Abstract

This work studies the behavior of both gate-to-channel capacitance (CGC) and source-channel-drain/well leakage in metal-gate/high-κ/Ge PMOS technology (W = 10 μm and L = 10; 5; 1 μm) under development at IMEC. The hole drift-mobility of germanium is ~4X that of silicon, leading researchers to evaluate germanium as a possible channel material replacement for PMOS expected at the 32 nm technology node. In particular this study focuses on—but is not restricted to—(1) the presence of a parasitic gate-to-channel capacitance (CGC), the large non-ideal trap assisted conductance which contributes to it, and its function versus Ge-PMOS architecture and gate length; (2) the existence of C-V tool compensation error due to CGC measurement technique resulting in conductance measurement error; (3) the presence of large source-channel-drain/well leakages characterized using a new MOS gated-diode measurement technique; (4) extrinsic capacitance (CEXT), flatband voltage (VFB), and effective oxide thickness (EOT) parameter extraction with discussion on inversion layer quantization. This study found that excessive current leakages from the Ge-PMOS source-anddrain into the channel led to a chuck-dependent parasitic capacitance during CGC measurement. This excessive leakage is identified as a trap-assisted leakage through both AC and DC analysis. The chuck-dependent parasitic capacitance was an unexpected side effect of the PMOS architecture: namely the lack of N-Well isolation. The parasitic capacitance—dependent on both applied bias and frequency—was separated into two main capacitive components: a frequency-dependent source/well and drain/well trapassisted leakage capacitance (CPara_SD) and a frequency-voltage-dependent gate-induced iv junction leakage capacitance (CPara_GIJL). A third parasitic capacitance due to interface trap (IT) contribution (CIT) during channel depletion was also identified. This study also found that the new MOS gated-diode measurement technique designed to separate and evaluate the source, channel, and drain leakage components is superior to typical VGS versus IDS methods when attempting to quantify the CGC measurement. The MOS gated-diode configuration allowed for temperature-dependent analysis and activation energy extraction (EA), thereby providing a means to confirm individual leakage components: diffusion; Shockley-Read-Hall (SRH); trap-assisted leakage (TAL). TAL components include: Poole-Frenkel (PF); phonon-assisted tunneling (PAT); trap-to-band tunneling (TBT). In conclusion, it was found that the source-channel-drain/well leakages and hence parasitic capacitances of PMOS built on relaxed germanium-on-silicon can be minimized by reducing the source/drain area, reducing the source/drain-to-gate contact distance, while increasing both the gate length and measurement frequency. The dominance of SRH and TAL during Ge-PMOS operation disagrees with diffusion dominance predicted by theory and as a result opens the door for future research. Future research includes Ge- PMOS fabrication on substrates free of dislocations—to minimize SRH and TAL current leakage contributions—so as to compare leakage performance.

Library of Congress Subject Headings

Metal oxide semiconductors, Complementary--Design and construction; Electric leakage--Prevention; Integrated circuits--Very large scale integration; Germanium--Industrial applications

Publication Date

2009

Document Type

Thesis

Department, Program, or Center

Microelectronic Engineering (KGCOE)

Advisor

Rommel, Sean

Advisor/Committee Member

Moon, James

Advisor/Committee Member

Kurinec, Santosh

Comments

Note: imported from RIT’s Digital Media Library running on DSpace to RIT Scholar Works. Physical copy available through RIT's The Wallace Library at: TK7871.99.M44 K76 2008

Campus

RIT – Main Campus

Download

Share

COinS