New memristor criticism

A paper was published last month in Nature's Scientific Reports from Isaac Abraham of Intel offering some criticism of the "memristor" concept as a fundamental circuit element.

https://rdcu.be/5f3j

https://arxiv.org/abs/1808.05982

It has been over 10 years since researchers from Hewlett-Packard first claimed to have made a scientific breakthrough by discovering what had previously been proposed to be a missing fourth "fundamental" circuit element called the "memristor." Originally, I had been a big supporter of the concept and had even been invited to speak at a memristor symposium at Berkeley in 2008 where I met Leon Chua, Stan Williams, and several other researchers in ReRAM and non-linear circuit theory. However, as I studied more about the memristor several problems and inconsistencies became apparent. In 2010 I was invited to speak at the IEEE International Symposium on Circuits and Systems (ISCAS 2010) in Paris where I tried to point out these problems. It seemed to go well but shortly thereafter the technical program chair of ISCAS 2010 (Wouter Serdjin) sent a few e-mails to my university adviser informing that there was a "negative reaction" to my presentation by an unnamed person in the audience. Luckily, I had already graduated the previous semester. My ISCAS 2010 paper was then removed from IEEE Xplore.

Obviously, I'm glad to finally see some criticism appearing in the scientific literature. Advancement in science and technology depends on criticism. Particularly of those already in power. It shouldn't have taken 10 years though.

My other blog may be of interest providing a timeline of my views on the "memristor" - http://nanomorphware.blogspot.com/ 

ReRAM Patent Landscape

ReRAM (Resistance RAM) is a type of memory which may one day replace Flash as the common form of non-volatile storage for electronic devices. I recently updated a file I keep on US patents related to ReRAM and some related non-volatile memory materials and posted the results on SlideShare available at this link.

A full copy of the patent data used to create this presentation can be obtained by sending an e-mail to tinytechip@gmail.com with the subject "Resistance switching materials and device patent landscape".

Top Ten Nanotechnology Patents of 2012

The following is a list of the most interesting nanotechnology patents this past year (in my opinion):

#10 - US 8093786 - Nanoscale piezoelectrics (Stevens Institute of Technology)

This patent teaches manufacturing piezoelectric nanofibers enabling the fabrication of nanoscale sensors and actuators.

#9 - US 8289352 - Erasable printing with nanoparticles (HJ Laboratories)

This patent teaches using nanomagnetic particles as an erasible ink in a printer.

#8 - US 8121162 - Nanocrystal laser (MIT)

This patent has priority going back to 2001 and includes some basic claims for nanocrystal coatings of diffraction gratings used for optical feedback in lasers.

#7 - US 8323976 - Genetic alteration using nanodiamonds (International Technology Center)

This patent includes basic claims reciting nanodiamonds as delivery particles used in ballistic DNA injection.

#6 - US 8113437 - "Memristor" RFID (Hynix SK)

Hynix SK is HP's manufacturing partner for a form of nanoscale ReRAM which HP (incorrectly) equates to the memristor theorized by Leon Chua in the 1970's. This is the first patent from Hynix mentioning memristors and may be indicative of the first commercial application.

#5 - US 8278757 - Printed graphene electronics (Vorbeck Materials Corp./Princeton University)

This patent incudes some basic claims for using graphene as an electrically conductive ink for printed electronics.

#4 - US 8182807 - Stroke treatment using nanoparticles (University of Nebraska)

Nanoparticles have a specific advantage in delivering drugs across the blood-brain barrier and this patent (priority 2004) seems to include some basic claims important to this application.

#3 - US 8147791 - Graphene oxide reduction (Northrop Grumman Systems)

This patent includes basic claims to graphene production via the reduction of graphene oxide.

#2 - US 8323607 - High tensile strength CNT wire (Hon Hai Precision)

Carbon nanotubes have been measured to have the highest tensile strength of any material tested but in macroscopic and composite structures the collective tensile strength is greatly reduced. This patent includes some basic claims for carbon nanotube structures with a tensile strength which, while less than the highest measured value of individual nanotubes, is compatible with larger structures (i.e. wires, films) and is greater than that of steel.

