News & Analysis
Comment
resistion
Also ionic charge should have different effect from electronic charge, but it's ...
resistion
They put up all those tilted figure 8 I-V's, each point is a different ...
HP, Hynix to commercialize the memristor
Mark Lapedus
8/31/2010 6:23 PM EDT
SAN JOSE, Calif. - Seeking to commercialize its memristor technology, Hewlett-Parkard Co. has entered into a joint development agreement with South Korea's Hynix Semiconductor Inc.
HP and Hynix will jointly develop new materials and process integration technology to transfer HP's memristor technology from R&D to commercial development in the form of resistive random access memory (ReRAM).
The deal is non-exclusive, according to HP, who said HP may work with others in the ReRAM arena.
HP itself does not want to be in the ReRAM business, said Stan Williams, senior fellow at HP and founding director of the Information and Quantum Systems Laboratory at HP Labs.
Eventually, HP hopes to use ReRAMs in its own, undisclosed products, Williams told EE Times. Initially, on the chip front, the company is working with Hynix. Then, HP hopes to work with other memory makers. This will allow the industry to purchase ReRAMs at competitive prices, he said.
Hynix will implement the memristor technology in its research and development fab. Hynix is hedging its bets, as the company is working on several rival memory technologies. Hynix and Samsung Electronics Co. Ltd. are jointly working on rival MRAM technology. Hynix and Grandis Inc. are also working on a next-generation MRAM technology called spin-transfer torque RAM.
FeRAM, MRAM, phase-change and ReRAM are next-generation memory technologies.
ReRAM is non-volatile memory with low power consumption that holds the potential to replace flash memory; it also has the potential to serve as a universal storage medium – that is, memory that can behave as flash, DRAM or even a hard drive, according to HP.
End-user products based on ReRAMs are due out by the end of 2013, Williams said. ''This is a darkhorse technology,'' Williams said. ''We think this will break out of the pack.''
HP and Hynix have not defined the first end-user products based on ReRAM. In any case, ReRAMs are ideal for solid-state storage, main memory for PCs and other products, he added.
The memristor, short for “memory resistor,” was postulated to be the fourth basic circuit element by Leon Chua of the University of California at Berkeley in 1971. It was moved into practice by researchers in HP Labs.
Earlier this year, HP announced the discovery that the memristor also can perform logic, showing that memristor-based devices could change the standard paradigm of computing by enabling computation to one day be performed in chips where data is stored, rather than on a specialized central processing unit.
HP has been talking about the technology for years, but it has yet to commercialize it. Others are also working on ReRAM, including 4DS, Adesto, Unity and several startups.
In order to explore the scaling limitations of conventional flash memory cells European research institute IMEC recently started looking at resistive RAM (RRAM) cells. Five of the leading memory makers — Samsung, Hynix, Elpida and Micron Technology — are involved in the IMEC core CMOS research program and are set to share the cost and benefit from the results of the research.
Resistive switching memories are based on materials whose resistivity can be electrically switched between high and low conductive states. RRAM is becoming of interest for future scaled memories because of their superior intrinsic scaling characteristics compared to the charge-based flash devices, and potentially small cell size, enabling dense crossbar RRAM arrays using vertical diode selecting elements. RRAM is seen as a potential candidate to replace conventional flash memory at or below the 22-nm manufacturing process technology node.
HP and Hynix will jointly develop new materials and process integration technology to transfer HP's memristor technology from R&D to commercial development in the form of resistive random access memory (ReRAM).
The deal is non-exclusive, according to HP, who said HP may work with others in the ReRAM arena.
HP itself does not want to be in the ReRAM business, said Stan Williams, senior fellow at HP and founding director of the Information and Quantum Systems Laboratory at HP Labs.
Eventually, HP hopes to use ReRAMs in its own, undisclosed products, Williams told EE Times. Initially, on the chip front, the company is working with Hynix. Then, HP hopes to work with other memory makers. This will allow the industry to purchase ReRAMs at competitive prices, he said.
Hynix will implement the memristor technology in its research and development fab. Hynix is hedging its bets, as the company is working on several rival memory technologies. Hynix and Samsung Electronics Co. Ltd. are jointly working on rival MRAM technology. Hynix and Grandis Inc. are also working on a next-generation MRAM technology called spin-transfer torque RAM.
FeRAM, MRAM, phase-change and ReRAM are next-generation memory technologies.
