A novel technique exploiting C–V, G–V and I–V simulations to investigate defect distribution and native oxide in high-κ dielectrics for III–V MOSFETs

Low leakage stoichiometric SrTiO3 dielectric for advanced metal–insulator–metal capacitors

“…The uniform defect distribution along the dielectric thickness ( N bulk ) is responsible for the trap‐assisted tunnelling (TAT) regime at low voltages, while the two near‐interface defect distributions ( N ni ) facilitate electron tunnelling at high fields. Minor discrepancies between measurements and simulations could be improved by further optimizing the shapes of the defect distributions, which is however out of the scope of this paper . These defects were considered to be associated with O vacancies because of their specific parameters (e.g.…”

(Invited) Defect Spectroscopy and Engineering for Nanoscale Electron Device Applications: A Novel Simulation-Based Methodology

“…The simulation-based spectroscopy is used to profile the defect distributions in a 1nm SiO x /5nm HfO 2 MOSFET with Si substrate. In order to identify the regions where defect effect is maximum on IV and BTI curves, we adopted the sensitivity maps (15). These maps are calculated by moving a narrow defect density distribution along a matrix of bin resulting from the discretization in space and energy of the dielectric stack.…”

“…The developed spectroscopic technique is based on the iterative procedure sketched in Fig. 4 (15), which relies on the interpretation in a consistent defect scenario of CV, GV and IV. Using complementary measurements allows widening the region within the dielectric bandgap along the thickness where defects can be profiled.…”

“…The energy-position coordinates of the defects assisting the IV current has been identified by the corresponding sensitivity regions, which represent the E-z domain within the stack bandgap hosting the defects that mostly contribute to the charge transport. On the other side, the defect distributions affecting the CV/GV characteristics (i.e., responding to the AC signal) were automatically extracted using the technique described in (15), consistently with the native oxide thickness. The defect distribution that allows consistently describing the experimental CV, GV and IV curves is shown in Fig.…”

“…6. The thickness of the native oxide layer (considered to be made of Ga 2 O 3 ) extracted from the procedure was t Ga2O3 ≈0.7nm (15). We considered the oxygen vacancies as the dominant defect species (24)(25) within the whole stack.…”