Date

2022-2-9

Author

Yolalmaz, Alim

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

448
views

569
downloads

Cite This

Spectral and spatial control of light strongly influences central fields such as solar energy, spectroscopy, holography, and imaging. Enhanced control of propagation, diffraction, scattering, and interference of light using micro-and nano-structures leads to superior performance in these fields. This thesis presents light control in spectral and spatial domains using wavefront control for solar energy, spectroscopy, holography, and imaging applications. Spectral splitting of the sunlight using diffractive optical elements (DOEs) is an effective method to increase the amount of converted solar energy. In this thesis, we design phase-only DOEs by using an iterative optimization algorithm to spectrally split and simultaneously concentrate the solar spectrum. We introduce the effective bandwidth approach, which reduces the computational time of DOEs from 89 days to 8 days while preserving the spectral splitting efficiency. Using our effective bandwidth method, we manage to spectrally split light into two separate bands between 400 nm - 700 nm and 701 nm - 1100 nm, with splitting efficiencies of 56% and 63%, respectively. Next, we present a hybrid design scheme, which relies on a deep learning model, the DOENet, and the local search optimization algorithm, to optimize a DOE that performs spectral splitting and spatial concentration of broadband light for solar cells. Our hybrid design approach both speeds up optimization of DOEs as well as provides better performance, at least 57% excess light concentration with spectral splitting. Here, we also design DOEs that concentrate and split the broadband light for angled illumination to minimize the variation of intensity on the targets when the angle of incident light changes leading to a decrease in the intensity of light for solar cells applications. We observe that spectral splitting of the broadband light with a DOE is less sensitive to variation of incident angle of the solar radiation once the DOE optimization is performed for the area which is half of the output plane. What's interesting is that less than 0.6% deviation in output intensity can be observed when a single DOE is illuminated at an angle that spans from 0 to 80 degrees. Then, we develop a neural network model to experimentally design and validate SpliCons, a special type of diffractive optical element that can achieve spectral splitting and simultaneous concentration of broadband light. Our results show that the neural network model, the SpliConNet, yields enhanced spectral splitting performance for the SpliCons with quantitative assessment compared to the SpliCons that are optimized via the local search optimization algorithm. The capabilities of the SpliCons optimized via the neural network are experimentally validated by comparing the intensity distribution at the output plane. Once the neural networks are trained, we manage to design the SpliCons with 96.6 ± 2.3% accuracy within 2 seconds, which is orders of magnitude faster than iterative search algorithms. Later, we present a DOE spectrometer design that performs spectral decomposition of the broadband light. Compared to an optical spectrometer constructed with a single prism or a diffraction grating, our DOE spectrometer consists of many DOEs which are designed for each wavelength of light and are placed on a moving part. With the moving part, the position of each DOE is altered to obtain desired spectral light from the broadband light. Each DOE in the DOE spectrometer splits the broadband light into a specific wavelength of light, so our DOE spectrometer operates under broadband light compared to conventional spectrometers. We believe that the DOE spectrometer will yield high throughput thanks to a single wavelength-based DOE design. We also focus on designing optical holograms to generate holographic images at multiple observation planes and colors via a deep learning model, the CHoLoNet. The CHoLoNet produces optical holograms, which show multitasking performance as multiplexing color holographic image planes by tuning holographic structures. Furthermore, our deep learning model retrieves an object/hologram information from an intensity holographic image without requiring phase and amplitude information from the intensity image. We see that reconstructed objects/holograms show excellent agreement with the ground-truth objects/holograms.

Subject Keywords

Wavefront shaping, holography, imaging, deep learning, solar energy, spectroscopy, computer-generated hologram, diffractive optical element, spectral splitting, neural network, artificial intelligence

URI

https://hdl.handle.net/11511/96256

Collections

Graduate School of Natural and Applied Sciences, Thesis