ReRAM is non-volatile memory with low power consumption that holds the potential to replace flash memory; it also has the potential to serve as a universal storage medium – that is, memory that can behave as flash, DRAM or even a hard drive, according to HP.
End-user products based on ReRAMs are due out by the end of 2013, Williams said. ''This is a darkhorse technology,'' Williams said. ''We think this will break out of the pack.''
HP and Hynix have not defined the first end-user products based on ReRAM. In any case, ReRAMs are ideal for solid-state storage, main memory for PCs and other products, he added.
The memristor, short for “memory resistor,” was postulated to be the fourth basic circuit element by Leon Chua of the University of California at Berkeley in 1971. It was moved into practice by researchers in HP Labs.
Earlier this year, HP announced the discovery that the memristor also can perform logic, showing that memristor-based devices could change the standard paradigm of computing by enabling computation to one day be performed in chips where data is stored, rather than on a specialized central processing unit.
HP has been talking about the technology for years, but it has yet to commercialize it. Others are also working on ReRAM, including 4DS, Adesto, Unity and several startups.
In order to explore the scaling limitations of conventional flash memory cells European research institute IMEC recently started looking at resistive RAM (RRAM) cells. Five of the leading memory makers — Samsung, Hynix, Elpida and Micron Technology — are involved in the IMEC core CMOS research program and are set to share the cost and benefit from the results of the research.
Resistive switching memories are based on materials whose resistivity can be electrically switched between high and low conductive states. RRAM is becoming of interest for future scaled memories because of their superior intrinsic scaling characteristics compared to the charge-based flash devices, and potentially small cell size, enabling dense crossbar RRAM arrays using vertical diode selecting elements. RRAM is seen as a potential candidate to replace conventional flash memory at or below the 22-nm manufacturing process technology node.
|
|
|
|
|
Navigate to related information
goafrit
8/31/2010 8:29 PM EDT
Good idea from HP. This is going to become a commodity business. I do not think it is safe to say that this is HP technology. Memristor has existed for more than 7o years. Even the paper that HP depended for their works exist, so I hope they are not claiming that this is HP thing. It is like someone trying to patent the Pythagorean Theorem, today
Sign in to Reply
junko.yoshida
9/1/2010 7:01 AM EDT
Hi, Goafrit. Just to set the record straight here. EE Times reported on Memristor back in 2008:
The long-sought after memristor--the "missing link" in electronic circuit theory--has been invented by Hewlett Packard Senior Fellow R. Stanley Williams at HP Labs (Palo Alto, Calif.)
Memristors--the fourth passive component type after resistors, capacitors and inductors--were postulated in a seminal 1971 paper in the IEEE Transactions on Circuit Theory by professor Leon Chua at the University of California (Berkeley), but their first realization was just announced today by HP.
According to Williams and Chua, now virtually every electronics textbook will have to be revised to include the memristor and the new paradigm it represents for electronic circuit theory.
"My situation was similar to that of the Russian chemist Dmitri Mendeleev who invented the periodic table in 1869," said Chua. "Mendeleev postulated that there were elements missing from the table, and now all those elements have been found. Likewise, Stanley Williams at HP Labs has now found the first example of the missing memristor circuit element."
Read the whole story entitled "
'Missing link' memristor created: Rewrite the textbooks?" here:
http://www.eetimes.com/electronics-news/4076910/-Missing-link-memristor-created-Rewrite-the-textbooks-?pageNumber=0
Sign in to Reply
resistion
9/1/2010 7:26 PM EDT
Thanks for the background. Yet, since it is so basic and universal, maybe every device in use today already incorporates some memristor, if unknowingly? The fact that Flash memory needs wear leveling, for example?
Sign in to Reply
Baolt
9/2/2010 6:32 PM EDT
Junko thanks for backround in detailed. However i wonder why lecturer insists to not teach memristor as 4 element of electronic. Here is an interesting post of spectrum, ( http://tinyurl.com/2g3hpn4 )which i believe details everything clearly. Even the experts couldnt believe that it would be feasible and real one day, it was just a theory.
Its clearly time for all lecturers all over to world to learn&teach to new generations of electronics more about missing element. So than it could be accepted and implemented widely.
Thanks to HP and Hynix to break down such thoughts like " "I don't think the memristor will be taught in undergraduate courses until it is widely adopted in industry for the simple reason that any circuit containing even only one memristor must be analyzed by nonlinear techniques,"
Sign in to Reply
mark.lapedus
8/31/2010 8:45 PM EDT
Will the memristor make it or not? It's interesting stuff.
