| André Chanthbouala, Vincent Garcia, Ryan O. Cherifi, Karim Bouzehouane, Stéphane Fusil, Xavier Moya, Stéphane Xavier, Hiroyuki Yamada, Cyrile Deranlot, Neil D. Mathur, Manuel Bibes, Agnès Barthélémy and Julie Grollier
The paper presents a novel ferroelectric memristor that demonstrates continuous tunable resistance, a key feature of synapse-like behavior. The device, composed of BaTiO₃/La₀.₆₇Sr₀.₃₃MnO₃ (BTO/LSMO) bilayers, exhibits resistance variations exceeding two orders of magnitude with a 10 ns operation speed. The resistance changes are attributed to voltage-controlled domain configurations in the ferroelectric tunnel barriers. Using piezoresponse force microscopy (PFM) and electrical transport measurements, the researchers show that the resistance can be continuously tuned by varying the amplitude, duration, and number of voltage pulses. The switching dynamics are modeled using a simple model of domain nucleation and growth, leading to an analytical expression for the memristive effect. This work suggests that ferroelectric tunnel junctions (FTJs) can serve as a new class of memristive systems, offering new opportunities for neuromorphic computing architectures.The paper presents a novel ferroelectric memristor that demonstrates continuous tunable resistance, a key feature of synapse-like behavior. The device, composed of BaTiO₃/La₀.₆₇Sr₀.₃₃MnO₃ (BTO/LSMO) bilayers, exhibits resistance variations exceeding two orders of magnitude with a 10 ns operation speed. The resistance changes are attributed to voltage-controlled domain configurations in the ferroelectric tunnel barriers. Using piezoresponse force microscopy (PFM) and electrical transport measurements, the researchers show that the resistance can be continuously tuned by varying the amplitude, duration, and number of voltage pulses. The switching dynamics are modeled using a simple model of domain nucleation and growth, leading to an analytical expression for the memristive effect. This work suggests that ferroelectric tunnel junctions (FTJs) can serve as a new class of memristive systems, offering new opportunities for neuromorphic computing architectures.