#1 - US 8101149 - "Purified cage molecules consisting of carbon atoms" (Mitsubishi)

This could be the most important submarine patent relevant to nanotechnology. The priority goes back to 1990 and the patent may have a life extending to 2029 (assuming 17 years from issuance). The quote above recites in full the first claim indicating the importance of this patent to purified carbon nanomaterials.

Memristor debate: "If it works who cares?"

I have made various public arguments over the past year about why Leon Chua and HP's "memristor" models are all hype rather than a legitimate scientific model for resistance switching (e.g. The Register, EETimes, New Scientist, Wired).  The response from Martin Reynolds, an electrical engineering analyst from Gartner, is that "..it doesn't matter how it works..". Recently Dr. Paul Meuffels also pointed out flaws in the memristor model and has commented to Chris Mellor of The Register that

  "We have shown by means of a thorough analysis in terms of electrochemistry that HP’s “memristor” model is misleading. Our arguments are based upon textbook electrochemistry and can be easily reproduced. There are no real devices which would operate in accordance with HP’s model because the model is by itself in conflict with fundamentals of electrochemistry. There seems to be no way out; otherwise, somebody would have tried to refute our argumentation in the meantime. Thus, HP’s memristor research group does not have found a realistic physical model for a working memristive device."

Violin Memory CTO Jon Bennet offered the response "If it works who cares?"

Now imagine if Violin Memory tried to manufacture flash memory arrays using incorrect models of transistors. Somehow I don't think that would work out too well for product development. In engineering good models are required to manufacture reliable products. The fact that Bennet (and others) do not fully grasp this point is illustrative of either their incompetence or an inability to grasp what the "memristor" argument is really about.

In response to my comments Chris Mellor created a topic on The Register Forum asking the question "Is HP delusional over its memristor technology and IP?" (link).

The memory resistor patent thicket

Patent thickets are overlapping patent rights controlled by different corporations which often act as an impedance to commercialization of new technology. One area which seems to have a big problem with patent thickets is memory resistor technology (e.g. ReRAM, CBRAM, phase change memory) in which over 3300 US patents have already issued (more information on the patent statistics available at this link). Despite this large number of patents there has been almost no indication of significant commercial development yet (other than HP's memristor hype).

At this point the large number of patents seem to indicate problems in terms of patent thickets for any start-up company interested in developing applications of memory resistor materials. I thought it might be a good idea to point out some of the broad patents which do not seem to meet the requirements for patentability (i.e. novelty or obviousness) based on uncited prior art and which may be some of the more stickier thorns of this patent thicket. A brief summary of these patents is organized below based on the listed assignee. I filed more thorough explanations with the USPTO under 37 CFR 1.501 (linked here).

AMD
US 6864522 - This patent filed in 2003 claims an ionic memory resistor with an active and passive layer. Some of the claims are broad enough to be anticipated by a chalcogenide ionic memory resistor patent filed by Micron in 2001 (US 6348365).

Axon Technology
US 6635914 and US 6985378 - These patents filed in 2001 claims an ionic memory resistor with an oxidizable electrode. The basic claimed structure was already described back in 1988 by a company called Maxdem Inc. (US 4945257).

US 7675766, US 7728322, US 7929331, and US 8022384 - These patents with priority going back to 2000 includes claims to ionic memory resistors. The basic claimed structure was already described back in 1988 by a company called Maxdem Inc. (US 4945257) and by NASA in 1991 (US 5278636).

Hewlett Packard
US 6456525 - This patent filed in 2000 includes claims to a crosspoint array of memory resistors with a serial linear resistor. The claimed structure was already described by Caltech researchers in US 4839859 in 1987.

US 6518156 and US 6891744 - These patents filed in 2000 claims memory resistor crossbar structure with nanoscale lines. The basic memory resistor crossbar structure was already described in 1983 by Energy Conversion Devices (US 4677742). Scaling to nanoscale dimensions would have been obvious using techniques such as nanoimprint lithography (US 5772905) to improve device density.

