| dc.description.abstract | In the age of big data, the challenge of efficient and rapid searching and comparing vast amounts of data has become unavoidable. At this point, Ternary Content Addressable Memories (TCAM) have emerged as an indispensable technology. TCAM, with their ability to handle both “Care” and “Don’t care” states, can effectively eliminate unnecessary comparison operations, significantly enhancing efficiency in search and comparison tasks. However, contemporary volatile TCAM, which are primarily based on SRAM circuits, face challenges such as high power consumption and low storage density per unit area. As a result, researchers have begun exploring the development of Non-volatile Ternary Content Addressable Memories (nv-TCAM) utilizing non-volatile memory technologies.
This paper presents a 2T32C nv-TCAM architecture designed using a Ferro-gated MOSFET, where the bottom electrode of the ferroelectric capacitor is connected to the gate terminal of the transistor. The polarization characteristics of the ferroelectric capacitor are utilized to modulate the MOSFET channel state, enabling control over the ＂Care＂ and ＂Don′t care＂ states, as well as data storage. Notably, the ferroelectric capacitor is constructed using angstrom-level ferroelectric materials stacked within the metal layers.
Experimental results show that this design achieves a memory window of 4.94V, a switching ratio of 5×10?, and 8-level storage states per memory cell. In terms of performance, the power consumption during search operations is as low as 9.6 μW/b, with a fast search speed of 4.5 ns. Each memory cell demonstrates an endurance of up to 1011 cycles and a data retention time of 10 years at 87.6°C.
This design resolves the data volatility issues of traditional SRAM-based TCAMs. By integrating ferroelectric capacitors into CMOS fabrication processes, it significantly reduces power consumption and improves storage density. Additionally, the unique properties of ferroelectric materials enable multi-bit storage states, offering a highly competitive solution for future search technology advancements. | en_US |