Sample preparation for in‐situ biasing TEM experiments

Abstract

Knowledge about the potential distribution of semiconductor devices at the nanometer scale is important for the development of new device structures. Electron holography is a technique sensitive to the electrostatic potential with a high spatial resolution up to the sub nanometer scale [1, 2]. The potential especially in semiconductors is highly sensitive to the generation of electron‐hole pairs, damage caused by the electron and ion beam during investigation and preparation [3], and the biasing of the sample. In‐situ Biasing of specimen within the transmission electron microscope (TEM) provides an external stimuli allowing to differentiate between various effects influencing the hologram. A possible investigation is the influence of reverse and forward bias on the width of the depletion region and on the potential difference between p‐ and n‐sides of the diode [4, 5]. For the comparison between the the bulk device and the TEM‐ specimen we monitored the influence of the sample preparation on the electrical behavior. For these investigations a silicon solar cell was chosen [6]. The used sample geometry is inspired by the experiments of A.C. Twitchett [7]. The first steps are cleaving and grinding of a wedge out of the wafer. Afterwards two copper clamps are bonded with conductive silver on the front and back contacts of the specimen. This is followed by the preparation of a TEM lamella with a focused ion beam (FIB). A sketch of the FIB prepared lamella is shown in Figure 1. The preparation of the biasing sample is finished by bonding of the clamped specimen on a custom build carrier chip with conductive silver. Figure 2 shows the finished sample. After each preparation step, an IV measurement was performed and the corresponding curves are shown in Figure 3. As result, the influence of the FIB preparation on the whole sample is negligible (solid red and dashed blue). The measurements are not significantly changed before and after inserting the sample holder into the vacuum of the TEM (solid green and dashed orange). In reverse bias direction however a change of the slope after each bonding step is visible. Before bonding the parallel resistance is 70.1 kΩ. After bonding the slope of the current increases. This corresponds to the presence of a parallel resistance. Conductive silver or residues from its solvent might shortcut the device and build a parallel resistance. This parallel resistance decreases from 10.3 kΩ to 1.95 kΩ after the bonding steps. During investigations within the TEM an influence of the electron beam on the measured shortcut current was recognizable. Illuminating the FIB‐cut lamella portion of the specimen (and partially the thicker parts below it (left of the lamella in Fig. 1) produced a current of ‐2.1 μA, while illuminating the hole near the specimen only produced an current of ‐8.8 nA.