Increasing interest in reducing the size, cost, and limit of molecular detection has led to the use of plasmonic structures to increase signal intensity and reduce detection limits dramatically. Of particular interest is vibrational spectroscopy, a powerful mid-infrared technique for measuring the presence of molecular bonds. Vibrational spectroscopy has traditionally been hindered by its inability to directly quantify the number of molecules sampled and by it's confinement to a laboratory bench, as the equipment necessary is bulky and impractical for field studies or embedding in small instrumentation. In this thesis, I present plasmonic structures and integrated devices designed to alleviate these issues and help enable miniaturized infrared spectroscopy.

The first half of this thesis is devoted to demonstrating how aluminum plasmonic antennas can be used for quantitative surface-enhanced infrared absorption (SEIRA). Based on the self-terminating aluminum oxide layer on the surface of the antenna, a self-calibrating metric for determining the number of molecules adsorbed within the hot-spot of the antenna is proposed. This metric is based on the ratio in signal strength between the target molecule of interest and the Al2O3 reference vibration. In order to measure both the target vibration of interest as well as the Al2O3 reference, I used asymmetric antennas with plasmonic resonances at both spectral positions. Experimental measurements demonstrate that this combination of a specifically-tailored antenna design with a mechanism for determining the relative strength of the targeted molecular vibrational resonance is capable of quantitative evaluations of the number of molecules sampled. 

The second half of this work is on the design and fabrication of an integrated optoelectronic spectrometer using plasmonic aluminum gratings on p-doped silicon. By using intra-band transitions in the silicon substrate, a photoresponse can be measured as a change in resistance through the device. An array of plasmonic gratings act as both filter elements and electrical contacts, each with a distinct response spectrum. By measuring the response from the gratings under an unknown illumination spectrum and using the previously determined photoresponse spectra the input spectrum can be reconstructed. Standard least-squares reconstruction algorithms are limited by the number of gratings, however the achievable resolution of the spectrometer can be improved by adjusting the reconstruction algorithm using techniques from the compressive sensing literature. Incorporating an L1 constraint on our reconstruction algorithm to select for sparsity enables sub-Nyquist resolution in our reconstructed spectrum. The spectrometer is also used for contact-free molecular detection, demonstrating the capability to function as a "nanophotonic nose". Finally, simulations are performed to demonstrate the ultimate achievable resolution of such a spectrometer with a larger number of gratings.