Real-space observation of a two-dimensional electron gas at semiconductor heterointerfaces.

Mobile charge carriers are essential components in high-performance, nano-engineered semiconductor devices. Employing charge carriers confined to heterointerfaces, the so-called two-dimensional electron gas, is essential for improving device performance. The real-space visualization of a two-dimensional electron gas at the nanometre scale is desirable. However, it is challenging to accomplish by means of electron microscopy due to an unavoidable strong diffraction contrast formation at the heterointerfaces. We performed direct, nanoscale electric field imaging across a GaN-based semiconductor heterointerface using differential phase contrast scanning transmission electron microscopy by suppressing diffraction contrasts. For both nearly the lattice-matched GaN/Al0.81In0.19N interface and pseudomorphic GaN/Al0.88In0.12N interface, the extracted quantitative electric field profiles show excellent agreement with profiles predicted using Poisson simulation. Furthermore, we used the electric field profiles to quantify the density and distribution of the two-dimensional electron gas across the heterointerfaces with nanometre precision. This study is expected to guide the real-space characterization of local charge carrier density and distribution in semiconductor devices.

Linear imaging theory for differential phase contrast and other phase imaging modes in scanning transmission electron microscopy.

We propose a linear imaging theory for differential phase contrast under the weak-phase-weak-amplitude object approximation. Contrast transfer functions are defined for thin and thick weak objects, and they successfully describe several imaging characteristics of differential phase contrast. We discuss the defocus dependence of the contrast for several examples: atomic resolution, a p-n junction, a heterointerface, and grain boundaries. Understanding the imaging characteristics helps in adjusting aberrations in DPC STEM.

Quantitative electric field mapping in semiconductor heterostructures via tilt-scan averaged DPC STEM.

Differential phase contrast (DPC) in scanning transmission electron microscopy can be used to visualize electric field distributions within specimens in real space. However, for electric field mapping in crystalline specimens, the concomitant diffraction contrast is seriously problematic. In particular, for heterostructures with large lattice distortions, such as GaN-based semiconductor devices, the diffraction contrast cannot be reduced using conventional methods such as DPC image acquisition under off-axis conditions. In the present study, the electric field imaging of heterostructures is shown to suppress the diffraction contrast by averaging multiple DPC signals, obtained under various beam-tilt conditions near the zone axis. The remaining diffraction contrast was quantitatively estimated through simulations. This technique was demonstrated to enable the quantitative evaluation of electric field distributions across GaN/AlGaN multi-heterostructures, with errors possibly attributed to the residual diffraction contrast.

Quantitative electric field mapping of a p-n junction by DPC STEM.

Local electromagnetic fields in a specimen is measured at high spatial resolutions using differential phase contrast (DPC) imaging in scanning transmission electron microscopy (STEM). According to previous studies, DPC signals can be quantified by measuring the center of mass of the diffraction pattern intensity and/or performing a deconvolution method based on a phase contrast transfer function (PCTF). However, when using a segmented detector, the field strength has been considerably underestimated for a very thick specimen. The main cause of the underestimation is assumed to be inelastic scattering, mainly bulk plasmon scattering. In this study, we develop a method to remove this inelastic scattering effect from segmented detector DPC signals by modifying the PCTF deconvolution method. Field quantification results using this new technique are compared with those using pixelated detector DPC and electron holography, and all results indicated good agreement within an error margin.