A new method to change the state of individual magnetic nanoparticles within a dense array using a low cost, low-power laser for data storage technology and potential application in neuromorphic ‘in-memory’ computing.
Proposed use
The technological benefits of optically-switched magnetism are enormous and achieving it has long been a goal of both industry and academia.
This technology allows low power, high density magnetic data storage in bit patterned media. A key future application could be in next-generation neuromorphic computation hardware as this technology is very suited to new technologies where multiple storage and/or logic functions are co-located e.g. ‘universal memory’, ‘memcomputing’, ‘hardware neural networks’.
Problem addressed
Computation is forecasted to reach 30% of global energy production by 2030. However, systems using nanomagnetics could be 100,000 times more energy efficient than standard electronics as information is transferred as a wave, reducing global energy need.
Established nanomagnetic technologies require ultrafast and intense pulses of light from expensive high-power lasers or otherwise are dependent on circular optical polarisation or a magnetic field. This technology uses extremely low power continuous wave lasers with linear polarization in the absence of a magnetic field.
This technique only uses aluminium, nickel and iron compared to rare and expensive elements used in existing techniques.
Technology overview
This invention is a new method to change the state of a magnetic particle within a dense nanomagnet array using continuous wave light from an extremely cheap low powered laser.
Hybrid magnetoplasmonic structures were developed combining a switchable magnetic bit with a light-focusing plasmonic antenna. Deterministic switching was demonstrated in high-density, strongly-interacting arrays including writing exotic high-energy states.
Benefits
- All optical magnetic switching with extremely cheap low power lasers in the absence of a magnetic field.
- Low power, high density data storage well suited to hardware neural networks.
- Potential application in next-generation neuromorphic computation hardware
- Magnetic computing is theoretically 100,000 times more energy efficient data storage compared to standard electronics.
- Low cost materials such as aluminium, nickel and iron compared to existing techniques
- Continuous wave laser uses linear polarization and so is not dependent on circular optical polarization.
Intellectual property information
UK Priority Application (Number: 2117279.6)