MRAM chip design (Credit: Freescale)

Non-volatile memory, on the other hand, can retain stored information permanently, absolving the need for a constant power supply. ROM (read-only memory), which stores information that does not require frequent changing (i.e., doesn't need rewriting), such as Firmware (a software embedded inside a hardware such as a BIOS [basic input-output system]), is typically non-volatile. So, even when the system is off, the data is stored.  

Modern types of ROM such as Flash, used in thumb drives and many MP3 players, are also non-volatile, but easily rewritable, making them more like RAM. This combination of qualities has made Flash memory highly popular in recent years and it is currently used to improve other types of storage such as hard drives or even replace them altogether. But although Flash is cheap and non-volatile, it still suffers from a relatively limited lifetime. Though this had improved considerably in the last few years, more importantly, Flash still has a much lower write speed than RAM.  

In an attempt to combine the speed of the faster volatile memory with the benefits of non-volatile memory, Freescale (which originated from Motorola Semiconductor about two years ago) created a new type of non-volatile memory - Magnetoresistive Random Access Memory, or MRAM. The roots of MRAM can be traced back to the 1940's at Harvard when physicists An Wang and Way-Dong Woo and later Jay Forrester and colleagues at MIT worked on developments that led to Magnetic Core Memory and later on to the discovery of the "giant magnetoresistive effect" in thin-film structures by researchers from IBM in the late 1980's. Like Flash, MRAM retains data after a power supply is cut off, potentially eliminating that seemingly endless boot time of conventional computers when data from the hard drive is transferred to RAM, as well as loss of data when the computer is suddenly shut off. MRAM has much faster write speeds than Flash and has an unlimited endurance, meaning that MRAM is not subject to the degradation suffered by Flash.  

Conventional RAM memory is made of transistors and capacitors that are paired to create a memory cell, which represents one bit of data (0 or 1). Memory cells are aligned in columns and rows, the intersections of which are known as addresses in which information is stored. Reading and writing information occurs by measuring or changing the charge at a specific address, accordingly.

MRAM read process - Current is passed through the bit, resistance of the bit is sensed (Credit: Freescale)

MRAM write process - Current is passed through the programming lines generating magnetic fields. The sum of the magnetic fields from both lines is needed to program the bit (Credit: Freescale)

Write is achieved by the alignment of the magnetic moments of the two memory layers into one or the other relative orientation. Current is passed through two sets of parallel wires or write lines (called a bit line and a digit or word line), which pass over and beneath the memory cells, respectively.The bit lines and the digit lines run perpendicular to one another and at their intersections lie the magnetic memory cells, each defined by one particular bit line and one particular digit line. To write to a particular memory cell (bit), current is passed through the two wires that intersect at that memory cell. The magnetic field that is generated from current passing through a wire can change the orientation of the magnetic moments of the particular memory cell.

Although MRAM has many advantages over virtually every existing memory type, it is still in its infancy. Many had hoped MRAM would usher in the age of instant-on computers able to replace the computer main memory and hard drives, but, due mainly to its cost, these hopes remain a dream. At $25 per 0.5 MB, MRAM has no chance of competing with existing RAM selling for $25 per 256 MB, not to mention Flash, which sells for $25 per 1 GB. The only place where MRAM might be widely utilized is in specialized markets, for example, as a Battery-Backed SRAM replacement. Only when it breaks its current high price per MB ratio (possibly with more advanced lithographic technology than its current 0.18 microns) will MRAM's unique qualities find widespread usage. 

MRAM comparison table - cost is still high (Credit: Freescale)

The MTJ is composed of a fixed magnetic layer, a thin dielectric tunnel barrier, and a free magnetic layer.  When an electrical bias is applied to the MTJ, electrons that are spin-polarized by the magnetic layers traverse the dielectric barrier through a process known as tunneling. The MTJ device has a low resistance when the magnetic moment of the free layer is parallel to the fixed layer, and a high resistance when the free layer moment is oriented anti-parallel to the fixed layer moment. This change in resistance with the magnetic state of the device is an effect known as magnetoresistance, hence the name “Magnetoresistive” RAM. During the write operation, current pulses are passed through a digit line and a bit line. The currents generate a magnetic field that writes only the bit in the array that is at the cross point of those two lines.  The “1” or “0” is stored by putting the free layer magnetic moment into the antiparallel or parallel state, and the data is read by passing a current through the device to determine if the resistance is high or low.

---

The MRAM chip (Credit: Freescale)

Q: What are Freescale's plans for MRAM over the next five years?
A: Freescale has no plans to compete in the High Density commodity memory market. It is possible that expansion into some SRAM and Flash markets will be possible in the future, especially as embedded memory in our microcontrollers.