This thesis investigates the fast strain response of electrical transport in GaAs-based semiconductor device structures (~ GHz). The transport was either across a semiconductor junction, or parallel or perpendicular to the plane of a quantum well (QW). Picosecond strain pulses were generated by thermalisation of femtosecond optical pump pulses in a metal transducer film adhered to the device's back surface. The QW-based device experiments employed picosecond strain pulses to modulate the photocurrent generated in a QW, excited by femtosecond optical pulses. Two devices were investigated: a p-i-n diode containing a QW in the intrinsic region (Q-p-i-n), and a device based on planar-transport across two QWs. The modelled photocurrent response of the Q-p-i-n diode was based on strain-induced changes in the QW's optical absorption. The predicted and the observed responses had the same order of magnitude. These experiments show that the photo carrier populations in a QW, which may be probed with picosecond resolution, are highly sensitive (fractional change in photocurrent/strain ~ 600) to sub-terahertz acoustic wave packets - several orders of magnitude greater than achievable with conventional pump probe techniques (reflectivity change/strain ~ 30). The junction-based experiments employed picosecond strain pulses to control device current on fast timescales. These devices were probed electrically using high-speed detection electronics (rise time ~ 30 ps). Three device structures were examined: p-n and Schottky diodes and a metal-semiconductor field effect transistor. The models attributed the changes in current to strain-induced, localised shifts in the semiconductor band structure. For the p-n and Schottky diodes, the predicted current response was within an order of magnitude of experimental values. These junction-based devices, operable at room temperature, also respond highly sensitively and on fast timescales (~ GHz) to sub-terahertz acoustic wave packets. Further development could potentially extend this to ultrafast timescales (~ THz). ·

CMOS scaling is rapidly reaching physical limits, forcing the industry to consider alternative routes to realise performance improvements. Strain-engineering is one such option and is already widely exploited in order to improve charge transport in the device channel. Almost every leading chipmaker has announced their version of strain-engineered CMOS and strain is forecast to play a major role in future device generations. The effects of strain-engineering are certainly not restricted to the device channel. For example, the heavily doped source/drain extension regions are subjected to high levels of strain that can play a major role in determining dopant activation, dopant diffusion and carrier mobility. A vigorous research effort is underway in order that highly active doping concentrations with minimal diffusion and good charge mobility are achieved. Since conventional dopant solubility limits already restrict dopant activation, scientists employ innovative methods to try to heighten doping ceilings. Such research has tended to focus on ordinary, unstrained silicon and up until now, no comprehensive experimental research has been carried out to address the subject of heavy doping in the context of strain-engineering. In this thesis, a detailed study into the effects of tensile strain on impurity activation and carrier mobility has been carried out for n-type dopants in silicon. Significant stress effects have been uncovered including a 30% strain-induced electron mobility enhancement to the universal mobility curve. This improves the resistance characteristic of heavily n-doped regions by 30%. For arsenic, tensile strain is shown to have little effect on dopant activation. Since tensile strain also increases arsenic diffusion, the effects of strain on arsenic doping are mixed since higher mobility is annulled by increased arsenic diffusion. On the other hand, tensile strain is shown to positively affect antimony doping in strained silicon. In addition to reducing antimony diffusion, modest amounts of tensile strain increase both electron mobility by 30% and increase antimony activation by a factor of 2 or more. These effects combine to create antimony ultrashallow junctions in strained silicon with a sheet resistance of < 600&ohm;/Sq, a junction depth of 10nm and junction abruptness of 2mn/decade - satisfying the requirements of the semiconductor technology roadmap for upcoming technology nodes.

Alloys of silicon (Si), germanium (Ge) and tin (Sn) are continuously attracting research attention as possible direct band gap semiconductors with prospective applications in optoelectronics. The direct gap property may be brought about by the alloy composition alone or combined with the influence of strain, when an alloy layer is grown on a virtual substrate of different composition. Si-GeSn nanostructures are also promising materials because they are compatible with Si-based technology, and have a high potential in many optoelectronic applications, such as silicon-based Ge/SiGeSn band-to-band and inter-subband lasers. In search for direct gap materials, the electronic structure of relaxed or strained Gel-xSnx and Si1-xSnx alloys, and of strained Ge grown on relaxed Gel_x_ySixSny, were calculated by the self-consistent pseudo-potential plane wave method, within the mixed-atom supercell model of alloys, which was found to offer a much better accuracy than the virtual crystal approximation. Expressions are given for the direct and indirect band gaps in relaxed Gel-xSnx, strained Ge grown on relaxed SixGel-x_ySny, and for strained Gel-xSnx grown on a relaxed Gel_ySny substrate, and these constitute the criteria for achieving a direct band gap semiconductor, by using appropriate tensile strain. In particular, strained Ge on relaxed SixGel_x_ySny has a direct gap for y > 0.12 + 0.20x, while strained Gel-xSnx on relaxed Gel_ySny has a direct gap for y > 3.2x2 - 0.07x + 0.09. In contrast, within the mixed-atom approach the SnxSi1- x alloys never show a finite direct band gap (while the VCA calculation does predict it). Self-assembled quantum dots in Si-Ge-Sn system attract research attention as possible direct band gap materials, compatible with Si-based technology, with potential applications in optoelectronics. In this work, the electronic structure near the f-point and interband optical matrix elements of strained Sn and SnGe quantum dots in Si or Ge matrix are calculated using the eightband k· p method, and the competing L-valley conduction band states were found by the effective mass method. The strain distribution in the dots was found with the continuum mechanical model. The parameters required for the k· p or effective mass calculation for Sn were extracted by fitting to the energy band structure calculated by the nonlocal empirical pseudopotential method (EPM). The calculations show that the self-assembled Sn/Si dots, sized between 4 nm and 12 nm, have indirect interband transition energies between 0.8 to 0.4 eV and direct interband transitions between 2.5 to 2.0 eV. In particular, the actually grown, approximately cylindrical Sn dots in Si with a diameter and height of about 5 nm are calculated to have an indirect transition (to the L valley) of about 0.7 eV, which agrees very well with experimental results. Similar good agreement with experiment was also found for SnGe dots grown on Si. However, neither of these are predicted to be direct band gap materials, in contrast to some earlier expectations. In order to extend a creativity in developing a complete suite of Si-base optoelectronic devices, SiGeSn alloys are considered as promising materials for optoelectronic applications because they offer the possibility for a direct band gap and are compatible with Si-based technology, therefore having a perspective of applications for interband lasers and detectors, solar cells, etc. In this work, another possible application of nanostructures based on these materials was considered: to extend the suite of Si-based optoelectronic devices, namely for interband electro-absorption modulators. Using the 8-band k.p method asymmetric double quantum wells have been designed and optimized, by varying the well and barrier widths and material composition, to show large optical transmission sensitivity to the applied bias. Generally, these structures are useful for electro-absorption modulators in the mid-infrared spectral range.

