“Scanning tunnelling microscopy is a very powerful tool to see structures and electronic states with high spatial resolution,” says Hiroshi Imada, a researcher at the Surface and Interface Science Laboratory at RIKEN in Japan. “It’s also a unique electron source as well.”

When there’s enough potential difference between a sharp metal tip, like that in an STM, and a substrate, electrons in the tip have a probability of quantum tunnelling through the gap of air or vacuum that’s that increases exponentially as the size of the gap decreases. Depending on the substrate material, these injected electrons may then be able to recombine in the substrate with a hole – a state with a missing electron – and dissipate energy through a number of processes that may or may not radiate light.

These processes have been well studied in the bulk, but as devices shrink, surfaces play a more significant role in the overall behaviour. A better understanding of electron recombination and energy dissipation at surfaces could aid the development of devices from solar cells and electronics to catalysts.

“With an STM we can select every position of electron injection with high spatial resolution, and we can see the excitation and de-excitation so we can imagine what is going on for a particular excitation,” says Imada.

Using scanning tunnelling luminescence measurements Imada and his colleagues at RIKEN and the Tokyo Institute of Technology mapped the intensity of the photon yield on an atomic scale, and identified the specific electron states responsible. Their studies explain some of the behaviour observed in GaAs, a well-studied semiconductor that is a very efficient light emitter, but not when measured using STM.

Surface states

The researchers measured the photon yield from tunnelling-induced luminescence at the surface of p-type GaAs as they increased the voltage across the gap. They found that despite a rapid rise at 1.5V the photon yield soon saturated.

Imada and his team compared STM images of the surface, which measure tunnelling current, with atomically resolved images of the photon yield intensity. The correlation was good except that the gallium atoms showed up dark instead of light in the photon maps. The darkness of the gallium atoms in the scanning tunnelling luminescence images, which show bright in the tunnelling current images, suggests that non-radiative recombination dominates at the surface.

In addition, analysis of the rate equations and density functional calculations indicated that fast scattering between surface states prevents the injected electrons from penetrating into the bulk. As a result GaAs does not emit light efficiently when measured with STM. These surface scattering and non-radiative energy dissipation processes become particularly significant in nanostructures.

Steady does it

The potential of STM has been clear since Heinrich Rohrer and Gerd Binnig received a Nobel Prize almost 30 years ago for its development. Work by Richard Berndt and colleagues in Switzerland in the 1990s and since first demonstrated the potential of scanning tunnelling luminescence as well, but until recently these experiments have been challenging in practice.

“The most difficult point is that the photon intensity is very weak so we need a very stable experimental set up,” says Imada. “Now we can use low-temperature STM that is very stable with almost no drift.”

Other approaches to studying recombination have had limitations: cathodoluminescence and photoluminescence can only probe processes that occur in the bulk, while two-photon photoemission measurements, which can provide information on surface phenomena, only offer low spatial resolution. As a solution to both these challenges, scanning tunnelling luminescence experiments may be increasingly valuable in fundamental studies of materials systems including quantum dots, local defects and organic molecules.

Full details are available at Nanotechnology 26 365402.