An improved analysis of low frequency trapping noise in a MOS device is proposed. This analysis takes into account the supplementary fluctuations of the mobility induced by those of the interface charge. It enables an adequate description of the gate voltage dependence of the input equivalent gate voltage noise to be obtained in various actual situations. The outputs given by the Hooge mobility fluctuation model are also presented and discussed with respect to those obtained by the carrier number fluctuation model. In particular, the impact of the channel length or channel width, and the model type on the input gate voltage and drain current noise characteristics is studied and compared to typical experimental data. Finally, a procedure for the diagnosis of the low frequency noise sources in a MOS transistor is proposed.

No abstract

A theoretical analysis for the calculation of the conductance fluctuations in small area metal–oxide–semiconductor (MOS) devices is presented. This analysis relies essentially on the assumption that the trap region in the device channel can be approximated by a rectangular region inside which the sheet conductivity differs from that in the region of the channel outside the trap. The sheet conductivity shift in the trap region is evaluated using the concept of flat band voltage shift associated with the trapping of one elementary charge in the trap region. The approach of flat band voltage shift is then advantageously used to derive first‐order analytic expressions of the random telegraph signal (RTS) amplitude. In particular, it is found both, analytically and numerically that, in the case of small sheet conductivity modulation, the RTS amplitude becomes independent of the trap size. The numerical simulations show that the maximum RTS amplitude occurring in weak inversion is not a monotonic function of the trap size. Indeed, the normalized conductance fluctuation does increase as the trap area is reduced, passes through a maximum and finally decreases towards zero. The comparison of the simulation results with typical experimental RTS data shows an overall good agreement. In particular, it is clearly demonstrated from the RTS measurements that the gate voltage dependence of the normalized RTS amplitude can be strongly correlated to that of the transconductance to drain current ratio gm/Id.