Ideal memristor based on the spin liquid state in magnetic heterostructures

Emory University, Atlanta GA

Investigators

Abstract

Computers operating with digital representation of information and Boolean logic have revolutionized science and technology. However, they remain far inferior to human brain due to their inability to adapt and to process massive amounts of imprecise information. These capabilities can be accomplished with neuromorphic computing, which can be efficiently implemented at the hardware level by utilizing memristors - electronic devices whose resistance depends on their electrical history. The existing memristors operate mostly like on/off switches, lacking the ability to continuously vary the resistance in response to electric stimuli required for neuromorphic applications. These devices commonly rely on the physical motion of atoms, and as a result lack durability and reproducibility. The proposed project will develop a new class of memristors based on the special magnetic properties of ferromagnet/antiferromagnet bilayers. The operation of the proposed devices will rely on the magnetic frustration at the interfaces between ferromagnets and antiferromagnets, resulting from the incompatible magnetic ordering of the two materials, which will lead to the formation of a viscous spin liquid state in the antiferromagnets. The project will explore the most suitable materials and geometries, electronic mechanisms enabling device operation, and device functionality at nanoscale. The proposed research will be integrated with STEM education through the development of a comprehensive undergraduate Materials and Engineering Physics program, which will include a hands-on freshman seminar and a state-of-art research training course for undergraduate students. The plan is also to organize a Science club at the local elementary school. A theoretically envisioned ideal memristor is an electronic device whose resistance is proportional to the total charge that passes through it, with the coefficient of proportionality known as memristance. The proposed project will experimentally realize ideal memristor nanodevices, by utilizing bilayers of low-anisotropy ferromagnets, such as Permalloy, and thin films of antiferromagnets such as NiO, CoO, or Fe50Mn50. A spin liquid state is expected to be formed in the antiferromagnets, due to the magnetic frustration associated with the random exchange interaction at the magnetic interface. The proposed project will utilize a combination of material and heterostructure engineering, nanofabrication, measurements of time-domain magnetic dynamics and transverse ac susceptibility, to address questions related to physical mechanisms of device operation, and device design. The proposed research will explore the possibility to engineer frustrated thin-film magnetic heterostructures that exhibit a spin liquid state with a well-defined and tunable viscosity, resulting in ideal memristive functionality with controlled memristance. Additionally, the project will identify and characterize the magnetoelectronic mechanisms that can facilitate writing, reading, and resetting of the memristive devices based on viscous spin liquids. The relevant length scales for the memristive properties will be established, to determine whether the proposed devices are scalable to the technologically relevant nanoscale dimensions. By addressing these questions, the proposed project will provide a transformative contribution to the research and development of memristive devices characterized by high-endurance, scalability, and tunable memristive properties, which will facilitate the implementation of efficient neuromorphic networks. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.