No description available.

This thesis describes a divide-and-conquer algorithm to improve the 3D boundary element method (BEM) for capacitance extraction. We divide large interconnect structures into small sections, set new boundary conditions using the borderfor each section, solve each section, and then combine the results to derive the capacitance. The target application is critical nets where 3D accuracy is required. The new algorithm is a significant improvement over the traditional BEMs and their enhancements, such as the "window" method where conductors far away are dropped, and the "shield" method where conductors hidden behind other conductors are dropped. Experimental results show that our algorithm is 25 times faster than the traditional BEM and 5 times faster than the window+shield method, for medium to large structures. The error of the capacitance computed by the new algorithm is within 2% for self capacitance and 7% for coupling capacitance, compared with the results obtained by solving the entire system using BEM. Furthermore, our algorithms gives accurate distributed RC, where none of the previous 3D BEM algorithms and their enhancements can.

In this work, electrical mechanism of Low Temperature Poly-Si Thin-Film Transistors (LTPS TFTs) and Nonvolatile memory TFTs was investigated. First, relationship between trap states in grain boundary and capacitance-voltage (C-V) transfer characteristic curve would be discussed. The experimental results reveal that the C-V curves were a function with the trap state distribution and the measured frequency. The threshold voltage was increased with increasing measured frequency and temperature. Besides, anomalous capacitance was generated in p-channel LTPS TFTs when the device was operated at off-state. In general, the effective capacitance of the LTPS TFTs was only dependent with the overlap area between gate and source/drain under the off-state. However, the experimental results reveal that the off-state capacitance was increased with decreasing measured frequency and/or with increasing measurement temperature. By fitting the curve of drain current versus electric field under off-state region, it was verified that the TAGIDL is consisted of the Pool-Frenkel emission and thermal field emission. In addition, the charge density calculated from the Cch-Vg measurement also the same dependence with electric field. This result demonstrates that the anomalous capacitance is mainly due to the trap-assisted-gate-induced-drain-leakage (TAGIDL). In order to suppress the anomalous capacitance, a band-to-band hot electron (BTBHE) stress was utilized to reduce the vertical electric field between the gate and the drain. The electric field simulation was also performed by ISE-TCAD software. In addition, the degradation mechanism in Nonvolatile memory TFTs under DC stress was discussed. The gate insulator of the Nonvolatile memory TFTs was stacked with oxide-nitride-oxide and the thickness was 40nm-20nm-10nm, respectively. The polarities of the gate insulator were included fresh state, programmed state and erased state. In order to compare the ONO with the STD TFTs, the STD TFTs was also discussed with the same DC stress condition. The experimental results reveal that the degradation phenomenon was not only oxide trapping (Nox) but alsointerface trap (Nit). Besides, the simulation software ISE-TCAD was used to demonstrate these results. This main degradation phenomenon was caused by carrier injecting into oxide which was due to the coulomb force. The Nox and Nit were increased while carrier injected into the gate oxide. On the other hand, there were showed identical degradation mechanism in fresh state and erase state SONOS TFTs under the positive gate bias stress, but in which were not consistent with the program state. In program state, the vertical electric field was released due to trapping electrons in nitride. Therefore, the electric property would slightly improve during the positive gate bias stress and the main degradation mechanism was become to the carrier detrapped from nitride to gate terminal. Beside, the off state C-V curve was slightly increased under the positive gate bias stress in program state. This result was contributed to the electrons trapped in the oxide near the gate insulator edge cause by the electric field corner effect. And then, the electric field corner effect was also verified by the simulation software ISE-TCAD. Finally, the TAGIDL in fresh and erase states is increased with increasing the stress time. On contrary, the situation in program state is decreased with increasing the stress time. These results are contributed to a large number electrons injection into the overlapped insulator region between the gate and S/D and enhancing the band bending in the overlapped region when SONOS TFT is operated at fresh and erase states. However, in program state, the electrons trapped in the nitride are flowed to the gate due to the positive bias.

This thesis describes a divide-and-conquer algorithm to improve the 3D boundary element method (BEM) for capacitance extraction. We divide large interconnect structures into small sections, set new boundary conditions using the borderfor each section, solve each section, and then combine the results to derive the capacitance. The target application is critical nets where 3D accuracy is required. The new algorithm is a significant improvement over the traditional BEMs and their enhancements, such as the "window" method where conductors far away are dropped, and the "shield" method where conductors hidden behind other conductors are dropped. Experimental results show that our algorithm is 25 times faster than the traditional BEM and 5 times faster than the window+shield method, for medium to large structures. The error of the capacitance computed by the new algorithm is within 2% for self capacitance and 7% for coupling capacitance, compared with the results obtained by solving the entire system using BEM. Furthermore, our algorithms gives accurate distributed RC, where none of the previous 3D BEM algorithms and their enhancements can.

Thesis (MTech (Mechanical Engineering))--Cape Peninsula University of Technology, 2005. / Word processed copy. Summary in English and Chinese. Includes bibliographical references. Also available online.

No description available.

