edfas.org ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 27 NO. 1 8 LOW FREQUENCY NOISE SPECTROSCOPY: A POWERFUL DIAGNOSTIC TOOL FOR TRAP IDENTIFICATION IN ACTIVE AND PASSIVE COMPONENTS B. Cretu1, A. Tahiat1, R. Coq Germanicus1, F. Bezerra2, C. Bunel3, A. Veloso4, and E. Simoen5 1Normandie Université, Caen, France 2Centre National d’Etudes Spatiales, Toulouse, France 3Murata Integrated Passive Solutions, Caen, France 4Imec Kapeldreef, Leuven, Belgium 5Ghent University, Gent, Belgium bogdan.cretu@ensicaen.fr EDFAAO (2025) 1:8-17 1537-0755/$19.00 ©ASM International® INTRODUCTION An essential aspect of evaluating the quality of materials and the device reliability is the use of fast, nondestructive, and accurate electrical characterization techniques that allow for the determination of key electrical parameters. Because of its sensitivity to defects and imperfections in the current path, low frequency noise has been employed as an efficient tool to assess device quality or reliability from both application and process optimization perspectives.[1-9] Low frequency noise studies are used to predict circuit performance, to obtain information on the charge transport mechanisms in devices, and to quantify device performances. Studying the generation-recombination (GR) noise as a function of the temperature gives information on the processing-induced defects in advanced MOSFET devices, identifying deep-level traps located in the gate stack or in the semiconductor material irrespective of the architecture and geometrical dimensions of the devices.[10-13] This diagnostic tool is more interesting as the low frequency noise related to free carrier trapping/ de-trapping phenomena increases with the decrease of the active surface of the components unlike other electrical characterization techniques for trap identification, e.g., the conventional deep level transient spectroscopy (DLTS) technique.[14] The methodology to estimate the noise parameters when one or several Lorentzian noise contributions appear in the total noise is described in references 10 and 15. The low frequency noise spectroscopy theory and methodology key points for traps located in the depletion zone are detailed in references 10 to 12. The work in this article focuses on traps located in the depleted region of the semiconductor material of already studied nanoscale transistor technologies. Technological and geometrical dimensions of the different architectures and typical low frequency noise spectra may be found in references 16 to 19 for ultra-thin buried oxide (UTBOX) transistors on silicon on insulator (SOI) substrates, in reference 15 for Si/SiGe superlattice I/O n-channel FinFETs, in reference 20 for gate all around (GAA) nanowires (NW), and reference 21 describes GAA vertically stacked lateral nanosheet (NS) FETs. An example is also given of a successful application of the low frequency noise spectroscopy to identify stable traps induced by proton irradiations on silicon passive devices.[22] THEORY AND METHODOLOGY The low frequency noise may contain flicker noise, white noise, and Lorentzian noise contribution. The generation-recombination (GR) noise corresponds to a Lorentzian spectrum, which is described by the expression:[10] (Eq 1) and is characterized by a corner frequency, f0, (or a char- acteristic time constant defined as τ0 = 1⁄((2πf0)) and by a plateau S0 that may be observed for frequencies f << f0. Considering uncorrelated noise sources and that the white noise level is Wn, the flicker noise level, Kf , at a frequency
RkJQdWJsaXNoZXIy MTYyMzk3NQ==