High-Resolution Perovskite Light-Emitting Diodes and MoS2 Synaptic Phototransistors for High-Definition Vision Systems

Abstract

Next-generation vision systems are expected to deliver ultrahigh-resolution image quality, on-device data processing, and low-power operation within compact and portable form factors. However, conventional architectures—where sensing, processing, and display are implemented as independent modules—inevitably suffer from latency, bandwidth bottlenecks, and power inefficiency, since their distributed operation must be sustained under a limited energy budget. As resolution and functional complexity continue to increase through pixel densification, these structural inefficiencies are further amplified, forcing each unit to emit and detect greater amounts of energy and information within confined physical space. The resulting energy concentration and signal load across the optical– electronic interface degrade overall optoelectronic efficiency, ultimately constraining both visual fidelity and computational performance. Therefore, the development of high-definition and compact vision platforms requires material- and device-level innovations that minimize energy loss and maximize functionality within each individual component. Semiconducting nanomaterials—such as colloidal nanocrystals, and low-dimensional semiconductors—exhibit exceptional optoelectronic tunability, mechanical flexibility, and process compatibility, providing a functional optoelectronic architecture for the optimization of optical and electronic performance. Nanomaterial-based optoelectronics offer a direct route toward this goal through the intrinsic coupling of optical and electronic processes, enabling high functionality and efficient energy utilization. In the first part of this work, nanocrystal-based light-emitting architectures are explored to realize ultrahigh-resolution and mechanically deformable display platforms. Double-layer transfer printing is introduced for the integration of perovskite nanocrystal (PeNC) films with organic charge transport layers, effectively suppressing mechanical fracture and mutual delamination during the printing process. This technique enables the formation of high-fidelity RGB PeNC arrays with pixel densities up to 2,550 pixels per inch (PPI) and monochromatic patterns up to 33,000 line pairs per inch, achieving a complete transfer yield. The transfer-printed perovskite light-emitting diodes (PeLEDs) exhibit superior electroluminescent performance, with external quantum efficiencies of 15.3%, 14.8%, and 2.5% for red, green, and blue devices, respectively. Moreover, the double-layer transfer printing allows the realization of ultrathin, multicolor PeLEDs (~2.6 m in total thickness) capable of stable operation on curved and deformable surfaces, including human skin, under various mechanical stresses such as bending, folding, wrinkling, and twisting. These results highlight the strong potential of perovskite emitters for next- generation, high-definition, full-color, and wearable display technologies. In the second part, synaptic optoelectronic devices are developed for in-sensor visual information processing, enabling precise feature extraction and energy-efficient decision-making. A multilevel image adjustment scheme based on gate-tunable synaptic phototransistors is introduced, allowing adaptive control of image brightness and contrast for distinct contour extraction. The device operation is inspired by dopamine-mediated regulation of neuronal excitability, where electrostatic gating facilitates or inhibits time-dependent photocurrent accumulation to modulate photoresponses under varying illumination conditions. By combining excitatory and inhibitory modes, the device enhances visibility in both dim and bright regions, thereby improving contour definition and subsequent segmentation accuracy. Quantitative evaluations using a road-scene dataset demonstrate significant improvements in object detection accuracy (from 83.2% to 86.7%) and Intersection over Union (from 0.73 to 0.79), along with substantial data compression to 8.2% of the original image volume. These results highlight the potential of synaptic optoelectronic devices for realizing high-clarity visual perception and energy-efficient in next-generation AI-native vision systems. Collectively, this dissertation demonstrates complementary nanomaterial-based approaches that link nanoscale optoelectronic behavior to system-level capability for high-definition visual information, establishing materials-driven enabling technologies for compact, energy-efficient, and high-definition vision systems. This dissertation is based on research work that has been previously published in peer-reviewed journals. The contents have been adapted with minor modifications to comply with the requirements of the dissertation format. The detailed list of publications is as follows: Chapter 2 is based on the article: Ultrahigh-Resolution Full-Color Perovskite Nanocrystal Patterning for Ultrathin Skin-Attachable Displays, Science Advances, 8, eadd0697 (2022). Chapter 3 is based on the article: In-sensor multilevel image adjustment for high-clarity contour extraction using adjustable synaptic phototransistors, Science Advances, 11, eadt6527 (2025). Minor edits, including reformatting, language revision, and the removal of journal-specific sections such as abstracts and reference list, have been applied to the published works for integration into this dissertation. Full citations are provided within the relevant chapters to acknowledge the original sources.