Researchers at the University of Santa Barbara in partnership with scientists from HP, published a study of atomic phenomena, heat and chemicals that occur within a memristor in titanium oxide. This research is so promising that HP envisions a marketing first memristors in 2013. The memristors are approached to replace DRAM and even Flash.
The paper was published today in the journal Nanotechnology. It is available this month on the Internet. It provides details on the yet unknown changes in strength of this component. The memristors represent the fourth passive component electrical capacitors, resistors and coils. Broadly, they have several levels of electrical resistance.
They are called memristors as they play the role of non-volatile memory. Instead of storing electrons or not they maintain a level of low or high resistance. The existence of these components has been theorized in 1971 and 2008, HP has proved the hypothesis of Professor Leon Chua. Palo Alto has already built memristors (cf.
"The first memristor in HP's 3D") and last year, he even struck a deal with Hynix for the mass production of these components in ReRAM, whose memories cells rely on memristors instead to conventional capacitors (see "HP and Hynix to sell memristors. The latest work has helped design a module with a density of 12 gigabits per square inch, manufactured in 15 nm and four eases piling on each other.
Even if HP can make memristors developed, the phenomena that occur within these components are poorly understood, mainly because the resistance changes occur in channels with a thickness of 100 nm. It's very small and it is difficult to design an experiment and measurement tools that do not alter the functioning of the memristor.
Researchers already knew that a memristor in titanium oxide (TiO2) has a high resistance. In this state, the electrons are scattered and do not pass. However, when heating the memristor, it switches to low resistance. A 10 nm channel is created in the anatase, the mineral composed of titanium oxide, and allows electrons to pass.
This tunnel is maintained even when the power comes more heat the memristor, which allows the memory to be nonvolatile. To better understand what is happening, the researchers placed a memristor between two electrodes. They changed the structure of the electrodes to get better results than the experiences already known.
They then proceeded to a change in resistance and tried to measure what was happening with an X-ray and electron diffraction. The lessons are rich. They were able to deduce that the titanium-oxygen ratio is essential for the proper functioning of the component that has higher yields when heated to a specific location (see the circle of the diagram cons).
The great advantage of this memory is that it is rapid, non-volatile and that it tolerates better the increase of etching finer than losing in the NAND write cycle for each new manufacturing process (see "Length Life of DSS in 25 nm, it would be dead "). Obviously, we are still far from a mass marketing and even if HP puts memristors market in 2013, which is far from guaranteed, it will not overtake immediately NAND or DRAM in cost of production.
However, HP could keep his hands the future memory. 
The paper was published today in the journal Nanotechnology. It is available this month on the Internet. It provides details on the yet unknown changes in strength of this component. The memristors represent the fourth passive component electrical capacitors, resistors and coils. Broadly, they have several levels of electrical resistance.
They are called memristors as they play the role of non-volatile memory. Instead of storing electrons or not they maintain a level of low or high resistance. The existence of these components has been theorized in 1971 and 2008, HP has proved the hypothesis of Professor Leon Chua. Palo Alto has already built memristors (cf.
"The first memristor in HP's 3D") and last year, he even struck a deal with Hynix for the mass production of these components in ReRAM, whose memories cells rely on memristors instead to conventional capacitors (see "HP and Hynix to sell memristors. The latest work has helped design a module with a density of 12 gigabits per square inch, manufactured in 15 nm and four eases piling on each other.
Even if HP can make memristors developed, the phenomena that occur within these components are poorly understood, mainly because the resistance changes occur in channels with a thickness of 100 nm. It's very small and it is difficult to design an experiment and measurement tools that do not alter the functioning of the memristor.
Researchers already knew that a memristor in titanium oxide (TiO2) has a high resistance. In this state, the electrons are scattered and do not pass. However, when heating the memristor, it switches to low resistance. A 10 nm channel is created in the anatase, the mineral composed of titanium oxide, and allows electrons to pass.
This tunnel is maintained even when the power comes more heat the memristor, which allows the memory to be nonvolatile. To better understand what is happening, the researchers placed a memristor between two electrodes. They changed the structure of the electrodes to get better results than the experiences already known.
They then proceeded to a change in resistance and tried to measure what was happening with an X-ray and electron diffraction. The lessons are rich. They were able to deduce that the titanium-oxygen ratio is essential for the proper functioning of the component that has higher yields when heated to a specific location (see the circle of the diagram cons).
The great advantage of this memory is that it is rapid, non-volatile and that it tolerates better the increase of etching finer than losing in the NAND write cycle for each new manufacturing process (see "Length Life of DSS in 25 nm, it would be dead "). Obviously, we are still far from a mass marketing and even if HP puts memristors market in 2013, which is far from guaranteed, it will not overtake immediately NAND or DRAM in cost of production.
However, HP could keep his hands the future memory.
- Researchers map out the switching location of a memristor (16/05/2011)
- HP Advances Next-gen 'memristor' Memory Technology (16/05/2011)
- HP Advances Next-Gen Memory Technology (16/05/2011)
- 'Computer synapse' analyzed at the nanoscale (16/05/2011)
- A New Processor For Big Data (16/05/2011)
Hewlett-Packard (geolocation) Hewlett-Packard (homepage) NYSE: HPQ (googlefinance) Hewlett-Packard Company (HPQ) (wikinvest) Hewlett-Packard (crunchbase) Hewlett-Packard (wikipedia) Memristor (wikipedia) 6-Minute Memristor Guide (youtube) Memristor and Memristive Systems Symposium (Part 1) (youtube) Finding the Missing Memristor - R. Stanley Williams (youtube) Versace Bionetics 1 (youtube)
No comments:
Post a Comment