Three main analog circuit building blocks that are important for a mixed-signal system are investigated in this work. New building blocks with emphasis on power efficiency and compatibility with deep-submicron technology are proposed and experimental results from prototype integrated circuits are presented. Firstly, a 1.1GHz, 5th order, active-LC, Butterworth wideband equalizer that controls inter-symbol interference and provides anti-alias filtering for the subsequent analog to digital converter is presented. The equalizer design is based on a new series LC resonator biquad whose power efficiency is analytically shown to be better than a conventional Gm-C biquad. A prototype equalizer is fabricated in a standard 0.18μm CMOS technology. It is experimentally verified to achieve an equalization gain programmable over a 0-23dB range, 47dB SNR and -48dB IM3 while consuming 72mW of power. This corresponds to more than 7 times improvement in power efficiency over conventional Gm-C equalizers. Secondly, a load capacitance aware compensation for 3-stage amplifiers is presented. A class-AB 16W headphone driver designed using this scheme in 130nm technology is experimentally shown to handle 1pF to 22nF capacitive load while consuming as low as 1.2mW of quiescent power. It can deliver a maximum RMS power of 20mW to the load with -84.8dB THD and 92dB peak SNR, and it occupies a small area of 0.1mm2. The power consumption is reduced by about 10 times compared to drivers that can support such a wide range of capacitive loads. Thirdly, a novel approach to design of ADC in deep-submicron technology is described. The presented technique enables the usage of time-to-digital converter (TDC) in a delta-sigma modulator in a manner that takes advantage of its high timing precision while noise-shaping the error due to its limited time resolution. A prototype ADC designed based on this deep-submicron technology friendly architecture was fabricated in a 65nm digital CMOS technology. The ADC is experimentally shown to achieve 68dB dynamic range in 20MHz signal bandwidth while consuming 10.5mW of power. It is projected to reduce power and improve speed with technology scaling.

Terahertz time-domain spectroscopy systems were developed and used for the anaylsis and characterization of various materials. By using ultra-fast Ti:Sapphire and Er-doped fiber lasers, terahertz time-domain spectrometers of different configurations were constructed and tested. To increase the accuracy and sensitivity of the measurements, the systems were optimized for spectroscopic analysis. MBE grown nominally undoped epitaxial GaAs samples were used for the spectroscopic measurements. These samples were first charactrized by electrical measurements in order to check the accuracy of the terahertz time-domain experiments. We have shown that the terahertz time-domin spectroscopic techniques provides a quick way of the determining the real ( ) and complex () components of the refractive index of material. In addition, we have investigated the photoexcitation dynamics of these GaAs samples. We have demonstrated that direct and photoexcited terahertz time-domain measurements give an estimate of the carrier densities and both the hole and electron mobility values with good precision. rnin An algorithm is developed to prevent the unwanted Fabry-Perot reflections which is commonly encountered in Terahertz Spectroscopy systems. We have performed terahertz time-domain transmission measurements on ZnTe &lt / 110&gt / crystals of various thicknesses to test the applicability of this algorithm. We have shown that the algorithm developed provides a quick way of eliminating the &ldquo / etalon&rdquo / reflections from the data. In addition, it is also shown that these &ldquo / etalon&rdquo / effects can be used for the frequency calibration of terahertz time-domain spectrometers.

Considering the communication systems with cyclic prefix (CP), such as orthogonal frequency-division multiplexing (OFDM) modulation and single-carrier cyclic prefixed (SCCP) modulation, when the length of CP is longer than the channel length, the use of cyclic prefix (CP) does not only eliminate the inter-block interference, but also convert linear convolution of the transmitted signal with the channel into circular convolution. Unfortunately, the use of CP significantly decreases the bandwidth utilization. Therefore, to reduce the length of CP is a critical issue. The thesis investigates that how to design a channel-shortening equalizer (CSE) at receiver which forces the length of the effective channel response as short as the CP length. The thesis describes the signal model as a matrix form. The effect channel response after CSE is investigated and then the coefficient of channel shortening filter is obtained using singular value decomposition method under various criterions. We further propose a novel CSE maximizing the shortening signal-to-interference ratio. In addition, it is demonstrated that the proposed CSE has the same performance as the conventional scheme but a lower computation complexity.

