American chip maker Freescale Semiconductor has today announced the development of a magnetoresistive random-access memory (MRAM) chip which can maintain data using magnetic properties and not the traditional electrical charge methodology.

Think of it in terms of storing data more like a hard drive, albeit a very small one indeed, and you are on the right conceptual tracks. Unlike flash memory, MRAM does not degrade over time, does not need any power in order to store the data and is lightning quick when it comes to read/write speeds. Think in terms of nanoseconds in fact!

Each MRAM chip is comprised of memory cells, hundreds of thousands of them, containing a magnetic electrode with a fixed magnetic field and another that can change polarization. The magnetic orientation of electrons being used to represent bits.

Like much of the innovation we see in technology today, this is not actually something new, as people (including IBM) have been experimenting with similar memory modules for many years. However, it is important because Freescale Semiconductor is the first to overcome the problem of producing it in any volume. Indeed, the only reason it has kept quiet about the breakthrough was to enable it to build the kind of inventory that would silence the critics and back up the claims. Production actually started at the Arizona factory over two months ago!

Will Strauss, an analyst with research company has gone so far as to describe it as being “the most significant memory introduction in this decade." Considering the pace of mobile device technology, and the problems inherent with flash memory power-down data loss, he could well be right.

It is the mobile market that holds the key for such memory advances, and no coincidence that Freescale Semiconductor is a spin-off from Motorola. Ultimately, it may reduce the cost, power consumption and form factor of mobile phones, MP3 players, PDAs and even notebook PCs.

The only downside at the moment, when compared with flash memory, is capacity: the Freescale Semiconductor MRAM chip is currently maxed out at 4GB.

Dani AI

Generated

A short, practical update that builds on the original announcement and the two follow-ups from and .

MRAM basics — MRAM stores bits in magnetic tunnel junctions and modern products write using spin-transfer or spin‑orbit torque, so the underlying physics and device stacks are very different from charge‑based flash. That physical difference is why MRAM combines non‑volatility with SRAM/DRAM‑class access times and very high write endurance. (journals.aps.org)

On stray magnetic fields (answering ): commercial MRAM devices are designed and packaged with magnetic‑immunity specs and soft‑magnetic shielding; vendor guidance and application notes say ordinary household magnets or nearby CRTs are extremely unlikely to flip stored bits once the device is mounted on a PCB. Handling with very strong permanent magnets (during unpacking or at wafer level) is where caution is warranted — vendors provide explicit limits and handling notes. (everspin.com)

On fragility and reliability (answering ): MRAM is a thin‑film stack, but parts are packaged and qualified like other ICs (solder‑reflow, MSL, mechanical and thermal tests). In normal use and shipping the package protects the MTJ film; mechanical shock that would flip a bit would need to be extreme or involve very strong local fields. Design‑in checklist: follow the MRAM datasheet for magnetic‑field and shock specs, run a PCB reflow + field‑exposure test, and validate endurance/retention under your operating temps. (everspin.com)

Will MRAM make flash disappear? Not overnight. MRAM is already displacing embedded NOR/eFlash in some nodes and niche high‑reliability applications (automotive, avionics, embedded controllers), but NAND flash remains far cheaper per bit for bulk storage. MRAM is most likely to expand into places where endurance, instant persistence, or operating-temperature/radiation tolerance matter; full replacement of mass NAND storage depends on future density and cost improvements. (semi.org)

Short practical references: vendor application notes and datasheets are the authoritative source for handling limits and design margins — consult them when planning a board change or a migration from flash.

Recommended Answers

All 2 Replies

My question is will the magnetic fields generated by everyday objects or the elecronics within a computer cause some errors in magnetic memory? What about your computer sitting next to your big crt monitor. Could this have the potential to perminatly change the magnetic pole orientation within the memory?

Well, my opinion, which is probably insanely misguided is that it is much like the magnetic charge on a floppy disk's magnetic tape. Which would take a bit more than a fridge magnet to wipe it out but then again memory is exponentially more volatile then the aforementioned floppy disk. Also, I've taken a look at the structure of the chips and it is an amazing idea, however it looks extremely, for lack of a better word, fragile and susceptable to damage from the slightest bumps and the magnetic charges albeit extremley small could perhaps interfere. As I said my opinion is perhaps very misguided and the developers/inventors have without a doubt thought of every point I have put across here and although my opinion is a little pessimistic, I do hope that MRAM will be the future.

Be a part of the DaniWeb community

We're a friendly, industry-focused community of developers, IT pros, digital marketers, and technology enthusiasts meeting, networking, learning, and sharing knowledge.