Sign in to Reply
selinz
8/31/2010 8:57 PM EDT
Check out the Youtube video by Stanley williams of HP at http://www.youtube.com/watch?v=bKGhvKyjgLY . Here is a highlight "... I believe that Memristers, at some point in time, and it could be as soon as 5 years from now could literally completely replace both DRAMS and hard disk.. And perhaps eventually CD's and DVD's." (and presumably bluray :-) Wow, that's saying something. Interesting stuff...
Sign in to Reply
selinz
8/31/2010 9:01 PM EDT
Another juicy tidbit from the video.. 1 petabit/cm2
Ok, that's a lot...
Sign in to Reply
resistion
8/31/2010 9:12 PM EDT
ReRAM or RRAM has some advantages like simple structure and use of fairly accessible materials. The key question is if any special material property is needed, for example ionic conductivity. It's possible any material can be an RRAM or memristor, then what happens? Another key focus should be the architecture for RRAM, as information is now stored as resistance, so any other resistance in series or parallel is considered parasitic. That said this is much simpler than trying to deduce how much charge is stored. To be honest, I think HP's main contribution is not in the RRAM or ReRAM aspect but the potential for logic operations. But it doesn't seem obvious the memristor would be a substitute for a transistor.
Sign in to Reply
Silicon_Smith
9/1/2010 3:32 AM EDT
Hustle to preserve the validity of Moore's law! The concept is interesting and has been fanning fantasies for long now. Whether this could be the node for Neural Networks, integration of Memory with CPU and ultra-dense storage et al. However, I havent read much about how good it fares on commercial terms as against the good old Flash and HDD. I mean, we have been hurtling at a good pace in Flash with 3D and more information per switch, and though there is a physical limit, will the benfits of the new technology be worth the cost? Rice fellows have shown confidence regarding the advance based on SiO2, but there arguments seem like they will take another five years to market. The phase change memory still seems to be more stable as against other novelties.
Sign in to Reply
resistion
9/1/2010 5:29 AM EDT
The Rice U SiOx result should make the Floating gate guys really nervous! Some groups tried NiO but don't feel so comfortable as with PCM. HP uses TiOx for the memristor.
Sign in to Reply
Sanjib.Acharya
9/1/2010 7:10 AM EDT
Thanks for the information provided in this article about another next generation memory technology ReRAM and for the thoughts shared by the readers. Looks like ReRAM is lagging behind MRAM with respect to the schedule for getting commercially available in the market? It would be great if some of our friends having expertise in this area could draw a comparison between these technologies: MRAM vs ReRAM.
Sign in to Reply
greenpattern
9/1/2010 7:25 AM EDT
MRAM and PCM are two more well-established types of ReRAM (resistance-based random access memory). MRAM is commercialized by several companies like Crocus, Grandis and Everspin. PCM is commericalized by Numonyx (Micron). Also some work by BAE for aerospace. A potential issue with these is the sensitivity to magnetism and temperature respectively.
Sign in to Reply
double-o-nothing
9/1/2010 9:00 PM EDT
It's true these are resistance-based technologies, but ReRAM or RRAM is usually referring to those systems where the conduction path effectively disappears and reappears within a normally expected insulator. With MRAM you always have tunneling current density. With PCM you have a semiconductor changing resistance.
Sign in to Reply
greenpattern
9/2/2010 12:04 PM EDT
Spin torque mram and pcm or pram qualify as memristors in strict sense.
Sign in to Reply
selinz
9/1/2010 1:35 PM EDT
With regard the comment about "any material can be a memrister." My understanding is that the requirements is that the ionic movement results in an a signficant change in electrical conductivty. Well, this is certainly true of many conducting polymers. And certain other cathode/anode matials in solid state batteries... But generally the polymer becomes charged with the concomitant movement of an associated anion (or less frequently, a cation). It seems like electroformed conducting polymers would be an ideal candidate. So while it's a stretch to say "any material," I do think that there are a large number and variety to choose from (beyond the TiO2-Ti4O7 material being studied by HP).
Sign in to Reply
resistion
9/1/2010 7:09 PM EDT
Yes, as hm reminded that memristance is the fourth basic property, like resistance, capacitance, inductance, independent of material, should we expect any material to have its particular memristance?
Sign in to Reply
hm
9/1/2010 2:03 PM EDT
As memristor is fourth basic electric element, it will have many potential new different applications. What these other applications will be? Will it be able to help make progress in solving problems related to Alzheimer and Parkinson’s diseases?