Στη συγκεκριμένη μεταπτυχιακή εργασία γίνεται ανάλυση των ηλεκτροεγκεφαλογραφικών καταγραφών επιληπτικών ασθενών με βάση την κλασματική διάσταση για τον εντοπισμό της έναρξης των επιληπτικών κρίσεων καθώς και για την ανίχνευση των επερχόμενων επιληπτικών κρίσεων. Πιο συγκεκριμένα, χρησιμοποιούνται τέσσερις μέθοδοι υπολογισμού της κλασματικής διάστασης στο πεδίο του χρόνου. Πραγματοποιείται μία συγκριτική μελέτη των συγκεκριμένων μεθόδων χρησιμοποιώντας δύο συνθετικά σήματα γνωστής κλασματικής διάστασης και στη συνέχεια οι μέθοδοι εφαρμόζονται σε ηλεκτροεγκεφαλογραφικές καταγραφές ασθενών που πάσχουν από επιληψία. Η συγκεκριμένη εργασία αποδεικνύει ότι όλες οι μέθοδοι μπορούν και εντοπίζουν την έναρξη των επιληπτικών κρίσεων ενώ η στατιστική ανάλυση δείχνει ότι κάποιες από αυτές τις μεθόδους μπορούν να χρησιμοποιηθούν και ως δείκτες πρόβλεψης των επερχόμενων επιληπτικών κρίσεων. Επίσης, διαπιστώνεται ότι ανάλογα με το είδος των ηλεκτροεγκεφαλογραφικών καταγραφών, το μέγεθος του παραθύρου και τη ζώνη διέλευσης του φίλτρου τα αποτελέσματα της κλασματικής διάστασης που προκύπτουν επηρεάζονται. / Τhe fractal dimension is a powerful tool in the analysis of electroencephalograms. In this work, four methods of estimating the fractal dimension of electroencephaolographic recordings of epileptic patients directly in the time domain are analyzed and compared. The analysis is performed over both synthetic data and real recordings of epileptic patients. The effects of the type of recordings, the window size and the frequencies in the passband filter are examined. This study shows that the four methods detect the beginning of the epileptic seizures, while the statistic analysis proves that some of these methods can be used for the prediction of the next seizures.

Innovation in wireless communication has resulted in accelerating demand for smartphones using multiple communications protocols such as WiFi, Bluetooth and the many cellular standards deployed around the world. The variety of frequency, bandwidth and power requirements associated with each standard typically calls for the implementation of separate radio frequency (RF) front end hardware for each standard. This is a less-than-ideal solution in terms of cost and device area. Software-defined radio (SDR) promises to solve this problem by allowing the RF hardware to be digitally reconfigurable to adapt to any wireless standard. The application of machine learning and cognition algorithms to SDR will enable cognitive radios and cognitive wireless networks, which will be able to intelligently adapt to user needs and surrounding radio spectrum conditions. The challenge of fully reconfigurable transceivers is in implementing digitally-controlled RF circuits which have comparable performance to their fixed-frequency counterparts. Switching-mode power amplifiers (SMPA) are likely to be an important part of fully reconfigurable transmitters since their switching operation provides inherent compatibility with digital circuits, with the added benefit of very high efficiency. As a step to understanding the RF requirements of high efficiency and switching PAs, an inverse class F PA in push-pull configuration is implemented. This configuration is chosen for its similarity to the current mode class D (CMCD) topology. The fabricated PA achieves a peak drain efficiency of over 75% with 42.7 dBm (18.6 W) output power at 2.46 GHz. Since SMPAs cannot directly provide the linearity required by current and future wireless communications standards, amplitude information must be encoded into the RF signal in a different way. Given the superior time resolution of digital integrated circuit (IC) technology, a logical solution is to encode this information into the timing of the signal. The two most common techniques for doing so are pulse width modulation and delta-sigma modulation. However, the design of delta-sigma modulators requires simulation as part of the design process due to the lack of closed-form relationships between modulator parameters (such as resolution and oversampling) and performance figures (such as coding efficiency and signal quality). In particular, the coding efficiency is often ignored although it is an important part of ensuring transmitter efficiency with respect to the desired signal. A study of these relationships is carried out to observe the tradeoffs between them. It is found that increasing the speed or complexity of a DS modulated system does not necessarily translate to performance benefits as one might expect. These observations can have a strong impact on design choices at the system level.

This thesis introduces spatial filtering, which is a technique to extend the time step size beyond the conventional stability limit for the Finite-Difference Time-Domain (FDTD) method, at the expense of transforming field nodes between the spatial domain and the discrete spatial-frequency domain and removing undesired spatial-frequency components at every FDTD update cycle. The spatially-filtered FDTD method is demonstrated to be almost as accurate as and more efficient than the conventional FDTD method via theories and numerical examples. Then, this thesis combines spatial filtering and an existing subgridding scheme to form the spatially-filtered subgridding scheme. The spatially-filtered subgridding scheme is more efficient than existing subgridding schemes because the former allows the time step size used in the dense mesh to be larger than the dense mesh CFL limit. However, trade-offs between accuracy and efficiency are required in complicated structures.

