Acronym · Memory & Storage
MRAM (Magnetoresistive Random Access Memory)
Non-volatile memory that stores bits in the relative magnetisation of two ferromagnetic layers separated by a thin tunnel barrier, read as a resistance difference.
It combines fast writes with high endurance compared with flash.
- Type
- Acronym
- Domain
- Memory & Storage, Emerging memory
Notes
The magnetic tunnel junction sits in the back end of line metal layers, so it adds masks but not front-end complexity.
Overview
MRAM (magnetoresistive random access memory) is a non-volatile memory that stores each bit as the magnetic orientation of a thin ferromagnetic layer and reads it back as a change in electrical resistance. Because it keeps data without power and can be built in the metal layers above ordinary logic, it is used both as standalone memory chips and as embedded memory inside microcontrollers and systems on chip. MRAM is a family rather than one device: field-switched, spin-transfer torque (STT) and spin-orbit torque (SOT) types write bits in different ways.
How it works
The core of every MRAM cell is a magnetic tunnel junction: a very thin insulating barrier sandwiched between a fixed magnetic layer and a free one. When the two layers are magnetised in the same direction, current tunnels through the barrier easily and the resistance is low; when they point in opposite directions, the resistance is higher. Reading means passing a small current and sensing which state the junction is in.
Writing means flipping the free layer. The earliest products, known as toggle MRAM, did this with a magnetic field from currents in nearby wires, which works reliably but does not shrink well. STT-MRAM instead drives a spin-polarised current through the junction itself, which scales far better and is the type foundries now offer. The write current stresses the thin barrier, so designers trade speed, endurance and data retention against one another. SOT-MRAM, still at the research and development stage, sends the write current along a separate layer beneath the junction, which promises faster writes and longer life at the cost of a larger, three-terminal cell.
MRAM cells are fabricated in the back end of line, so adding them to a logic process needs relatively few extra masks. That is a large part of their appeal at nodes where embedded flash has become impractical.
Examples
- Embedded STT-MRAM replacing embedded flash in automotive microcontrollers built at advanced logic nodes.
- Standalone toggle MRAM used in spacecraft, where its stability and radiation tolerance outweigh its high cost per bit.
- MRAM used as a buffer in enterprise storage so that data survives a sudden power loss.
- Imec research on SOT-MRAM as a last-level cache, aiming to bring MRAM into roles now filled by SRAM.
Why it matters
MRAM matters to embedded memory and device engineers at foundries, to designers of microcontrollers and automotive chips who need on-chip non-volatile memory at advanced nodes, and to reliability and radiation-effects engineers, since magnetic storage is inherently resistant to radiation.
Related terms
Part of the Memory & Storage glossary, with every term in the topic and the lifecycles, tools and standards that use them.
Where it appears on the Hub
Used on 6 pages of the Hub. In each group, the pages that mention it most come first.
Suggest an improvement
Suggest an improvement to this entry
A clearer definition, an example, a correction or a source an editor can check. Editors review every suggestion before the entry changes.
Last reviewed . Written by The Semiconductor Hub editorial team. The full glossary