Nanoscale characterization of metal/dielectric/semiconductor interfaces using ballistic electron emission microscopy

• Main Author: Qin, Hailang
• Other Authors: Cedric Troadec
• Format: Theses and Dissertations
• Language: English
• Published: 2012
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Nanoelectronics
• Online Access: https://hdl.handle.net/10356/50608
• Institution: Nanyang Technological University

Ballistic electron emission microscopy (BEEM) was employed to study metal/dielectric/semiconductor devices with high spatial resolution. A comprehensive study of the subthreshold characteristics of the BEEM spectra shows that the BEEM current in the subthreshold region decreases at a rate of ~60 mV/decade. A comparative BEEM study of Au/SiO2/n-Si devices with in-situ and ex-situ deposited metal suggests that an ultra¬¬clean metal surface is critical for the BEEM studies of the devices with high electron barriers. Investigation of the ultrathin native oxide and high-κ dielectrics shows that an ultrathin dielectric (probably no thicker than 6–8 Å) does not exhibit the same electronic properties as a thicker dielectric. Spatial non-uniformity of these oxides on the nanometer scale was also observed by BEEM. We also determined the electron barrier height for a few important dielectrics (oxidized GaAs, Al2O3 and HfO2) on GaAs. An analytical equation describing the subthreshold BEEM behavior is derived from the basic BEEM model, i.e., the Bell-Kaiser model. This analytical equation shows that the BEEM current in the subthreshold region decreases exponentially with a subthreshold swing of about 60 mV/decade as the tip bias decreases. The equation is further verified by simulated and experimental BEEM spectra. Realizing the subthreshold characteristics, a new simplified model was proposed for the BEEM spectrum fitting. This proposed model takes the subthreshold behavior into account and it is to the best of our knowledge the first simplified analytical BEEM model that considers the temperature dependence. This model is shown to significantly improve the fitting of the BEEM spectrum compared with using the well-known simplified square model. It also provides a quick method to estimate the BEEM current at and below the threshold. It is shown that the BEEM measurement of samples with high barriers (~3.0 eV or above), cannot be performed properly if the metal base electrode is deposited ex-situ, while it can be performed consistently if the metal is deposited in-situ in an ultra-high vacuum. We further show that the main reason is that contamination of the Au surface strongly affects the tunneling between the tip and Au base in the high bias range, thus affecting the BEEM spectroscopy.