The work described in this thesis concerns studies of damage and annealing processes in ion implanted Si, relevant for the formation of source / drain extensions in sub 100 nm CMOS devices. Implants were carried out using 1-3 keV As, BF2 and Sb ions and implanted samples were annealed at temperatures between 550 °C and 1130 °C. The principal analysis technique used was Medium energy ion scattering (MEIS), which yields quantitative depth profiles of displaced Si atoms and implanted dopants. The results obtained have been related to comparative analyses using SIMS, TEM and X-ray techniques. Heavy ion damage evolution and the concomitant dopant redistribution as a function of ion dose was investigated using As and Sb implantation into Si. It was found that for low doses the damage build up does not follow the energy deposition function. Instead a ~4 nm wide amorphous layer is formed initially under the oxide that grows inwards into the bulk with increasing dose. For low doses As is seen to have migrated into the damaged regions near the surface, where it appears to be more readily accommodated. Both effects are ascribed to the migration of interstitials. Various annealing studies have been carried out to investigate the regrowth behaviour of the damaged Si and the redistribution of the dopant. Effects of recrystallisation, dopant movement into substitutional positions, dopant segregation and diffusion are observed. Annealing studies of implanted Silicon on Insulator (SOI) wafers have shown a regrowth that has a wavy a/c interface unlike the layer by layer mode that is typical in solid phase epitaxial regrowth. This effect is ascribed to localised damage accumulation at the buried oxide layer. Following a BF2 implant into Si, pre-amorphised by a Xe bombardment, an interaction between Xe, F and B has been observed to occur upon annealing during which the implanted species conglomerate at depths related to the end of range of the BF2 implant in the amorphous Si.

No description available.

This thesis describes an investigation by computer simulation of the performance of two semiconductor device types at the heart of optical high speed data communications, namely the PIN photodiode and the electroabsorption modulator. Both device types operate by light absorption and are therefore likely to have similar factors that limit their performance at high speed. In order to have high speed detection, the PIN photodiode has been investigated through varying the materials, and used a photodiode structure to improve its bandwidth. If output signals beyond the 3dB frequency limit can be well detected by the photodiode, then significant improvements in the detection speed can be achieved. This possibility is a motivation of this thesis. In this study the InP /InGaAs/InP PIN photodiode is chosen because the light at 1.55 urn wavelength can be absorbed by InGaAs. At 1.55J.lm, the fibre is on low dispersion and low loss. A numerical model of a PIN photodiode has been written in C. Comprehensive modelling ofthe PIN photodiode requires a self-consistent solution of the Poisson's equation for calculating the electrostatic potential and the continuity equations for the electron and hole currents. The PIN photo diode model has included the therm ionic current over the hetero-junction, drift current and diffusion current that other models often ignore. After completing an extensive study of the large signal performance of PIN photodiodes at data rates much higher than the conventional 3dB bandwidth, the model was extended to investigate InP/InGaAsP/lnGaAs MQW -EAM under high speed applied bias pulses. The numerical modelling of the MQW-EAM requires a self-consistent solution of the Poisson's equation, the Schrcdinger 's equation and the current continuity equation. The Schrodinger 's equation is for the estimation of carrier concentration in the quantum wells. The MQW-EAM numerical model has applied a special technique for adding the carrier concentration in the quantum wells to the charge density. 11 Large performance has been successfully analysed on PIN photodiode to reveal that optical pulses at repetition frequencies are substantially higher than the conventional 3dB limit can detect (1 ps FWHM and up to 240Gb/s repetition rate) to give photocurrent pulses with an open eye diagram even in the presence of simulated noise, however, these output current pulses tend to spread and merge together sometimes. This tendency can be counteracted to a reasonable extent by using a suitable repetition time, and a large input average power. Similar Gaussian shaped applied bias pulses have also applied to the MQW-EAM, in order to generate fast Gaussian shaped light power, however; the output light pulses spread out nearly 3 times compared with its applied bias pulses under 5ps FWHM and 48Gb/s repetition rate. Thinner l-layer and less quantum wells in the MQW-EAM might be the solution. iii