(a) Schematic of the unipolar switching in SiOx, showing the typical set (red) and reset (blue) I-V curves (9). (b) Schematic of the SiOx thin-walled structure for in situ TEM imaging. The switching region in SiOx is localized by the nanogap generated in the α-C layer and imaged by TEM. (c) and (d) Schematics of the multi-stage structure and its arrangement with respect to the TEM stage. The pink arrows indicate the TEM electron beam imaging direction. Credit: Yao et al.; Nature Science Reports.
Two separate but recent stories underline the versatility of silicon oxides and involve two very different digital memory applications.
The first story comes from Rice University, where a team of researchers continues to build on a discovery several years ago of how to make a reversible silicon/silicon-oxide nanowire switch. This tale begins around 2008 when a team lead by James Tour (Rice’s T.T. and W.F. Chao chair in chemistry and professor of mechanical engineering and materials science and of computer science) reported a nanometer reversible switching ability in a set up where ribbons of graphite had been placed on a SiOx substrate. They assumed that the breaking and reformation of the graphite ribbons was the basis of the switching.
When we first wrote about this group’s work two years ago, grad student Jun Yao had doubts about the role of the graphite, thinking that the SiOx was really the key ingredient. To prove his point, Yao had demonstrated how to build a switching “sandwich” composed of a layer of SiOx between semiconducting sheets of polycrystalline silicon. He argued that the switching occurs through the creation of 5 nm silicon filaments that can be repeatedly broken and reconnected by applying a pulse of varying voltage through the silicon oxide.
(Yao presented more evidence to support his theory about the role of the SiOx in January in an online paper published as part of Nature’s Scientific Reports. This paper presented TEM evidence of the growth and shrinkage of silicon nano crystals, with the silicon coming from the SiOx matrix.)
Conceptually, Yao’s sandwich was intriguing because it could operate as a two-terminal resistive switch, not three (as in flash memory), because the physical process does not require the device to hold a charge. It also held promise as a platform for building 3D memory units. Scalability was an issue, and the breakthrough in 2010 was that the group was able to build a prototype with 1,000 memory elements. The 5 nm filament size caught the eye of electronics makers because the state-of-the-art has been 22 nm connections.
The latest breakthrough is that Tour’s group and Yao (who is now doing post-doc work at Harvard) have demonstrated a new type of SiOx sandwich, where outer single layers of graphene crossbars serve as the terminals. This results in transparent, two-terminal memory material that can be formed on a flexible plastic substrate that are again suitable for 3D memory packages. They also report that indium tin oxide can be used as the electrode material. Another benefit, they say, is that these packages are suitable for form-fitting applications, ranging from hand-held devices to transparent heads-up displays. Finally, a Rice press release reports that the materials are radiation and heat tolerant (up to about 700°C), and prototypes are being tested at the International Space Station. More details are available in a new Nature Communications paper, “Highly transparent nonvolatile resistive memory devices from silicon oxide and graphene” (doi:10.1038/ncomms2110).
Tour recently discussed the these developments in a news conference held at an August American Chemical Society meeting:
100-Million-year memory in fused silica?
The second silicon oxide memory story has to do with an announcement from Hitachi that the company had perfected a way to use fused silica glass to create a “semi-perpetual” memory material, with an information density a little better than a CD, which could be decoded—if necessary— with something as simple as an optical microscope.
Hitachi’s idea is to use fused silica because of its rugged and waterproof characteristics. The company says, “[I]t was confirmed that data could be read without degradation even after subjecting the fused silica to an accelerated temperature test of 1,000℃ for 2 hours, indicating that long-term data storage in the order of several hundred million years is possible.”
Hitachi admits that the idea of using fused silica as a memory medium goes back to 2009, and that the company had previously figured out how to crudely and slowly encode and decode the system. One big problem was that the decoding mechanism required a computed tomography–LED system. The company understood that this mechanism could end up being too complex in some far-into-the-future scenarios.
So, to be practical, Hitachi spent the next few years devising a data-dense system that could be rapidly encoded and easily decoded. The company reports
To resolve these issues, Hitachi has developed multilayer recording technology and simultaneous multibit recording technology to enable high-speed high-density recording. Furthermore, read technology using a conventional commercially available optical microscope was developed, achieving simple access to the recorded data.
The Hitachi system stores binary information in a series of dots (and “no-dots”) placed in the glass material using a femtosecond laser. The encoding laser is also equipped with a spatial light modulator that made it “capable of recording a batch of one hundred dots.”
Credit: Hitachi.
Regarding the use of a microscope to read the data, the company reports
Normally, if an optical microscope is used to observe multilayer recorded fused silica, the image of the dots in the other layers appear as noise, reducing the image quality of the layer being read. To enable the data recorded in each of the four layers to be accurately read, technology was developed to enhance contrast by using two data block images taken at different focal points. Further, by emphasizing the outline of the dots through signal processing, a signal to noise ratio of 15 dB, equivalent to zero read error, was achieved for all four layers.
Hitachi says it plans on elaborating on this system by working to further increase the recoding density.
A story from the AFP New Service reports that Hitachi is demonstrating prototypes two centimeters square and two millimeters thick, and says the company hasn’t decided on a route to commercialization, but believes it would start with offering it as a storage service for government agencies, museums and religious groups.