With the rapid advancement in portable electronics, artificial intelligence, and the Internet of Things, there is an escalating demand for high-density, low-voltage non-volatile memory (NVM) technologies. Germanium (Ge) nanocrystals (NCs) have emerged as a promising candidate for NVM applications; however, traditional synthesis methodologies suffer from limitations in achieving precise control over the size and density of these nanocrystals, which exert a significant influence on device performance. This study presents an innovative Ag-catalyzed chemical vapor deposition (CVD) methodology for the synthesis of Ge NCs with precisely controllable size and density on SiO₂/Si substrates, tailored for NVM applications. Scanning electron microscopy characterization confirms the successful growth of faceted Ge NCs. Electrical characterization of the fabricated devices reveals that Ge NCs grown at temperatures ranging from 700 to 1000 °C exhibit memory windows spanning from 3.0 to 6.8 V under a ±6 V bias. Notably, the device synthesized at 900 °C demonstrates an exceptional memory window of 7.0 V under a ±8 V bias. Furthermore, the Ge NC-based NVM devices exhibit excellent charge retention characteristics. Specifically, for the device with Ge NCs grown at 700 °C, the time required to retain charge from 100% to 95% of its initial value exceeds 10 years, demonstrating long-term stable charge storage capability. These findings underscore the significant potential of this approach for the development of high-performance NVM technologies.

Self-assembled nanosphere structures are widely used as bottom-up platforms for ordered micro- and nanostructures, with applications in photonic crystals, sensing platforms, functional coatings, drug delivery, and nanosphere lithography. Their performance and reproducibility depend on structural order, packing density, interparticle spacing, and defect density and distribution. Thus, reliable imaging and quantitative defect detection are needed for quality evaluation and process optimization. This review provides an overview of defect characteristics, imaging systems, and defect detection algorithms for self-assembled nanosphere structures. It first introduces representative zero-, one-, two-, and three-dimensional assemblies, followed by a summary of common defects, including vacancies, interstitial particles, dislocations, grain boundaries, stacking faults, voids, and cracks. Optical microscopy, electron microscopy, atomic force microscopy, scanning near-field optical microscopy, and correlative techniques are compared in terms of resolution, field of view, temporal resolution, contrast mechanism, in situ capability, and compatibility with feedback control. Algorithmic approaches are also reviewed, encompassing classical image processing, machine learning, and deep learning, along with their applications in segmentation, localization, classification, and high-throughput analysis. Overall, reliable defect inspection requires integrated workflows. These workflows should combine appropriate imaging systems, image quality control, transferable algorithms, standardized datasets, and closed-loop feedback.