Experimental Physics/Electron- and Ion-Nanooptics

Electric Potentials in Nanostructures

Determination of the Built-in Voltage at p-n Junctions

Electron holography in a TEM is often used to measure the built-in voltage of p-n junctions in doped semiconductors. The lateral distribution of p-n junctions is generally easy to image. However, the measured built-in voltage is always smaller than the expected voltage resulting from the doping levels. It is reported that, at least for silicon that is given a conductive path to ground by coating with carbon, the discrepancy is greatly reduced. Consequently, the discrepancy between measurement and expected value can be inferred to undefined band bending towards the surfaces, unknown surface potentials and the resulting stray electrostatic fields.

As part of an experiment, a needle has been prepared from a GaN wafer with a p-n junction using the Focused Ion Beam (FIB) and examined electron holographically. The key advantage of this geometry is the known sample thickness, so that the built-in voltage at the p-n junction can be determined from the measured phase difference between p- and n-region. By modifying the sample surface it was shown that the too low measured built-in voltage is not only due to the above mentioned experimental uncertainties, but that another physical effect must play a very important role.

p-n Junction under Electron Illumination

An effect only partially considered so far is the influence of illumination by the electron beam in the TEM, which generates electron-hole pairs, so that the p-n junction behaves like a solar cell illuminated by electrons in the TEM instead of light. In addition, the high-energy electron beam leads to the generation of secondary electrons, which are observed as a positive charge on poorly grounded samples.

Again, needle-shaped GaN p-n junctions were prepared by FIB and studied by electron holography. Here, the electron dose rate was reduced by three orders of magnitude. In order to still achieve a sufficient signal-to-noise ratio, hologram series with an accumulated exposure time of up to 1000 s have been recorded.

It turns out that the generation of electron-hole pairs indeed plays an important role in explaining the discrepancy, while the generation of secondary electrons can be neglected since they do not produce a net current across the p-n junction. Thus, of general importance for defined experimental conditions is a small short-circuit resistance, which can be easily achieved in the silicon system by a carbon coating, while in the GaN system there must be a much larger interfacial resistance between bulk and conducting surface layers [1]. These findings are important for further measurements, where a good contact of the bulk sample with the ground of the TEM has to be ensured, and they open the window for quantitative in-situ measurements under bias voltage.

[1] Jae Bum Park et al.,  Appl. Phys. Lett. 105 (2014) 094102.