This thesis introduces spatial filtering, which is a technique to extend the time step size beyond the conventional stability limit for the Finite-Difference Time-Domain (FDTD) method, at the expense of transforming field nodes between the spatial domain and the discrete spatial-frequency domain and removing undesired spatial-frequency components at every FDTD update cycle. The spatially-filtered FDTD method is demonstrated to be almost as accurate as and more efficient than the conventional FDTD method via theories and numerical examples. Then, this thesis combines spatial filtering and an existing subgridding scheme to form the spatially-filtered subgridding scheme. The spatially-filtered subgridding scheme is more efficient than existing subgridding schemes because the former allows the time step size used in the dense mesh to be larger than the dense mesh CFL limit. However, trade-offs between accuracy and efficiency are required in complicated structures.

Le présent projet de maîtrise a pour but d’étudier les interactions optiques des films de nanotubes de carbone (FNTCs) avec les ondes THz. Des expériences d’absorption térahertz faites par spectroscopie THz dans le domaine temporel ont été entreprises sur les films dont l’épaisseur varie. Les films d’épaisseurs allant de 14 à 145 nm, sont des couches minces de nanotubes de carbone (NTCs) empilés les uns sur les autres et sont déposés sur substrats (GaAs et silicium). Une caractérisation comparative des épaisseurs des films est entreprise dans un premier temps par AFM et par ellipsométrie spectroscopique. À cause de la rugosité de la surface et de porosité des films qui compliquent les interactions de la lumière avec les films, les épaisseurs déterminées par AFM sont gardées au détriment de celles d’ellipsométrie. La relation entre les épaisseurs mesurées par AFM en fonction des épaisseurs nominales s’est révélée linéaire. Les couleurs des FNTC sont aussi caractérisées en fonction de leurs épaisseurs. L’expérience d’absorption THz sur les films consiste à enregistrer la transmission d’une impulsion THz à large bande à travers les échantillons. Sur les spectres, on détecte aussi l’impulsion de réflexion, l’écho de réflexion de l’impulsion principale THz à l’intérieur du substrat séparé par un délai temporel. La diminution du pic de l’impulsion principale THz en fonction de l’épaisseur est non linéaire et atteint une saturation pour les films les plus épais. Ce résultat est en lien direct avec les mesures quatre pointes de conductivité dc des films où l’inverse de la résistivité de feuille sature à partir des mêmes épaisseurs de film. L’écho de réflexion de l’impulsion principale à l’intérieur du substrat perd de l’amplitude plus rapidement en fonction de l’épaisseur à cause de près de deux passages supplémentaires de l’impulsion dans le film au moment de la réflexion. Finalement, une disparition de l’impulsion de réflexion à une épaisseur particulière de film (100 nm pour le GaAs et 60 nm pour le Si) démontre les propriétés antiréfléchissantes des FNTCs. / In the present masters project, the goal is to study the optical interactions of carbon nanotube films (CNTFs) with terahertz (THz) waves. The THz absorption experiments made by time domain THz spectroscopy have been undertaken on thickness-variable films. CNTFs, which have their thicknesses range from 14 to 145 nm, are thin CNT layers that are piled one on another are deposited on a substrate (GaAs or silicon). First, a comparative characterization of film thicknesses is undertaken with AFM and with spectroscopic ellipsometry. Because of surface rugosity and film porosity which has the effect of complexifying the interaction of light with the films, AFM thicknesses are held for the rest of the analysis instead of those determined with ellipsometry. AFM measured thicknesses scale linearly with respect to nominal thicknesses that are proportional to the CNT density. CNTFs’ colors reveal to be correlated with their thicknesses. THz absorption experiments consist of taking the transmission spectrum of a broad band THz pulse through the samples. On the spectra, we also detect the reflection pulse, which is the echo of the main THz pulse inside the substrate separated by a time delay. The decrease of the main THz pulse with respect to the film thickness is non linear and reaches a saturation plateau for the thickest films. This finding is in direct relationship with four-point probe sheet conductivity measurements made on the films where a saturation is also observed from the same thicknesses. The reflection pulse loses amplitude more rapidly as the film thickness increases because of two additional wave passages in the film during reflection. Lastly, a quenching of the reflection pulse which is observed at a particular film thickness (100 nm for GaAs and 60 nm for silicon) demonstrates antireflection properties for the CNTFs. / Les expériences de spectroscopie ont été réalisées en collaboration avec Jean-François Allard du groupe de Denis Morris de l'Université de Sherbrooke.

