Infoscience

Thesis

Ferroelectric Field Effect Transistor for Memory and Switch Applications

Silicon technology has advanced at exponential rates both in performances and productivity through the past four decades. However the limit of CMOS technology seems to be closer and closer and in the future we might see an increasing number of hybrid approaches where other technologies add to the CMOS performance, while maintaining a back-bone of CMOS logic. Ferro-electricity in ultra-thin films has been investigated as a credible candidate for nonvolatile memory thanks to the bistability of polarization. 1 transistor (1T) ferroelectric memory cells have been proposed and experimentally studied in order to reduce the size of 1T-1C (1Transistor-1Capacitor) design with consequent advantages in terms of size, read-out operation and costs. More recently ferroelectrics have been proposed by Salahuddin and Datta as dielectric materials in order to lower the 60mV/dec limit of the subthreshold swing (SS) in silicon Metal Oxide Semiconductor Field Effect Transistors, MOSFETs. The objective of this thesis is to study the ferroelectric transistor performance for both memory and switch application. For this purpose different Ferroelectric Field Effect Transistors, Fe-FETs, structures have been designed, fabricated and characterized. An organic ferroelectric polymer, vinylidene fluoride trifluorethylene, P(VDF-TrFE), of 100nm and 40nm thickness has been successfully integrated into the gate stack of bulk and SOI MOSFET and, later, on a Tunnel FET, TFET, structure. The 1T ferroelectric FET memory cells have shown a programming time in the order of ms at 9V as programming voltage. The retention of a few seconds, however, is the main limiting factor for the usage of this device for NV-memory applications. The retention failure mechanisms have been studied and investigated for future improvement. For the first time this work experimentally demonstrates that a subthreshold swing lower than 60mv/dec can be achieved in a ferroelectric transistor thanks to the voltage amplification arising from the ferroelectric material. This unique finding has been first measured in a 40nm P(VDF-TrFE)/10nm SiO2 gate stack MOSFET and then, confirmed, in a 100nm P(VDF-TrFE)/10nm SiO2 gate MOSFET with an intermediate contact between the two dielectrics. This internal node contact allows the study of the voltage amplification due to the ferroelectric material. Finally a temperature study of the performance of a ferroelectric Fully Depleted Silicon on Insulator, FD SOI, transistor has been done. A model based on Landau's theory has been carried out and it has been experimentally validated for both the subthreshold and the strong inversion regions. It has been demonstrated for the first time that, because of the divergence of the ferroelectric permittivity at the Curie temperature, Tc, a ferroelectric transistor has a maximum and a minimum, respectively of its transconductance and subthreshold swing, at Tc.

Related material