There is a need for a high voltage, capacitance voltage (HV, CV) measurement system for the measurement and characterization of silicon carbide (SiC) power MOSFETs. The following study discusses the circuit layout and automation software for a measurement system that can perform CV measurements for all three MOSFET capacitances, CGS, CDS, and CGD. This measurement system can perform low voltage (0–40V) and high voltage (40–5kV) measurements. Accuracy of the measurement system can be safely and effectively adjusted based on the magnitude of the MOSFET capacitance. An IRF1010N power MOSFET, a CoolMos, and a prototype SiC power MOSFET are all measured and their results are included in this study. All of the results for the IRF1010N and the CoolMos can be verified with established characteristics of power MOSFET capacitance. Results for the SiC power MOSFET prove that more testing and further development of SiC MOSFET fabrication is needed. / Master of Science

Transient photocapacitance (TPC) and transient photocurrent (TPI) spectroscopy have been used to examine the defect structure in the upper-half of the bandgap of CdTe solar cells, with an emphasis on understanding the effect of copper. TPC spectra reveal two defects in the CdTe devices at optical energies of 1.2eV and 0.9eV with respect to the valence band. The origin of the 1.2eV defect could not be associated with a particular element, although copper and zinc were ruled out as sources. TPI spectra were used to observe that the density of the 1.2eV defect was dramatically reduced by thermally annealing the devices, suggesting that the defect itself is annealed during the treatment. The set of CdTe samples examined used a rapid thermal processing treatment to control the amount of copper that diffused into the CdTe layer from the Cu:ZnTe interfacial layer at the back of the device. Comparison of devices with varying amounts of copper in the CdTe layer revealed that the 0.9eV defect seen in TPC was associated with the presence of copper in the absorber layer. TPI spectra confirmed the association of the 0.9eV with copper and showed that the magnitude of the 0.9eV defect signal increased as more copper was diffused into the CdTe layer. A proportional link between the density of the 0.9eV defect observed in TPI spectra and the amount of copper in the absorber layer observed via ToF-SIMS further established that copper is responsible for the existence of the defect. Numerical modeling of the CdTe devices was used to confirm that the spatial distribution of copper observed in ToF-SIMS was consistent with the relative variation of defect magnitudes observed in TPI. The fact that the copper-associated 0.9eV defect lies close to mid-gap suggests that it will act as an efficient recombination center in CdTe. Therefore, it is suggested that this work has detected the deep defect that is responsible for the decreased minority carrier lifetime that has been previously associated with the amount of copper in the CdTe layer

An all-electrical system is developed to actuate and detect single biological cells in a microfluidic channel for diagnostic applications. Interdigitated electrodes fabricated on the channel floor transfer a high frequency signal for capacitance detection and a low frequency signal for dielectrophoretic actuation. In the fluid-filled channel, a pressure-driven flow propels single biological cells, which induce time-dependent capacitance signatures as they pass over the electrodes. With a sub-attofarad (~0.15 aF RMS, 53 Hz bandwidth) capacitance resolution, this system detects biological cells (e.g., 1 yeast cell ~ 50 aF) and their deflections (1 micrometer ~ 5 aF) from exerted dielectrophoretic forces (> 5 pN). Electrical detection of cell actuation by strong DEP forces provides an avenue for both inducing and monitoring the deformation of viscoelastic cells. A strong and repulsive dielectrophoretic force can be used to press a biological cell into a channel wall. When this occurs, the mechanical properties of the cell can be investigated by capacitively monitoring the cell-to-wall interaction. The nature of the resulting interaction is shown to depend on the mechanical properties of the cell (surface morphology and viscoelastic properties). Various mammalian cell types such as Chinese Hamster Ovary (CHO) cells, mouse fibroblasts, human blood cells, human breast cells and their tumorogenic phenotypes are investigated using this system. Between these populations, the effective Young's modulus varies widely from 20 Pa (neutrophils) to 1-2 GPa (polystyrene microspheres). The viability and phenotype of a biological cell are known to reflect its mechanical and electrical properties. Consequently, this work investigates whether dielectrophoretically induced cell deformations are correlated with corresponding variations in capacitance, which could be used for discriminating cell phenotypes in the future.

An all-electrical system is developed to actuate and detect single biological cells in a microfluidic channel for diagnostic applications. Interdigitated electrodes fabricated on the channel floor transfer a high frequency signal for capacitance detection and a low frequency signal for dielectrophoretic actuation. In the fluid-filled channel, a pressure-driven flow propels single biological cells, which induce time-dependent capacitance signatures as they pass over the electrodes. With a sub-attofarad (~0.15 aF RMS, 53 Hz bandwidth) capacitance resolution, this system detects biological cells (e.g., 1 yeast cell ~ 50 aF) and their deflections (1 micrometer ~ 5 aF) from exerted dielectrophoretic forces (> 5 pN). Electrical detection of cell actuation by strong DEP forces provides an avenue for both inducing and monitoring the deformation of viscoelastic cells. A strong and repulsive dielectrophoretic force can be used to press a biological cell into a channel wall. When this occurs, the mechanical properties of the cell can be investigated by capacitively monitoring the cell-to-wall interaction. The nature of the resulting interaction is shown to depend on the mechanical properties of the cell (surface morphology and viscoelastic properties). Various mammalian cell types such as Chinese Hamster Ovary (CHO) cells, mouse fibroblasts, human blood cells, human breast cells and their tumorogenic phenotypes are investigated using this system. Between these populations, the effective Young's modulus varies widely from 20 Pa (neutrophils) to 1-2 GPa (polystyrene microspheres). The viability and phenotype of a biological cell are known to reflect its mechanical and electrical properties. Consequently, this work investigates whether dielectrophoretically induced cell deformations are correlated with corresponding variations in capacitance, which could be used for discriminating cell phenotypes in the future.