In 2000, a controversial paper by John Pendry surmised that a slab of negative index material could act as a perfect lens, projecting images with resolution detail beyond the limits of conventional lensing systems. A thin silver slab was his realistic suggestion for a practical near-field superlens - a 'poor-mans perfect lens'. The superlens relied on plasmonic resonances rather than negative refraction to provide imaging. This silver superlens concept was experimentally verified by the author using a novel near-field lithographic technique called Planar Lensing Lithography (PLL), an extension of a previously developed Evanescent Near-Field Optical Lithography (ENFOL) technique. This thesis covers the computational and experimental efforts to test the performance of a silver superlens using PLL, and to compare it with the results produced by ENFOL. The PLL process was developed by creating metal patterned conformable photomasks on glass coverslips and adapting them for use with an available optical exposure system. After sub-diffraction-limited ENFOL results were achieved with this system additional spacer and silver layers were deposited onto the masks to produce a near-field test platform for the silver superlens. Imaging through a silver superlens was achieved in a near-field lithography environment for sub-micron, sub-wavelength, and sub-diffraction-limited features. The performance of PLL masks with 120-, 85-, 60-, and 50-nm thick silver layers was investigated. Features on periods down to 145-nm have been imaged through a 50-nm thick silver layer into a thin photoresist using a broadband mercury arc lamp. The quality of the imaging has been improved by using 365 nm narrowband exposures, however, resolution enhancement was not achieved. Multiple layer silver superlensing has also been experimentally investigated for the first time; it was proposed that a multi-layered superlens could achieve better resolution than a single layer lens for the same total silver thickness. Using a PLL mask with two 30-nm thick silver layers gave 170-nm pitch sub-diffraction-limited resolution, while for a single layer mask with the same total thickness (60 nm) resolution was limited to a 350-nm pitch. The proposed resolution enhancement was verified, however pattern fidelity was reduced, the result of additional surface roughness. Simulation and analytical techniques have been used to investigate and understand vi ABSTRACT the enhancements and limitations of the PLL technique. A Finite-Difference Time- Domain (FDTD) tool was written to produce full-vector numerical simulations and this provided both broad- and narrowband results, allowing image quality as a function of grating period to be investigated. An analytical T-matrix method was also derived to facilitate computationally efficient performance analysis for grating transmission through PLL stacks. Both methods showed that there is a performance advantage for PLL over conventional near-field optical lithography, however, the performance of the system varies greatly with grating period. The advantages of PLL are most prominent for multi-layer lenses. The work of this thesis indicates that the utilisation of plasmonic resonances in PLL and related techniques can enhance the performance of near-field lithography.

Nowadays, silicon photonics is a widely studied research topic. Its high-index-contrast and compatibility with the complementary metal-oxide-semiconductor technology make it a promising platform for low cost high density integration. Several general problems have been brought up, including the lack of silicon active devices, the difficulty of light coupling, the polarization dependence, etc. This thesis aims to give new attempts to novel solutions for some of these problems. Both theoretical modeling and experimental work have been done. Several numerical methods are reviewed first. The semi-vectorial finite-difference mode solver in cylindrical coordinate system is developed and it is mainly used for calculating the eigenmodes of the waveguide structures employed in this thesis. The finite-difference time-domain method and beam propagation method are also used to analyze the light propagation in complex structures. The fabrication and characterization technologies are studied. The fabrication is mainly based on clean room facilities, including plasma assisted film deposition, electron beam lithography and dry etching. The vertical coupling system is mainly used for characterization in this thesis. Compared with conventional butt-coupling system, it can provide much higher coupling efficiency and larger alignment tolerance. Two novel couplers related to silicon photonic wires are studied. In order to improve the coupling efficiency of a grating coupler, a nonuniform grating is theoretically designed to maximize the overlap between the radiated light profile and the optical fiber mode. Over 60% coupling efficiency is obtained experimentally. Another coupler facilitating the light coupling between silicon photonic wires and slot waveguides is demonstrated, both theoretically and experimentally. Almost lossless coupling is achieved in experiments. Two approaches are studied to realize polarization insensitive devices based on silicon photonic wires. The first one is the use of a sandwich waveguide structure to eliminate the polarization dependent wavelength of a microring resonator. By optimizing the multilayer structure, we successfully eliminate the large birefringence in an ultrasmall ring resonator. Another approach is to use polarization diversity scheme. Two key components of the scheme are studied. An efficient polarization beam splitter based on a one-dimensional grating coupler is theoretically designed and experimentally demonstrated. This polarization beam splitter can also serve as an efficient light coupler between silicon-on-insulator waveguides and optical fibers. Over 50% coupling efficiency for both polarizations and -20dB extinction ratio between them are experimentally obtained. A compact polarization rotator based on silicon photonic wire is theoretically analyzed. 100% polarization conversion is achievable and the fabrication tolerance is relatively large by using a compensation method. A novel integration platform based on nano-epitaxial lateral overgrowth technology is investigated to realize monolithic integration of III-V materials on silicon. A silica mask is used to block the threading dislocations from the InP seed layer on silicon. Technologies such as hydride vapor phase epitaxy and chemical-mechanical polishing are developed. A thin dislocation free InP layer on silicon is obtained experimentally. / QC20100705