Back to search

Aristotle University Of Thessaloniki (AUTH)

Low frequency noise in advanced CMOS/SOI nanoscale multi-gate devices and noise models for applications in electronic circuits

Abstract

dc:description

The continuous electronic device scaling forced by the need for smaller and faster circuits led to the development of new structure approaches, such as the multi-gate transistors. Among the most common solutions for high performance nano-scale area devices are the fin-shaped field-effect transistors (FinFETs) and the fully-depleted silicon-on-insulator MOSFETs, which can provide a high immunity to the short-channel effects (SCEs), low threshold voltage variability and an improved drain-induced barrier level (DIBL). Furthermore, advanced single-gate or double-gate thin-film transistors (TFTs) of amorphous metal oxides or polycrystalline silicon are key building blocks for active-matrix-driven flat panel displays (FPDs), supporting a variety of functions beyond pixel switching in FPD applications. However, an issue that has not been yet fully resolved and has become very crucial in nano-scale technology devices is the low-frequency noise (LFN) of the drain current. The LFN is a major subject in nowdays semiconductor device research, because of its dependence on the oxide geometry: the noise level is inversely proportional to the device area. Thus, the performance of both analog and digital circuits consisting of nano-scale transistors may be dramatically affected by noise and affect their functionality. Furthermore, in very small area devices new types of noise introduce high levels of fluctuations and further performance risks can appear.The present thesis is making an effort to deal with the problem of the LFN in modern technology MOSFETs, through an analysis based on both experimental and simulation data of advanced multi-gate devices. Emphasis is given on the co-existence of more than one silicon/insulator interfaces and the impact on the LFN levels and behavior. In addition, all types of observed noise are analyzed, in order to define the origin of the current fluctuations in the most accurate way. Moreover, the LFN is used as a diagnostic tool to characterize the interface quality of advanced transistors. Finally, our research results are implemented for the development of a Verilog-A model used in circuit simulations.First, the LFN analysis of amorphous Indium-Gallium-Zinc Oxide bottom-gate (α-IGZO) TFTs is presented, where it is shown that the drain current flicker noise can be sufficiently described by the carrier number fluctuations (CNF) model, whereas the generation-recombination (g-r) noise is originating from traps located in a thin depletion layer of the Si film. The noise behavior of double-gate poly-Si TFTs is also studied, showing an agreement with the carrier number fluctuations with correlated mobility fluctuations (CNF/CMF) approach regarding the 1/f noise, while some interface traps with uniform energy distribution are creating g-r noise. In addition, low-frequency measurements were performed in DG poly-Si TFTs after electrical stress, to identify the stress-induced degradation mechanisms. Aside from the importance of the qualitative results helping the research and development engineers for better understanding and improving the device quality, a 1/f noise model for TFT circuit simulations can be easily implemented based on the equations supporting the experimentally validated CNF/CMF approach.Second, we present an analysis of the current noise behavior in tri-gate n-channel FinFETs. While the CNF/CMF model can describe the flicker noise behavior, the total normalized noise level seems to be dependent on the fin width and length: the product of the Coulomb scattering coefficient αsc and the effective mobility μeff is current independent however it is increasing exponentially with the decrease of the fin width. Regarding the g-r noise observed in narrow-fin FinFETs, it is attributed to traps in the sidewall areas and interfaces, whereas short-channel FinFETs exhibit g-r noise originating form traps near the drain and source contacts. The successful description of the flicker noise behavior by the CNF/CMF model and furthermore the experimentally observed behavior of the CMF coefficient can lead to a direct application in circuit simulations.Following the FinFET LFN study, the current fluctuations origins observed in FDSOI MOSFETs are discussed, covering a major part of the thesis. We present a new approach of the CNF/CMF model, taking into account the dependence of the Coulomb scattering coefficient αsc on the carrier position in the channel, which is defined by the bias conditions. In this way, we explain the flicker noise increase observed when applying a negative voltage on the bottom-gate (BOX). Moreover, we developed a method for determining the trap densities of both front and back interfaces using different bias conditions measurements. Using this method, we confirmed that the trap density of midgap/high-k metal gate stacks is much higher than the typical SiO2 gate oxides. The CMF factor was found to increase with decreasing the channel width, which is in agreement with the CMF behavior of FinFETs. A low-frequency dominance of g-r noise was observed in ultra-thin body and BOX (UTBB) FDSOI MOSFETs and a clear dependence on the transconductance and channel width is shown. The traps creating the g-r noise are located at the interfaces and in thin layers of the body, near the drain and source contacts. As a final step of the FDSOI MOSFET LFN analysis, a comparison between three FDSOI technologies with different body and BOX thicknesses is presented. In addition to the multi-interface CNF/CMF approach to predict the 1/f noise level, the main conclusions of the above comparison are: (i) the CMF flicker noise coefficient is significantly affected by the channel cross-section area W•tSi, where W the channel width and tSi the silicon thickness and (ii) the trap density of the BOX is increased with decreasing its thickness.The last part of our research work concerns the application of the developed noise approaches in circuit simulations. To achieve this, Verilog-A code is used to develop single device modules based either on experimental/simulation data or on analytical expressions. In both cases, the noise behavior of the single device simulation output is first validated, thus giving the ability to perform circuit simulations. The output noise of a FinFET CMOS inverter can be described using the threshold voltage fluctuations approach. The phase noise of a FinFET CMOS ring oscillator near the carrier frequency is caused by the up-conversion of the 1/f noise and it can be predicted by applying the CNF/CMF model in the phase noise mathematical expression. The noise behavior of a FDSOI CMOS inverter is also presented. It is shown that the back-biasing of the FDSOI device can dramatically affect both the DC response and the output voltage noise levels.In conclusion, we believe that the qualitative results of our study can function as a guideline for new experimental work on advanced device noise and easily be implemented in circuit simulation processes.

Degree

thesis:*
Grantor dc:publisher
Aristotle University Of Thessaloniki (AUTH)
Year dc:date
2013

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Theodorou, Christoforos
  • Θεοδώρου, Χριστόφορος

Subjects

dc:subject × 14

Rights

dc:rights
Statement dc:rights
  • BY_NC
Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/36981

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Theodorou, Christoforos; Θεοδώρου, Χριστόφορος. Low frequency noise in advanced CMOS/SOI nanoscale multi-gate devices and noise models for applications in electronic circuits. Aristotle University Of Thessaloniki (AUTH), 2013. http://hdl.handle.net/10442/hedi/36981