Andrew C. Pineda

POINT DEFECTS IN Si-SiO2 SYSTEMS: CURRENT UNDERSTANDING

Perhaps no other technology developed in the 20th century plays such an important role on the daily life of today’s civilized society as microelectronics. The rapid growth experienced by complementary-metal-oxide-semiconductor (CMOS) technology since the first metal-oxide semiconductor field effect transistor (MOSFET) was realized by Kahng [1] some 40 years ago, accompanied by the advances in integrated circuit (IC) fabrication, has been revolutionizing the field of electronics. The past thirty years have also witnessed tremendous progress toward the miniaturization of CMOS devices, a trend that continues toward further downscaling of the device feature size. While miniaturization of CMOS devices has resulted in higher packing density (more devices per unit area), higher circuit speed (faster computers), and lower power dissipation, it has also created new problems and issues that need resolution for the reliability of the contemporary and future generation technology. In order to appreciate the problems and reliability issues associated with the steady downscaling of CMOS devices, a schematic design of a MOSFET is shown in Fig. 1. The top metal, generally a polycrystalline-silicon (poly-Si) acts as a gate. A thin amorphous SiO1 dielectric layer underneath the gate electrode, normally referred to as the “gate oxide”, lies above the channel regions which separates the “source” (carrier donor) and the “drain” (carrier acceptor) layers. The distance between the source and the drain under the gate dielectric is called the “channel length”. Upon biasing the gate electrode, an image charge builds up under the gate initially forming a “depletion region” and eventually at a certain voltage (called the threshold voltage, Vth) inverts the silicon surface and current starts flowing in the channel between the source and the drain. “Decreasing the feature size” of MOSFET generally means reducing the channel length. The shorter the channel length, the faster the carrier flow and the higher the drive current resulting in higher speed. Also, continued reduction in the supply voltage has lowered power consumption. Of course with the miniaturization of the device components, the primary benefit is that much larger numbers of transistors can be integrated per unit area on the wafer, thus increasing the device density in very large integrated circuits (VLSI). Such desirable features have been the driving force toward miniaturization of the MOSFET.

Self-trapping of single and paired electrons in Ge2Se3

We report the theoretical prediction of single and paired electron self-trapping in Ge(2)Se(3). In finite atomic cluster, density functional calculations, we show that excess single electrons in Ge(2)Se(3) are strongly localized around single germanium dimers. We also find that two electrons prefer to trap around the same germanium dimer, rupturing a neighboring Ge-Se bond. Localization is less robust in periodic, density functional calculations. While paired electron self-trapping is present, as shown by wavefunction localization around a distorted Ge-Ge dimer, single-electron trapping is not. This discrepancy appears to depend only on the boundary conditions and not on the exchange-correlation potential or basis set. For single- and paired-electron trapping, we report the adiabatic barriers to motion and we estimate hopping rates and freeze-in temperatures. For the single trapped electron, we also predict the (73)Ge and (77)Se hyperfine coupling constants.

The effect of network topology on proton trapping in amorphous SiO/sub 2/

We report the results of first-principles quantum chemical calculations of the interactions of H/H/sup +/ with oxide ring structures of varying sizes. The calculations suggest that the binding and stability of protons in amorphous SiO/sub 2/ depend upon the topology of the interacting network. A neutral hydrogen atom, H/sup 0/, does not bind to bridging O atoms in the rings, but may occupy voids within larger rings.

First-Principles Study of 75 As NQR in Arsenic-Chalcogenide Compounds

We present a theoretical study of the nuclear quadrupole interaction, ν(Q), of (75)As in crystalline and amorphous materials containing sulfur and selenium, and compare them with experiment. We studied a combination of hydrogen-terminated molecular clusters and periodic cells at various levels of quantum chemical theory. The results show clearly that the standard density functional theory (DFT) approximations, LDA and GGA, underestimate the nuclear quadrupole (NQR) interaction systematically, while Hartree-Fock theory overestimates it to an even greater degree. However, various levels of configuration interaction and the B3LYP hybrid exchange-correlation functional, which includes some exact exchange, give very good quantitative agreement for As bonded only to the chalcogen species. As-As bonds require highly converged basis sets. We have performed a systematic study of the effect of local distortions around an arsenic atom on ν(Q) and η. Using a simple, semiclassical model, we have combined our total energy results with our NQR calculations to predict ν(Q) lineshapes for bond angle and bond length distortions. Our predictions for lineshape, including first and second moments, are in excellent agreement with the results of Su et al for a-As(2)S(3), a-As(2)Se(3) and a-AsSe. We offer new insight into the distortions that led to this inhomogeneous broadening. Our results show clearly that, for trivalent arsenic atoms with zero or one arsenic nearest neighbor, symmetric bond stretching is the predominant contributor to the ν(Q) linewidth. However, in the presence of two arsenic nearest neighbors, distortions of the As-As-As apex angle dominates and, in fact, leads to a much larger second moment, in agreement with experiment.

Interface Effects on Total Energy Calculations for Radiation-Induced Defects

We present a new, approximate technique for estimating the polarization energy of point defects near interfaces in layered systems using semiconductor device simulation combined with a finite element quadrature technique. We show that we recapture the original, spherical Jost approximation in a homogeneous, infinite solid, as well as reproducing the exact result for a point charge near the interface of two dielectrics. We apply this technique to the silicon-silicon dioxide system for doped substrates, and for devices under bias. We show that the correction to calculated, bulk defect levels depends mildly on the distance from the interface. It depends more strongly on the substrate doping density. Finally, there is a significant dependence on gate bias. These results must be considered for proposed models for negative bias temperature instability (NBTI) that invoke tunneling from the silicon band edges into localized oxide traps.

Size-Dependence of the Linear and Nonlinear Optical Properties of GaN Nanoclusters

Abstract : In this paper, we present the results of our first-principles quantum mechanical studies of the electronic structure, geometry, and linear and nonlinear optical (NLO) properties of tetrahedral GaN atomic clusters. Our calculated results suggest that the linear and NLO properties both exhibit a strong dependence upon cluster size and shape (geometry). However, the size- and the geometry-dependences are more pronounced for the NLO properties than for the linear optical properties. For clusters containing equal numbers of Ga and N atoms, an open-structure with no network-forming ring has a much larger second-order NLO coefficient than a cluster with a closed ring structure.