Search for: All records

Abstract Due to its transparent and conductive nature, indium tin oxide (ITO) offers substantial benefits in several industries, such as thin film transistors, displays, and nanophotonics. Previous studies on ultrathin ITO have so far focused on its electrical properties but have neglected the technologically important epsilon-near-zero (ENZ) optical features due to the difficulty of extracting the refractive index and the thickness-dependent degradation of the optical properties. Here, we demonstrate a complementary metal-oxide-semiconductor (CMOS)-compatible deposition procedure for sub-percolation thickness (below 4 nm) ITO using a dry-etch assisted radiofrequency magnetron sputtering technique that yields continuous films in a precisely controlled manner. Through interface engineering and post-deposition annealing optimization, we show that these ITO films can retain high carrier mobility (43 cm²V⁻¹s⁻¹) while achieving a tunable near-zero-index (NZI) regime throughout the telecommunications band using a Berreman-assisted optical characterization technique. Our result opens the possibility of efficiently designing ENZ/NZI materials at the nanoscale using a robust fabrication approach for applications in nanophotonics. 

Concolic execution is a powerful technique in software testing, as it can systematically explore the code paths and is capable of traversing complex branches. It combines concrete execution for environment modeling and symbolic execution for path exploration. While significant research efforts in concolic execution have been directed toward the improvement of symbolic execution and constraint solving, our study pivots toward the often overlooked yet most common aspect: concrete execution. Our analysis shows that state-of-the-art binary concolic executors have largely overlooked the overhead in the execution of concrete instructions. In light of this observation, we propose optimizations to make the common (concrete) case fast. To validate this idea, we develop the prototype, SYMFIT, and evaluate it on standard benchmarks and realworld applications. The results showed that the performance of pure concrete execution is much faster than the baseline SYMQEMU, and is comparable to the vanilla QEMU. Moreover, we showed that the fast symbolic tracing capability of SYMFIT can significantly improve the efficiency of crash deduplication.