Democritus University of Thrace (DUTH)
Design, fabrication and applications of memristive devices and circuits for in-memory and edge computing using silicon nitride dielectrics
Abstract
dc:descriptionThis thesis investigates the potential of silicon nitride (SiNx)-based resistive random-access memory (RRAM) or memristor devices as high-density, low-power, and high-speed nonvolatile memory solutions capable of addressing limitations faced by traditional flash memory and advancing in-memory computing. As technology demands increase, memristor has emerged as a promising candidate due to its resistive switching phenomenon, which supports efficient scaling and operational advantages over flash memory. This research explores the fundamental properties of SiNx in RRAM, examining its suitability for next-generation applications in areas like artificial intelligence (AI) and edge computing. Key research objectives include optimizing fabrication techniques, evaluating interlayer impacts on device performance, and investigating substrate effects on noise stability. By analyzing random telegraph noise (RTN) and defects management through doping, this study addresses challenges in device reliability, retention, and endurance. Innovative device architectures, including SiNx-based crossbar arrays, aim to enhance integration of memory and logic operations, while silicon-on-insulator (SOI) substrates reduce current overshoot, supporting multi-level resistance switching. Further, this work introduces a SiNx-based true random number generator (TRNG) for secure computing applications, leveraging high-entropy multistate resistance properties. The research contributes substantially to advancing SiNx-based RRAM by revealing insights into noise behavior, multilevel switching, and substrate-dependent performance. Collectively, these findings underscore the importance of SiNx’s compatibility with complementary metal-oxide-semiconductor (CMOS) processes, emphasizing its potential as a stable, scalable material for future RRAM technologies and efficient in-memory computing. This thesis presents a comprehensive study on SiNx memristive devices, guiding advancements in RRAM technology through detailed assessments of device stability, innovative architectures, and material interactions, establishing a foundation for robust nonvolatile memory solutions and sophisticated computing architectures.
Degree
thesis:*- Grantor dc:publisher
- Democritus University of Thrace (DUTH)
- Year dc:date
- 2025
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Vasileiadis, Nikolaos
- Βασιλειάδης, Νικόλαος
Subjects
dc:subject × 44- Νανοδιατάξεις Memristor
- Διηλεκτρικά υλικά
- Νιτρίδιο του Πυριτίου (Si₃N₄)
- Μη πτητική μνήμη
- Κατασκευή διατάξεων
- Θόρυβος χαμηλής συχνότητας
- Δικτυώσεις τύπου σταυροειδούς πλέγματος
- Υπολογισμός εντός μνήμης
- Γεννήτρια αληθινών τυχαίων αριθμών (TRNG)
- Τεχνητή νοημοσύνη
- Memristive devices
- RRAM (Resistive Random Access Memory)
- Silicon nitride (SiNx)
- Non-volatile memory
- Resistive switching
- Device fabrication
- Dielectric materials
- Silicon oxide interlayer
- Doping
- Defects engineering
- Low-frequency noise (LFN)
- Random telegraph noise (RTN)
- Device reliability
- Noise analysis
- Crossbar arrays
- In-memory computing
- Edge computing
- In-sensor computing
- Memristor-based logic
- CMOS compatibility
- True random number generator (TRNG)
- Multilevel resistance states
- Artificial Intelligence (AI)
- Secure computing
- Silicon-on-insulator (SOI)
- Substrate effects
- Current overshoot
- Memristor ratioed logic (MRL)
- Επιστήμη Ηλεκτρολόγου Μηχανικού, Ηλεκτρονικού Μηχανικού, Μηχανικού Η/Υ
- Επιστήμες Μηχανικού και Τεχνολογία
- Υπολογιστές, Υλικό (hardware) και Αρχιτεκτονική
- Electrical Engineering, Electronic Engineering, Information Engineering
- Engineering and Technology
- Computer science, Hardware and Architecture
Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/58822