US 6625055 - This patent filed in 2002 claims storing multibit data in individual memory resistance cells. Multibit memory resistor storage was earlier described by Energy Conversion Devices such as in US 4646266 filed in 1984.

US 6870751 - This patent filed in 2002 claims a memory resistor crosspoint array with a current concentrating feature. This was earlier described by Energy Conversion Devices in US 5687112 filed in 1997.

US 7034332 - This patent filed in 2004 claims a nanoscale memory resistor array with self-aligned rectifying elements. This was earlier described by Macronix in 2002 for phase change materials (US 6579760).

US 7443711 - This patent filed in 2004 claims a nanoscale memory resistor with tunable impedance. This was earlier described by Unity Semiconductor in 2003 (US 7038935). It is notable that the attorney arguments in this case were based on misunderstanding of the definition of electrical impedance versus electrical reactance which may also affect the validity of the patent.

US 7741638 - This patent filed in 2005 for a control layer of a memory resistor for controlling electrochemical or electrophysical reaction paths. The claimed structure was earlier described by Macronix in 2003 (US 6873541) and by Sharp in 2003 (US 6972238).

US 7763880 - This patent filed in 2007 claims a 3-terminal ionic memory resistor but this concept was earlier described by NEC in 2004 (US 2006/0164880).

US 7902869 and US 7982504 - These patents filed in 2010 claim a 3D crossbar memory resistor circuit. The claimed structure was earlier described by Toshiba in 2003 for 3D phase change memory resistor crossbars (US 7335906).

US 7985962 - This patent filed in December of 2008 claims a memory resistor structure with a metal oxide electrode. This was earlier described by Samsung in US 7417271 filed in the US in January of 2007.

US 8063395 - This patent filed in 2009 claims an ionic memory resistor with an electrode formed of amorphous material. This was earlier described by AMD in US 6838720 filed in 2003.

IMEC
US 7960775 - This patent filed in 2008 claims a metal oxide memory resistor with an oxygen gradient. This was earlier described by Sharp in 2003 (US 6972238), IBM in 2006 (US 7569459), and Samsung in 2007 (US 7417271).

Micron Technology
US 6891749 - This patent filed in 2002 claims a memory resistor structure having electrodes with different electrochemical potential. This was earlier described by AMD in 2002 (US 6781868).

Panasonic
US 7369431 and US 7826247 - These patents filed in 2004 claims determining a relationship between voltage polarity and increase/decrease of resistance in a memory resistance cell. This was earlier described in 1994 for a synaptic memory resistance cell (US 5541869).

US 7577022 - This patent filed in 2005 claims a memory resistor structure having both variable resistance and diode characteristics. This was earlier described in Energy Conversion Devices in 1984 (US 4646266), Micron in 2002 (US 6891749), and Spansion in 2004 (US 7157732).

Samsung
US 7085154 and  US 7180771, - These patents filed in 2004 claims a method for setting the pulse width duration for programming a memory resistor. The claimed method was earlier described in 1994 for a synaptic memory resistance cell (US 5541869).

US 7400027 - This patent filed in 2005 claims a multilayer memory resistor cell with a resistance changing layer and a threshold switching layer. This was earlier disclosed by Energy Conversion Devices in 1985 (US 4597162) and again by Unity Semiconductor in 2003 (US 6870755).

US 7417271 - This patent filed in 2007 claims a multilayer oxide memory resistance cell. This was earlier described by Sharp in 2003 (US 6972238).

US 7656696 - This patent filed in 2006 claims a memory resistor with a resistor to control a switching window. The claimed structure was already described by Caltech researchers in US 4839859 in 1987.

US 7935952 - This patent filed in 2008 claims a memory resistor including a threshold switching resistor, an intermediate electrode, and a resistance changing layer. This was earlier described by Unity Semiconductor in 2002 (US 6753561) and by Sharp in 2003 (US 6849564).

US 8054672 - This patent filed in 2007 claims a chalcogenide memory resistor operating without phase change and with a limited electrode contact region. This was earlier described by Axon Technology in 2002 (US 6798692).

Sandisk
US 7733685 - This patent filed in 2008 claims a memory resistor structure sandwiched between distributed diodes. This was earlier described in 2006 (US 7378870).

