An Alternative Way for Reconfigurable Logic-in-Memory With Ferroelectric FET

Highly Reliable Logic-in-Memory by Bidirectional Built-in Electric-Field-Modulated Multistate IGZO/AFE Nonvolatile Memory

“…Recently, some LIM investigations of multistate FE NVMs based on partial switching have been demonstrated experimentally. 10 However, the partial switching of the ferroelectric multidomain will significantly amplify the electrical variations of states in cycleto-cycle (C2C) operations and device-to-device (D2D) distributions and therefore lead to serious signal overlap phenomena between different polarization states, 11 which are expected to greatly degrade the computational accuracy and integration efficiency in LIM circuits. Furthermore, as the number of ferroelectric domains decreases as the device is scaled down, the variation becomes more significant for arraylevel performance.…”

“…Remarkably, a multistate storage capability can be achieved in one FE NVM by partial switching of the ferroelectric multidomain, which greatly improves computational efficiency by improving the reconfigurability of logic operations compared with the two-state FE NVMs. Recently, some LIM investigations of multistate FE NVMs based on partial switching have been demonstrated experimentally . However, the partial switching of the ferroelectric multidomain will significantly amplify the electrical variations of states in cycle-to-cycle (C2C) operations and device-to-device (D2D) distributions and therefore lead to serious signal overlap phenomena between different polarization states, which are expected to greatly degrade the computational accuracy and integration efficiency in LIM circuits.…”

Optically Controllable Organic Logic-in-Memory: An Innovative Approach toward Ternary Data Processing and Storage

“…After implementing the two-level memory, it was extended to a ternary memory inverter to further improve the data integration density and achieve three memory states beyond binary storage. Finally, the operating voltage of the devices was significantly reduced to below 10 V using a high- k (dielectric constant) hafnium oxide (HfO 2 ) insulating layer, making it comparable with other binary and ternary memory devices with inorganic materials. ,,,,, Thus, the proposed devices could realize the necessary characteristics required for energy-efficient LIM units.…”

“…Finally, the operating voltage of the devices was significantly reduced to below 10 V using a high-k (dielectric constant) hafnium oxide (HfO 2 ) insulating layer, making it comparable with other binary and ternary memory devices with inorganic materials. 16,17,20,22,24,25 Thus, the proposed devices could realize the necessary characteristics required for energy-efficient LIM units.…”

“…The major performance bottlenecks of the current von Neumann computing systems are the physically separated memory and logic modules, which cause significant energy consumption and latency. − LIM computation can be a potential substitute for the prevailing von Neumann paradigm. Unlike von Neumann computing, LIM units can directly perform data computations inside nonvolatile memory, enabling space-saving, high-speed, and energy-efficient computations. ,− Several approaches for implementing efficient LIM units, such as floating gates, molecular switching, and ferroelectric technologies, have been investigated. − Nevertheless, most of the reported devices are limited to binary logic circuits and one-bit (two states: “0” and “1”) storage, which restrict their application in binary-weighted networks. ,− However, multivalued logic (MVL) circuits can solve this problem, especially for organic LIM units, as MVLs enable high data integration by increasing the number of logic states. In addition, the high photosensitivity of organic semiconductors can give rise to optoelectronic LIM units with multilevel storage.…”