PhD Defense of Thibauld CAZIMAJOU

Random networks of nanowires, sometimes called nanonets, could be promising candidates for the 3D integration of CMOS biosensors. In this thesis we present characterization and simulation results of field effect transistors based on silicon nanonets (Si NN-FET). We show that measurements cannot be understood without account for dispersions within the nanonet.
The static electrical characteristics of these Si NN-FETs were measured for different geometric parameters (channel length and nanowire density) on a large number of devices, in order to obtain statistically significant orders of magnitude for the main electrical parameters (apparent low field mobility, subthreshold slope ideality factor and threshold voltage), which were extracted by means of a compact model. In parallel, the theoretical variations of these parameters were evaluated using percolation theory and Monte Carlo simulations. Compared to the usual approaches found in the literature for percolating networks, the originality of our simulations is to take into account both field-effect and dispersions. Threshold voltage dispersions proved to be essential to understand the experimental dependence of electrical parameters with network parameters. The analysis of Si NN-FET low frequency noise (LFN) made it possible to estimate the variation, with nanowire density, of the electrical area of the nanonet. From the temperature variation of Si NN-FET electrical parameters, it was found that inter-nanowire junctions were thermally activated. The unexpected variation of mobility with temperature suggests that junction barrier heights are widely dispersed, an assumption which was validated by the Monte Carlo simulations.

Jury members :

• Mireille MOUIS : Supervisor
• Emmanuel DUBOIS : Reviewer
• Philippe DOLLFUS : Reviewer
• Gérard GHIBAUDO : Co-Supervisor
• Edwige BANO : Examiner
• Cristell MANEUX :  Examiner
• Stéphane MONFRAY : Examiner