Dispersion-free accurate color estimation using layered excitonic 2D materials and machine learning

Abstract

Dispersion is accepted as a fundamental step required for analyzing broadband light. The recognition of color by the human eye, its digital reproduction by a camera, or detailed analysis by a spectrometer utilize dispersion; it is also an inherent component of color detection and machine vision. The central thesis of this dissertation is that dispersion is not a prerequisite for a machine to recognize color. The research pursued in this dissertation establishes this novel idea beyond reasonable doubt and leads to the development and demonstration of a transformative new method for color recognition and spectral estimation that does not require traditional dispersion of light. Instead, we exploit spectral variations of optical transmittances of wide-band transmissive windows to accurately estimate the broadband colors. Presented in this dissertation are the experimental, simulation, and machine learning innovations that we developed, step- by-step, to accomplish this unusual new technique.In chapter 1 of this dissertation, we present an overview of the current state of the art in the materials used in this dissertation. Then, we present the main spectroscopy techniques that we used extensively in this dissertation. Next, we discuss the progresses that advanced data analytic methods such as machine learning have brought into the domain of materials science. Finally, we direct the discussion into the area of spectrum and color estimation and cast doubt on the efficacy of the existing tools. We present a method and device for estimating spectrum and color by introducing the advanced data analytic techniques, uniquely, into the optical properties of layered excitonic 2D materials and briefly mention the advantages that our method has over the existing technologies. In chapter 2, we present the method for estimating the wavelength of any monochromatic/near- monochromatic light within the visible range of wavelengths (325-1,100 nm). There, we present an easy- to-fabricate array of wide-band transmissive windows, discuss how statistical tools such as Bayesian theory can effectively extract the wavelength information from features of the transmission spectrum, which others might consider simply as noise, and study the sources of estimation error, as well as transmissive window stability and reusability over time using analytical methods. In chapter 3, we present an overview of a few machine learning algorithms and study their efficacies in monochromatic wavelength estimation. We primarily consider their differences in testing time for real-time applications and their differences in situations where the data size is of concern, i.e., where the data is scarce. Finally, we expand on work in chapter 2 by introducing an analytical method to predict and correct for the drift -- that naturally happens to the physical properties of materials when exposed to ambient conditions - hence, making it possible to use the same transmissive windows over extended periods of times, without compromising the accuracy. In Chapters 4 and 5, we present a device (called artificial eye or A-Eye) that accurately recognizes and reproduces tested colors without any spectral dispersion. Instead, A-Eye uses N=3-12 transmissive windows, each with unique spectral features resulting from the broadband transmittance and excitonic peak features of 2D transition metal dichalcogenides. Colored light passing through (and modified by) these windows and incident on a single photodetector generated different photocurrents, and these were used to create a reference database (training set) for 1,337 "seen" and 0.55 million synthesized "unseen" colors. By "looking" at test colors modified by these windows, A-Eye can accurately recognize and reproduce "seen" colors with zero deviation from their original spectra and "unseen" colors with only ~1% median deviation, using the k-NN algorithm. A-Eye can continuously improve color estimation by adding the corrected guesses to its training database. A-Eye's accurate color recognition dispels the notion that dispersion of colors is a prerequisite for color identification and paves the way for ultra-reliable color recognition by machines with reduced engineering complexity. Finally, in Chapter 6, we summarize the outlook and future directions likely to emerge from this dissertation.--Author's abstract