Browsing by Subject "Silicon crystals"

Now showing 1 - 6 of 6

An accurate and computationally-efficient model for boron implants through an overlying oxide layer into single-crystal (100) silicon
(1993) Lim, Hsuan-Yu, 1967-; Tasch, Al F., Jr.
In many of the current implant applications in integrated circuits, a thin overlying amorphous oxide layer is often used to reduce the depth of dopant profiles by reducing the amount of channeling. The oxide layer is believed to randomly scatter a well-collimated (≤ 0.5° divergence) ion beam into a cone-shape angular distribution prior to its entry into the underlying single-crystal silicon. In this manner, most of the ions are scattered away from major axial and planar channels. A computationally-efficient 1-D model for boron implantation into single-crystal silicon through a screen oxide layer was developed. This model is of great interest and importance to the semiconductor industry for understanding process control issues in manufacturing and for guiding technology development. In developing this model, approximately 400 SIMS experimental profiles were obtained. In addition, the UT-MARLOWE Monte Carlo ion implant code was improved to generate boron profiles for part of the implant parameter space. It has been observed that boron implanted profiles are significantly dependent on the implant dose, energy, oxide layer thickness, tilt angle and rotation angle. A curve-fitting software program that uses the Dual Pearson Distribution function was used to extract the nine parameter values which define each profile. The parameters extracted for each profile are arranged into a lookup table where each set of nine parameter values corresponds to the profile for a particular combination of implant dose, energy, tilt angle, rotation angle and oxide thickness. Linear interpolation functions are employed to generate profiles for which there are no explicit set of parameters. This computationally-efficient model is able to generate as-implanted profiles of boron implantation through thin overlying oxides ranging from bare silicon (with a native oxide of approximately I.6nm) to 40nm, implant energies ranging from 15keV to 80keV, doses up to 1x10¹⁶cm⁻², tilt angles ranging from 0°-10°, and rotation angles ranging from 0°-360°. This model is being implemented into SUPREM III, a widely used process simulation program used in the semiconductor industry in order to demonstrate the model. In this manner, the model will allow users to generate, in a highly computationally-efficient manner, accurate 1-D boron profiles as a function of implant dose, energy, oxide thickness, tilt angle and rotation angle
First principles modeling of arsenic and fluorine behavior in crystalline silicon during ultrashallow junction formation
(2006) Harrison, Scott Anthony; Hwang, Gyeong S.; Edgar, Thomas F.
The 2005 International Technology Roadmap for Semiconductors predicts ultrashallow junctions (USJs) less than 7 nm deep with unprecedented dopant activation levels will be required for silicon transistors to be manufactured in 2010. To meet these requirements, it is necessary to have a better understanding of the dopant transient enhanced diffusion (TED) and clustering behaviors that undermine the achievement of these manufacturing specifications. Arsenic (As) is a commonly used n-type dopant in USJ formation and fluorine (F) is an impurity commonly co-implanted with dopants to reduce dopant diffusion and clustering during USJ formation. In this dissertation, density functional theory within the generalized gradient approximation is used to understand the behavior of As and F in crystalline silicon during USJ formation. In the first part of this dissertation, the influence of silicon interstitials on As behavior during thermal annealing that follows dopant implantation is investigated. As a result of dopant implantation, a net excess of silicon interstitial defects exist in the silicon. First, it is shown that silicon interstitials can easily annihilate existing Asvacancy complexes in silicon with negligible recombination energy barriers. Second, experimentally observed As TED mediated by interstitials is explained by the formation of a highly mobile As-silicon interstitial pair that can exist in positive, neutral, and negative charge states. Finally, it is shown that large As-silicon interstitial complexes may form when excess interstitials are present and provide a kinetic route to As clustering that leads to As deactivation. In the second part of this dissertation, the interaction of F impurities with silicon interstitials and B dopants is investigated. First, the formation and diffusion of a highly mobile fluorine-silicon interstitial pair which has been suggested by experiment is detailed. Second, an immobile B-Sii-F structure is identified in which B has a deactivated configuration. This structure may play a role in deactivating and immobilizing B when implanted B and F profiles coincide. This research provides fundamental insight into the behavior of As dopants and F impurities during USJ formation. As the future of silicon-based devices relies on the ability to perform precise doping, these findings should be of great importance to device manufacturers.
