 K. All the emission spectra were acquired at constant-current mode.
Owing to their characteristic dielectric function, noble metals support strong surface plasmon oscillations in the energy range corresponding to the visible lights.[26,27] The shape and intensity maxima of the light emission spectrum correspond to particular modes of plasmon resonance, determined by both the tip and the sample. For Au(111), it emits red lights when a bias of a few volts is applied. Figure
For semiconductor, such as TiO2, the mechanism of STM-induced light emission is similar with that of noble metal samples (Figs.
Figure
In order to understand the origin of the band edge shift, we carried out systematic dI/dV measurements for TiO2 surface. A series of STS spectra were taken at the defective area with different tip heights (Fig.
Such energetic shifts of band edge and defect states might be caused by the tip-induced band bending (TIBB).[20] In low-conductivity semiconductors, the potential between tip and sample drops over not only the vacuum gap but also an extended region in the semiconductor, which gives rise to TIBB. As a result, the dI/dV spectra may exhibit a considerable variation of the apparent bandgap. However, based on TIBB, the bandgap should appear substantially larger at smaller tip heights, which is contrary to our experimental results above. In addition, in our experiment, the EF is very close to ECB, which means that the sample is highly n-doped due to the high concentration of surface and subsurface defects. The high doping level of the sample may lead to efficient screening and thus negligible band bending.[30] Therefore, we can exclude TIBB as the origin of the energy shift.
The most possible origin of the bandgap narrowing is the image charge effect. When a point charge is placed before a metal tip, it induces a polarized cloud of opposite charges in the tip, which in turn lowers the energy of the point charge (Fig.
When an electron is injected into the conduction band of TiO2, the induced positive image charge in the tip will lower the electron energy, leading to the downward shift of the conduction band edge (ECB). When the electron tunnels from the defect state to the tip, the hole created in the defect state also feels an attractive interaction from the negative image charge in the tip, which pushes up the defect state (Ed) towards EF. Such a picture nicely explains the observed experimental results of light emission and dI/dV measurements.
We have investigated the plasmon-enhanced light emission of rutile TiO2(110) surface by a low-temperature STM. The photon emission arises from the irradiative decay of LSP excited by the inelastic electronic transition between the STM tip and the conduction band or defect states of TiO2(110). In contrast to the Au(111) surface, the maximum photon energy is not linear with the sample bias voltage, suggesting that the band edges are shifted towards the EF when the tip height decreases. By performing differential conductance (dI/dV) measurements, we exclude the tip-induced band bending as the origin of band edge shifts. Instead, we propose that such band shifts arise from the image charge effect of the metal tip. This work not only sheds new lights onto the understanding of plasmon-enhanced light emission of semiconductor surfaces, but also opens up a new avenue for engineering the plasmon-mediated interfacial charge transfer in molecular and semiconducting materials.
[1] | |
[2] | |
[3] | |
[4] | |
[5] | |
[6] | |
[7] | |
[8] | |
[9] | |
[10] | |
[11] | |
[12] | |
[13] | |
[14] | |
[15] | |
[16] | |
[17] | |
[18] | |
[19] | |
[20] | |
[21] | |
[22] | |
[23] | |
[24] | |
[25] | |
[26] | |
[27] | |
[28] | |
[29] | |
[30] |