Improving TSV Efficiency in 3D ICs Using Perylene-N for Enhanced Noise Reduction
The increasing demand for smaller and faster electronic devices has driven continuous innovation in integrated circuit (IC) design within the semiconductor industry. Conventional long metal interconnects not only contribute to larger system sizes but also introduce RC delays, which negatively impact performance by limiting communication bandwidth. To overcome these challenges, 3D IC integration has emerged as a promising solution. This approach involves stacking multiple functional layers on a silicon (Si) substrate, with electrical connections facilitated by through-silicon vias (TSVs). TSVs are essential for enabling high-speed interlayer communication in 3D architectures. However, one of the primary challenges in TSV-based 3D ICs is signal degradation caused by noise coupling between signal TSVs (aggressors) and ground TSVs (victims). Maintaining signal integrity requires effective isolation between TSVs and the Si substrate, which depends on optimized liner materials and structural configurations. Among the various dielectric materials available, Perylene-N has demonstrated exceptional potential in reducing area and power consumption. This study compares Perylene-N with traditional liner materials such as silicon dioxide (SiO₂), Benzocyclobutene (BCB), and Teflon AF 1600 for embedded TSVs (ETSVs). Noise coupling is analyzed under different conditions using a dielectric-metal-dielectric configuration surrounding copper TSVs. The results reveal that Perylene-N significantly enhances noise isolation, reducing noise coupling by 51.28% at terahertz (THz) frequencies compared to SiO₂, as confirmed through rigorous validation. Across all frequency ranges, Perylene-N consistently surpasses SiO₂ in minimizing noise coupling, achieving an average reduction of 10–13 dB.
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2025-04-01
- DOI
- https://doi.org/10.5281/zenodo.15118682
- Primary Topic
- Machine Learning in Bioinformatics
- Type
- article
- Field-Weighted Citation Impact
- 0.00