Monte Carlo simulation of arsenic ion implantation into single- crystal silicon
(1995-12) Yang, Shyh-horng; Not available
Second harmonic spectroscopy of silicon nanocrystals
(2007) Figliozzi, Peter Christopher, 1972-; Downer, Michael Coffin
Using a novel two-beam technique developed to greatly enhance quadrupolar contributions to the second-order nonlinear polarization, we performed a nonlinear spectroscopic study of silicon nanocrystals implanted in an SiO₂ matrix.
Selective silicon and germanium nanoparticle deposition on amorphous surfaces
(2007-08) Coffee, Shawn Stephen, 1978-; Ekerdt, John G.
This dissertation describes the development of a process for the precise positioning of semiconductor nanoparticles grown by hot wire chemical vapor deposition and thermal chemical vapor deposition on amorphous dielectrics, and it presents two studies that demonstrate the process. The studies entailed growth and characterization using surface science techniques and scanning electron microscopy. The two systems, Ge nanoparticles on HfO₂ and Si nanoparticles on Si₃N₄, are of interest because their electronic properties show potential in flash memory devices. The positioning technique resulted in nanoparticles deposited within 20 nm diameter feature arrays having a 6x10¹⁰ cm⁻² feature density. Self-assembling diblock copolymer poly(styrene-b-methyl methacrylate) thin films served as the patterning soft mask. The diblock copolymer features were transferred using a CHF₃/O₂ reactive ion etch chemistry into a thin film SiO₂ hard mask to expose the desired HfO₂ or Si₃N₄ deposition surface underneath. Selective deposition upon exposed pore bottoms was performed at conditions where adatom accumulation occurred on the HfO₂ or Si₃N₄ surfaces and not upon the SiO₂ mask template. The selective deposition temperatures for the Ge/HfO₂ and Si/Si₃N₄ systems were 700 to 800 K and 900 to 1025 K, respectively. Germanium nucleation on HfO₂ is limited from hot wire chemical vapor deposition by depositing nanoparticles within 67% of the available features. Unity filling of features with Ge nanoparticles was achieved using room temperature adatom seeding before deposition. Nanoparticle shape and size are regulated through the Ge interactions with the SiO₂ feature sidewalls with the adatom removal rate from the features being a function of temperature. The SiO₂ mask limited Ge nanoparticle growth laterally to within ~5 nm of the hard mask at 800 K. Silicon deposition on patterned Si₃N₄ has multiple nanoparticles, up to four, within individual 20 nm features resulting from the highly reactive Si₃N₄ deposition surface. Silicon nucleation and continued nanoparticle growth is a linear function of deposition flux and an inverse function of sample temperature. Diblock copolymer organization can be directed into continuous crystalline domains having ordered minority phases in a process known as graphoepitaxy. In graphoepitaxy forced alignment within microscopic features occurs provided certain dimensional constraints are satisfied. Graphoepitaxy was attempted to precisely locate 20 nm diameter features for selective Ge or Si deposition and initial studies are presented. In addition to precise nanoparticle positioning studies, kinetic studies were performed using the Ge/HfO₂ material system. Germanium hot wire chemical vapor deposition on unpatterned HfO₂ surfaces was interpreted within the mathematical framework of mean-field nucleation theory. A critical cluster size of zero and critical cluster activation energy of 0.4 to 0.6 eV were estimated. Restricting HfO₂ deposition area to a 200 nm to 100 [mu]m feature-width range using SiO₂ decreases nanoparticle density compared to unpatterned surfaces. The studies reveal the activation energies for surface diffusion, nucleation, and Ge etching of SiO₂ are similar in magnitude. Comparable activation energies for Ge desorption, surface diffusion and cluster formation obscure the change with temperature an individual process rate has on nanoparticle growth characteristics as the feature size changes.
A universal species ion implantation model for implants into topographically complex structures with multiple materials
(2001-08) Chen, Yang, 1973-; Tasch, Al F.
A physically-based model for ion implantation of any species into single crystal silicon and a few amorphous materials has been developed, tested and implemented in ion implant simulator UT-MARLOWE. In this model, an interpolation scheme, based on mathematical properties of ion-target interatomic potential, was employed and implemented to calculate the scattering process. Using this scheme, the resulting energy, direction and momentum of the ion and target can be derived from the existing scattering tables of UT-MARLOWE without calculating the whole scattering process. The method has advantages in both accuracy and computational efficiency, as well as significantly reduced cost of code development. The impurity profiles and damage profiles predicted by the model simulations have been compared with SIMS and RBS, and excellent agreement with experimental data has been achieved.