Power/Performance Bits: Dec. 31

Three-valued memory; zip compression and entropy.


Three-valued memory
Scientists at the Tokyo Institute of Technology and the University of Tokyo developed a new three-valued memory device inspired by solid lithium-ion batteries which could potentially serve as low power consumption RAM.

The new device consisted of a stack of three solid layers made of lithium, lithium phosphate, and gold. This stack is essentially a miniature low-capacity battery that functions as a memory cell; it can be quickly switched between charged and discharged states that represent the two possible values of a bit. However, gold combines with lithium to form a thick alloy layer, which increases the amount of energy required to switch from one state to the other.

Design of the layered three-valued memory cell. The stacked layers in the proposed memory device form a mini-battery that can be quickly and efficiently switched between three different voltage states (0.95 V, 1.35 V, and 1.80 V). (Credit: Tokyo Institute of Technology)

To counteract this, the researchers substituted nickel for the gold. They expected better results using nickel because it does not easily form alloys with lithium, which would lead to lower energy consumption when switching. The memory device they produced was much better than the previous one; it could actually hold three different voltage states instead of two, meaning that it is a three-valued memory device. “This system can be viewed as an extremely low-capacity thin-film lithium battery with three charged states,” said Taro Hitosugi, a professor at Tokyo Tech. Three-valued memory implementations have the potential to be more area efficient.

The researchers also found that nickel forms a very thin nickel oxide layer between the Ni and the lithium phosphate layers, which is essential for the low-energy switching of the device. The oxide layer is much thinner than that of the gold-lithium alloys that formed in their previous device, which means that this new “mini-battery” cell has a very low capacity and is therefore quickly and easily switched between states by applying minuscule currents. “The potential for extremely low energy consumption is the most noteworthy advantage of this device,” said Hitosugi.

Zip software and entropy
Researchers at Tel Aviv University discovered a way to calculate the entropy of complex physical systems, such as tendency of a properly folded protein to unravel, using common compression software.

“We discovered a way to calculate entropy using a standard compression algorithm like the zip software we all have on our computers,” said Roy Beck, a professor at TAU’s School of Physics and Astronomy. “Supercomputers are used today to simulate the folding or misfolding of proteins in diseased states. Our study demonstrated that by using a standard compression algorithm, we can provide new insights into the physical properties of these proteins by calculating their entropy values using a compression algorithm.”

Calculating entropy is an important aspect of many fields, from biomedical simulations to basic research conducted in physics, chemistry or material science, the team said.

They approached entropy from the point of view of information theory. “They simulated a few standard physical systems with entropy values they can compare to,” said Beck. “Soon they found that the simulation data file size after compression rises and falls just as the expected entropy should. Shortly after that, they realized they could convert the compressed file size into a usable value — the physical entropy. Surprisingly, the simple conversion they used was valid for all the systems studied.”

At present, the team is mostly focusing on simulation of protein folding, which has important implications for treating disease. But they think the approach could be applied to a wide selection of systems.

“A high school student used our concept to calculate the entropy of a complex physical system — the XY model,” said Beck. “Although this is considered a challenging problem with regard to entropy, the student accomplished it with very little guidance. This demonstrates how easily this method can be used by almost anybody to solve very interesting problems.”

Leave a Reply

(Note: This name will be displayed publicly)