FeRAM gets speed boost, lower power from new modeling

Philadelphia (PA) - Researchers have developed a new theory which removes much of the mystery surrounding Ferroelectric RAM (FeRAM) and its operation. Armed with this new model, FeRAM may finally live up to its promise and become the front-runner for a Flash memory replacement technology.

The Discovery

The new model aligns very accurately with real-world observations. It’s even provided researchers with theoretical insight into not only why FeRAM works the way it does, but also areas of previously unknown potential. For example, the model shows that with minor changes in materials or design, power levels can be brought lower than was previously thought. In addition there are faster speeds and greater densities possible through other alterations. The new model’s findings are adding up to make FeRAM a very viable (and desirable) path to extremely high-speed, low-power, permanent, non-volatile storage. It might even be the ideal candidate for DRAM replacement, especially since FeRAM is manufactured very similarly to DRAM.

How Does it Work ?

FeRAM gains its non-volatility from the ferroelectric properties of the substrate material, typically PZT (lead zirconate titanate), though tests using PbTiO3 (lead titanate) and BaTiO3 (barium titanate) were used in the modeling. Instead of using a capacitor to store a charge as DRAM does, FeRAM uses something called a domain wall. Basically, the domain walls create physical barriers which move up and down based on the data’s state. An application could use the “up position” for storing a 1, and the “down position” for storing a 0, for example. This change in state involves several atoms physically moving location. But, because of the properties of atoms in this particular material, the move happens very fast—in about 1ns. This makes its potential maximum speed much greater than DRAM. However, there are similar external circuitry requirements to DRAM. This creates a speed hindrance in current designs, not from charging the capacitor as it is in DRAM, but rather from the switching circuitry used to open the paths to each FeRAM bit.

The read cycle determines the up/down state and sends back the appropriate result. Write cycles are designed to move the domain wall to its new location. This could also mean leaving it in the same state if a it happened to already contain the same value. Because of this design, and because FeRAM does not require significant power to move this group of atoms, it uses much less power than DRAM. It also generates less heat and has the potential to become much faster. And thanks to this new theory, it is likely we’ll actually see those improvements, as well as making it more dense.

What Does Having FeRAM Bring Us ?

FeRAM has been had many proponents who’ve hail it as a desirable all-around memory solution. Because of its comparable speed to DRAM, non-volatility and lower power consumption, many believe its potential is unrivaled. However, until this new model was created scientists lacked the tools necessary to fully understand and wield it. Several research efforts have involved tweaking the design through experimentation, though the actual mechanics were never completely understood. This reality hindered nearly all efforts to move FeRAM forward and improve its design. Because of this, even in 2006 worldwide FeRAM sales accounted for well under 1% of the total non-volatile memory market. This was true despite the products of the time already having many more-desirable qualities, such as an estimated 10^16 read/write cycles before failure, compard to Flash’s 10^5, faster read/write times, less power consumption, etc.

The significance of this new finding brings much to scientists. FeRAM now has the potential to make possible the long sought after instant-on computer. The kind that doesn’t require booting up and, when powered on, has everything exactly where it was, windows and all. All files are open, all programs are running, all everything is just as it was when it was powered down. The complete machine state was saved in FeRAM because of its non-volatile nature, even what was in memory.

Imagine being able to have notebooks, PDAs, MDIs or whatever, all of which can be used without worry right up until the time they shut down due to battery depletion. After pulling out the power cord and plugging it in, such a system would be able to continue from exactly where it was before without loss. No rebooting. No resuming from low-power state. No reading from the hard drive to restore memory. In fact, nothing would be any different than how it was when it shut off.

That is the potential FeRAM offers when employed as a DRAM replacement. And when employed as a Flash memory replacement, the potential exists for DRAM speeds in place of hard drive or flash, along with less power consumption. In fact, the real possibility here is that a combination main-memory / permanent storage could exist with FeRAM, allowing for a truly capable machine to say the least.

Conclusion

Lower power, comparable-to-DRAM speeds with greater headroom for speed increases on the horizon, non-volatile and now well understood. It can be used as DRAM or Flash memory, a viable replacement for both. FeRAM may finally live up to its promised potential. The dreams of 1999 may see the light of day in 2009.