Sign in to Reply
Code Monkey
9/1/2010 2:14 PM EDT
I was going to write something interesting, but I forgot what it was. Flash has a lot of process steps and time-consuming test needs (retention bake, test patterns, etc.), so I'll be glad to see it replaced by something more economical. Writes will be much faster. In 10 years I might have ReRAM in my pocket, on my 1TB thumb drive.
Sign in to Reply
David Ashton
9/4/2010 8:23 PM EDT
I'll bet 1TB thumb Drives come a lot sooner than 10 years. By then we'll be into Petabytes, probably.
Sign in to Reply
Gil Russell
9/1/2010 5:51 PM EDT
I am curious about financial details. HP Labs may be looking at commercialization activities as a way to finance their long term research goals.
I am left wondering whether HP's long term goal is not so much the ReRAM per se, but a derivation of the technology in the form of High Density Configurable Logic (HDCL)and a nicely fattened IP portfolio.
Should Xylinx and Altera be concerned?
As usual this will all take much longer than anyone at first anticipates as nothing is easy...,
Sign in to Reply
iniewski
9/1/2010 7:40 PM EDT
thank you Junko for straightening memristor story...this story is very intriguing...commercializing a completely new technology in 2-3 years sounds almost impossible, this is not a process shrinkage, new gate material, or new dielectric stack, this is something completely new and not well understood, there is not a single electronic textbook that covers this 4th missing electronic component (after R, L, C)...hoping that you can derive millions in revenue at 22nm node seems like a big stretch to me, I am putting my money on flash ;-)...Kris
Sign in to Reply
Helicopter
9/2/2010 1:30 AM EDT
me too
3d mlc nand 2012
3d tlc nand 2014
4d pram 2017 ?
Sign in to Reply
resistion
9/2/2010 1:27 AM EDT
Ok, hold on guys. Was just pointed out to me, memristance means the resistance is a function of the charge that has passed through. In that case, this would not make a good resistance-based memory, as each time you read it, you are passing charge through it!
Sign in to Reply
hlavac
5/18/2011 5:09 AM EDT
Not only charge passed, but its direction too.
You just read it using AC current pulse to keep the resistance unchanged.
Sign in to Reply
resistion
5/18/2011 6:42 AM EDT
They put up all those tilted figure 8 I-V's, each point is a different resistance, depending where you stop!
Sign in to Reply
resistion
5/18/2011 8:00 AM EDT
Also ionic charge should have different effect from electronic charge, but it's not distinguished in the memristance concept.
Sign in to Reply
unknown multiplier
9/2/2010 2:34 AM EDT
I don't think HP's stuff will commercialize soon, and Hynix still wants to sell flash. But the crossbar architecture looks interesting, looks cheaper than flash.
Sign in to Reply
http://www.lulu.com/spotlight/poconoarmchairreview
9/2/2010 2:59 AM EDT
Who will write the new electronics textbooks covering memristors? Contact me if you are up to it! My e-mail is rich (over here at) richardkrajewski.com.
Sign in to Reply
kinnar
9/2/2010 6:19 AM EDT
eeTimes is a great site, I just come across this great invention going on. ReRAM is something very new for me. I came across it at this site only. But this memory has got very good potential of changing the present day use of Memories and Storage. Hope it will get standardized earlier. But since the giants are after it this memory will come with some royalties, if it is like that then will not survive in the market rather will not be able to get positioned in the market.
Sign in to Reply
p51
9/2/2010 11:12 AM EDT
Interesting this device - one whose resistance can alter with charge. However I have difficulty imagining a scenario where this can be used in a circuit. Can someone elaborate a bit more on the possible applications already mentioned in the article? Thanks!
Sign in to Reply
3D Guy
9/3/2010 12:54 AM EDT
I guess it helps that HP buys a lot of DRAM from Hynix :-) If a big customer of yours comes to you and asks you to license their technology for a small sum, it makes sense to oblige.
Hynix is also way behind on RRAM compared to Samsung (who has been working on it since 2000, and has a lot more work/IP on it than HP). So it makes sense for them to use HP's know-how to be able to start competing with Samsung.
Sign in to Reply
mark.lapedus
9/7/2010 1:38 AM EDT
So which will make it? FeRAM, MRAM, PRAM or RRAM? My friend, Bob Merritt, says they will all find a home. Any thoughts?
Sign in to Reply