Search

A material can be uniquely identified by its optical properties. Based on this principle,the Kretschmann configuration is widely used in chemical analysis to identify materials by determining optical properties of the sample material upon surface plasmon resonance.This conventional approach measures average optical properties within the diffraction-limited illumination beam spot, therefore is incapable of identifying or mapping samples that are inhomogeneous on the nanometer scale. On the other hand, near-field optical microscopy is capable of nanoscale imaging but quantitative optical property maps cannot be inferred from the obtained images. This work demonstrates the principles of surface plasmon scanning tunneling microscopy (SPSTM), a new method to map the optical properties of a material at the nanometer scale, thus providing the foundation of a novel approach towards nanoscale chemical mapping. In this demonstration of principles, surface plasmons are excited in a gold thin film that supports micrometer-size flakes of a two-dimensional (2D) layered material in some areas. The surface plamons are coupled to a gold-coated, nanometer-sized optical fiber tip, and the resulting optical signal is guided by the fiber to a detector. The nanoscale imaging capability stems from the exponential decay of the evanescent field of surface plasmon modes, resulting in the exponential decrease in plasmon coupling as the tip-substrate separation increases, as well as the sharpness of the probe tip. By fabricating Au-coated, sharp optical fiber tips and constructing an SPSTM setup, I have demonstrated exponential decrease in optical signal intensity with increasing probe height, with faster decay in the presence the 2D material flakes than on bare Au, consistent with theoretical prediction, laying down the groundwork for a future SPSTM prototype.

A broadband spectrum channelizer divides the spectrum of an input signal into sub-bands prior to baseband processing. Efficient spectrum channelizer architectures can serve as enablers for new systems and applications for communications, signal analysis and detection. Channelizers can be used to implement broadband receivers for mobile applications, that are capable of receiving signals corresponding to multiple standards, and general-purpose hybrid frequency-and-time domain analog to digital converters (ADCs). These designs can help to relax baseband design specifications, and can also be more spectrum-aware than a broadband direct-sampling ADC front-end. Circuits and architectures for the implementation of broadband spectrum channelizers are proposed in this research. First, a channelizing broadband receiver that employs a bank of two-stage harmonic rejection mixers (HRMs) is described. Each HRM internally synthesizes a distinct downconversion LO, while using a fixed high-frequency master clock. The HRMs also implement a quadrature phase matching technique for enhancing image rejection. The technique counters the impact of transistor mismatches on the phase matching between quadrature paths. This reduces the two-dimensional calibration for enhancing image performance, that requires amplitude and phase correction, into a one-dimensional problem, thereby significantly reducing calibration complexity. A prototype receiver employing this principle has been implemented in a 65nm CMOS process. The design down-converts an I/Q input of bandwidth 250 MHz to baseband and channelizes the signal concurrently into 16 sub-bands of bandwidth 15.625 MHz each. It achieves an image rejection of 56 dB with amplitude calibration alone, and a harmonic rejection of 56.5 dB without any calibration at a 1 MHz baseband output. Second, a frequency-folding channelizer architecture is proposed, which can span bandwidths of several GHz. A digital post-processing approach for reducing the anti-aliasing requirement is described. A signal of bandwidth (N/2)f [subscript LO] is downconverted into N paths that are clocked using rectangular, non-overlapping pulse waveforms at a fundamental frequency of f [subscript LO] with a duty-cycle of 1/N. All parts of the input are aliased onto a baseband signal of bandwidth f [subscript LO]/2, which is low-pass filtered and applied to an ADC, where each path employs identical sampling clocks. A digital equalization technique is presented that can be utilized to recover the complete broadband input, similar to a high-speed ADC, or recover each channelized and down-converted sub-band, similar to a bank of mixers, while performing digital-domain harmonic and image rejection. Aliasing can be reversed through equalization and this is used to reduce the order of the anti-aliasing filters in the analog signal paths, potentially reducing power requirement and design complexity. A theoretical and simulation-based analysis of channelizer performance after signal reconstruction is presented. Impact of filter-order on SNR affected by noise and impairments is also analyzed. The equalization technique is analyzed in the context of a previously reported channelizer IC. It is shown that the application of the equalization can result in 50 dB broadband SNR over 1 GHz input bandwidth, using 8 paths with 3rd-order anti-aliasing analog filters and 10-bit sub-ADCs.

A wearable device that accurately records a user's daily activities is of substantial value. It can be used to enhance medical monitoring by maintaining a diary that lists what a person was doing and for how long. The design of a wearable system to record context such as activity recognition is influenced by a combination of variables. A flexible yet systematic approach for building a software classification environment according to a set of variables is described. The integral part of the software design is the use of a unique robust classifier that uses principal component analysis (PCA) through singular value decomposition (SVD) to perform real-time activity recognition. The thesis describes the different facets of the SVD-based approach and how the classifier inputs can be modified to better differentiate between activities. This thesis presents the design and implementation of a classification environment used to perform activity detection for a wearable e-textile system.