Prof. Lanza’s Work Published in Nature Electronics
The research group led by the Prof. Mario Lanza, based at the Institute of Functional Nano & Soft Materials of Soochow University, has developed memristors capable of partially emulating the behavior of biological synapses in the human brain. Synapses are nerve endings that connect neurons in the brain, and have the function of transmitting (or not) the electrical impulses that these generate. To do this, the synapses change their resistivity by secreting calcium and sodium ions. In a very similar way, memristors are two terminals electronic devices nano-structured in the form of conductor/insulator/conductor, in which the insulating layer can change its resistivity depending on the electrical impulses applied to the input by secreting ions of oxygen or metal.
Despites memristors were considered the most promising electronic device to emulate the behavior of biological synapses, until now their main limitation was that memristors made of traditional materials (metals and oxides) had many difficulties in emulating the dynamic behavior of a synapse. In other words, the changes in resistivity in a traditional memristor were produced very abruptly between two states depending on the applied electrical impulses, while in a synapse the modulation of the resistivity is more progressive and its relaxation depends much more on the time. The memristors developed by the group of Prof. Mario Lanza show, for the first time, a very progressive, stable and repetitive relaxation process, which represents a new world record. In addition, these memristors allow to work in volatile or non-volatile mode depending on the electrical impulses applied to the input, as biological synapses do. These advances have been possible thanks to the introduction of two-dimensional materials in the structure of the memristors, such as multi-layer hexagonal boron nitride (also called white graphene). These results, published this week in Nature Electronics, represent a major breakthrough for the manufacture of artificial neural networks, which are computational structures essential for the development of advanced artificial intelligence systems.