US 8208282 - This patent with priority going back to 1998 claims a memory resistor cell including a vertically aligned steering element (e.g. diode) and state change element. This was earlier described by Energy Conversion Devices in 1991 (US 5296716).

Sony
US 7583525 - This patent filed in 2007 claims a method of driving a memory resistor by applying multiple voltage pulses to change the resistance state. The claimed method was earlier described in 1994 for a synaptic memory resistance cell (US 5541869).

Spansion
US 7183141 - This patent filed in 2004 claims a method of reversibly programming a memory resistor. This was earlier described by researchers at CalTech in 1990 (US 5272359).

US 7564708 - This patent filed in 2006 claims an ionic memory resistor including an active and passive layer. This was earlier described by Micron Technology in 2001 (US 6348365) and by AMD in 2003 (US 6864522).

US 7646624 - This patent filed in 2006 claims selecting an electrode material of a memory resistor to determine the on-state resistance. This was earlier described by Micron Technology in 2002 (US 6891749).

Unity Semiconductor
US 6870755 - This patent filed in 2003 claims a reversibly writable memory resistor. This was earlier described by Energy Conversion Devices such as in US 4597162 filed in 1985.

US 6906939 - This patent filed in 2003 claims a 3D stacked memory resistor array. This was earlier described by Energy Conversion Devices such as in US 5912839 filed in 1998.

Fundamental Issues and Problems in the Realization of Memristors

It appears that some in the scientific community are beginning to catch on to the problems of HP's "memristor" models. A recent preprint was posted in arXiv by Paul Meuffels and Rohit Soni criticizing Leon Chua and HP's memristor framework on the grounds that it does not obey the "no energy discharge property" and is not consistent with principles of thermodynamics (link). Dr. Meuffels co-authored an earlier paper in Applied Physics A pointing out an incorrect assumption about ionic conductivity by the HP Labs group in one of their early memristor papers (link).

Back in 2010 I was invited to speak at the IEEE International Symposium on Circuits and Systems during a special session on memristors. At the time I noted some of the inconsistencies with HP's and Chua's arguments including that it is not properly considered a fourth fundamental circuit element and it was never missing as claimed by Chua (link). Last year I also posted a paper on arXiv pointing to more realistic dynamic systems models for thin film resistance switching devices (link). Hopefully the efforts of Dr. Meuffels and others will be constructive to move beyond the hype of HP and Chua and open the doors for more realistic models applicable to ReRAM and other resistance switching technologies.

US Patent 8233017 - Electrostatic printing with nanocarbon image generator

http://www.freepatentsonline.com/8233017.html

Xerox continues its efforts in using nanostructured materials to improve the operability of electrophotographic printers. This latest patent teaches using graphene or carbon nanotubes in forming an electrostatic image for smaller, more energy efficient printers. Claim 1 reads:

1. An electrostatic latent image generator comprising:

a substrate;

an array of pixels disposed over the substrate, wherein each pixel of the array of pixels comprises a layer of one or more nano-carbon materials, and wherein each pixel of the array of pixels is electrically isolated and is individually addressable; and

a charge transport layer disposed over the array of pixels, wherein the charge transport layer comprises a surface disposed opposite to the array of pixels, and wherein the charge transport layer is configured to transport holes provided by the one or more pixels to the surface.

US Patent 8232722 - Nanocrystal LED with reduced radiative loss

http://www.freepatentsonline.com/8232722.html

This patent from MIT teaches using nanocrystals to form an LED structure including a lightly doped layer to improve the efficiency of photon generation. Claim 1 reads:

1. A light emitting device comprising:

a first charge transporting layer including a first inorganic material in contact with a first electrode arranged to introduce charge in the first charge transporting layer;

a second electrode;

a plurality of colloidally-grown semiconductor nanocrystals disposed between the first electrode and the second electrode; and

a lightly doped layer proximal to the nanocrystals, wherein the lightly doped layer is doped in an amount effective to minimize non-radiative losses due to exciton quenching by unbound charge carriers in the transporting layer.