Magnetic Memory Using Spin Torque Transfer to Reduce Writing Current
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Magnetic random access memories (MRAMs) face challenges in reducing writing current while maintaining thermal stability and scalability, with existing methods like spin torque transfer and current-driven domain wall motion facing issues of rewriting resistance and reliability, and requiring large chip areas and high power consumption.
Innovation Solution
A magnetic memory structure incorporating a ferromagnetic body film with perpendicular magnetic anisotropy, utilizing a stacked film of Co and Ni on a Pt layer, with a specific underlying layer structure and thermal treatment to enhance perpendicular magnetic anisotropy and domain wall motion suitability, and a middle nonmagnetic layer to prevent magnetic coupling and ensure robustness against thermal changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If a magnetic field is used to switch magnetization direction in conventional MRAM, then high-speed operation is achieved, but writing current increases to several milliamperes and chip area expands
Solution Approach 1:
The patent replaces the conventional magnetic field-based magnetization switching mechanism with a spin torque transfer mechanism. Instead of using external magnetic fields generated by current-carrying wires, the invention uses spin-polarized electrons to directly exert torque on the magnetic moments, achieving magnetization reversal through quantum mechanical spin transfer rather than classical magnetic field interaction. This substitution fundamentally changes the physics from electromagnetic induction to quantum spin transport.
Solution Approach 2:
The patent changes the operating parameters by utilizing spin-polarized current with specific polarization direction and magnitude to switch magnetization. The writing current is reduced from several milliamperes to lower levels by exploiting the spin transfer torque effect, where the key parameter becomes the spin polarization of the current rather than the magnetic field strength. This parameter change enables efficient magnetization switching at lower current levels.
2Area of moving object
If element size is reduced for scalability, then integration density improves, but writing current increases in conventional domain wall motion MRAM
Solution Approach 1:
The patent replaces the domain wall motion mechanism with direct spin torque transfer for magnetization switching. In domain wall motion, current drives domain walls to move along the magnetic strip, but this requires relatively high currents that increase as elements are scaled down. The spin torque transfer mechanism directly reverses magnetization without requiring domain wall motion, eliminating the scalability penalty and enabling consistent low-current operation across different element sizes.
Solution Approach 2:
The patent extracts and eliminates the domain wall motion step from the magnetization switching process. By using spin torque transfer, the invention directly switches magnetization between parallel and antiparallel states without requiring the formation and movement of domain walls. This extraction of the domain wall intermediate state simplifies the switching mechanism and removes the current-scaling penalty associated with domain wall dynamics.
3Reliability
If spin torque transfer is used with insulating layer between magnetic layers, then magnetization reversal is achieved, but rewriting resistance increases and reliability decreases
Solution Approach 1:
The patent uses a conductive or semiconductive spacer layer as an intermediary between the magnetic layers, replacing the insulating layer used in conventional spin torque transfer. This intermediary layer allows writing current to flow between the magnetic layers with lower resistance, reducing power consumption and improving reliability. The spacer layer serves as a current pathway that maintains the spin torque transfer mechanism while eliminating the high resistance problem of insulating barriers.
4Device complexity
If writing current path is identical to reading current path, then device complexity is reduced, but incorrect data may be written during readout
Solution Approach 1:
The patent segments the current paths into separate writing and reading paths. The writing current flows through the spin torque transfer path between magnetic layers, while the reading current flows through a separate magnetoresistance measurement path. This segmentation prevents reading operations from inadvertently causing magnetization switching, eliminating the risk of incorrect data writing during readout while maintaining manageable device complexity through dedicated current pathways.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach allows for a significant reduction in writing current, improved scalability, and enhanced thermal stability of the magnetic memory, enabling efficient and reliable data storage with reduced power consumption and increased competitiveness with other random access memories.
Implementation Method 1
a data memorizing layer (53) laid over the first nonmagnetic layer (52, 521) and comprising a ferromagnetic body having perpendicular magnetic anisotropy
Implementation Method 2
the ferromagnetic body film (53a) for the data memorizing layer (53) having perpendicular magnetic anisotropy
Implementation Method 3
a stacked film of Co and Ni on a Pt layer, with a specific underlying layer structure and thermal treatment to enhance perpendicular magnetic anisotropy and domain wall motion suitability
Implementation Method 4
a middle nonmagnetic layer to prevent magnetic coupling and ensure robustness against thermal changes
Implementation Method 5
with a specific underlying layer structure and thermal treatment to enhance perpendicular magnetic anisotropy and domain wall motion suitability
Data Source
AI summary
A memory includes an underlying layer of a ferromagnetic body, a first nonmagnetic layer on the underlying layer, a data memorizing layer laid on the first nonmagnetic layer and made of a ferromagnetic body having perpendicular magnetic anisotropy, a reference layer coupled through a second nonmagnetic layer with the data memorizing layer, and first and second magnetization fixed layers laid underneath the underlying layer to come into contact with the underlying layer. The data memorizing layer includes a magnetization liberalized region having reversible magnetization, and overlapping with the reference layer, a first magnetization fixed region coupled with an end of the magnetization liberalized region, and having a magnetization direction fixed to +z direction by the first magnetization fixed layer, and a second magnetization fixed region coupled with a different end of the magnetization liberalized region, and having a magnetization direction fixed to −z direction by the second magnetization fixed layer.