iBRAIN

Growing Big Data workloads are pushing conventional, current-driven computing architectures toward energy-efficiency limits. Neuromorphic computing offers a more efficient alternative, but existing artificial synapses suffer from high energy losses and poor durability. The iBRAIN project tackles these issues by developing magneto-ionic artificial synapses that operate through voltage-driven ion migration, avoiding energy-intensive electric currents. To enable efficient room-temperature operation, iBRAIN overcomes key magneto-ionic challenges, ion mobility and endurance, using defect engineering, advanced material design, and device miniaturization. Experiments show that Co-based oxides exhibit strong magneto-ionic performance when engineered via ion implantation and tailored thin-film growth, while Ni-based oxides perform worse due to higher kinetic barriers to ion motion. A major breakthrough is the demonstration of voltage-controlled switching between antiferromagnetic and ferromagnetic states, including room-temperature exchange bias without thermal annealing. Antiferromagnetic oxides and nitrides offer enhanced stability, immunity to external magnetic fields, and enable dense, crosstalk-free synapse integration. Finally, device miniaturization further improves ionic control and stability, supporting scalable neuromorphic architectures.