Data Storage Systems Center

In this project we are developing a model that describes the electric field-driven resistance switching behavior demonstrated by perovskite oxides. Switches made from these resistance change materials can have an OFF to ON resistance ratio of as high as 107. Our goal is to discover the fundamental mechanism that is at work behind this resistance switching phenomenon. In our model, we identify mobile oxygen vacancies as the key feature that leads to the observed switching behavior. In our previous work, we have demonstrated, under some limiting conditions, how motion of oxygen vacancies can lead to hysteretic current-voltage relationship in these devices which gives rise to the bi-stable resistance characteristics. In our recent work, we are exploring how electric field dependent mobility of oxygen vacancies affect the vacancy concentration profile inside these devices and how that leads to hysteresis is current-voltage relationship.

In this project we are developing a model that describes the electric field-driven resistance switching behavior demonstrated by perovskite oxides. Switches made from these resistance change materials can have an OFF to ON resistance ratio of as high as 107. Our goal is to discover the fundamental mechanism that is at work behind this resistance switching phenomenon. In our model, we identify mobile oxygen vacancies as the key feature that leads to the observed switching behavior. In our previous work, we have demonstrated, under some limiting conditions, how motion of oxygen vacancies can lead to hysteretic current-voltage relationship in these devices which gives rise to the bi-stable resistance characteristics. In our recent work, we are exploring how electric field dependent mobility of oxygen vacancies affect the vacancy concentration profile inside these devices and how that leads to hysteresis is current-voltage relationship.

NAND flash memories have developed a significant presence in the non-volatile memory (NVM) market and are now attempting to move into the computer storage market in the form of Solid State Drives (SSDs). Although NAND flash based SSDs offer lower power, faster speed, and better mechanical reliability, they face a challenge in overcoming the price advantage of HDDs due to their dependence upon lithographic resolution as well as fundamental physical limitations beyond the 22nm process node. Thus, to replace HDDs, alternative non-volatile memory technologies that can overcome shortcomings of NAND flash and compete on a cost per TB basis with HDDs must be found.

In this poster, a dozen different alternative NVM technologies are studied and evaluated with regard to cost, scalability, performance and likelihood of success in 2020.

Magnetic hard disk drives (HDD’s) recently entered the 1.5TB range. Further advances in areal density will enable yet higher capacities; however, in order to overcome density limits imposed by the inability to write on high coercivity media, new technologies such as heat-assisted magnetic recording and/or patterned magnetic media are being investigated.

Solid state drives (SSD’s) using NAND flash memory are today competing with HDD’s in mobile, low-capacity applications and are threatening to make inroads into high performance enterprise applications as well. The SSD provides very high robustness against mechanical vibrations, fast read access time and low power consumption, but has a relatively high cost per gigabyte compared to a HDD.

As the NAND flash memory scales down beyond 20nm, the memory cell integration will encounter large challenges due to degraded device performance and lithography limitations. New types of non-volatile memory will need to be developed in order to continue scaling density and capacity as HDD’s have done for the past 50 years.

In this poster, the device performance, cost and physical challenges of emerging non-volatile memory devices such as phase change memory, spin-transfer-torque memory, racetrack memory and molecular memory, are analyzed and compared to those of HDD’s and SSD’s.

In traditional optical and magnetic data storage, bits are stored at only a 2D cross section of the medium. Although multi-layer media is common, such as using both sides of a magnetic disc or fabricating multiple layers in an optical disc, true volumetric recording is regarded as data storage throughout a uniform material.

Volumetric data storage is categorized into to forms, holographic and bit-wise. In holographic data storage two coherent light beams recombine and interfere to create a pattern of high and low intensity areas. Pages of bits are stored in the interference pattern by passing one beam through a mask with high and low transmission areas representing the data to be written (spatial light modulators). Readback occurs by exposing the desired page with the reference beam and detecting the resulting pattern with a CCD. In bit-wise recording each bit is recorded individually at a point in the volume where a beam comes into focus. The high fluence at the focal point modifies the material properties of the media, i.e. absorption or fluorescence, while leaving the surrounding volume unchanged. Readback is achieved through confocal microscopy allowing for high depth resolution.

Many of these technologies have matured significantly in recent years. The focus of this project will be to evaluate the feasibility of product commercialization by collecting BER data through spin stand experiments.

Data are continuously piling up in the modern world at higher rate every day. The HDDs (Hard Disk Drives) have changed the world of computing dramatically during the past 20 years. However, they have recently suffered from the physical limitations of the media & head and the rotating disks and other moving parts. The flash-based SSDs (Solid State Drives) offer a feasible alternative to the traditional HDDs. They have been used in military, aviation, enterprise applications due to their rapid speed and robust ability to perform in extreme conditions. But they have not been used in commercial applications of the general consumers these days due to the pricing. Even though The price ($/GB) gap between HDD and SSD is getting smaller steadily, it is still relatively large. However, Si-based FETs (Field Effect Transistors) could be scaled down up to 10 nm but they will face big challenges with leakage current, high K materials, power consumption and reliability of complex process integration due to physical and lithography limits. Therefore, there is no doubt that we need the alternative technologies to replace the HDD and SSD in the future. Those potential technologies may not be limited to the structure, the recording media and the mechanism of the current devices. In this poster, the several emerging non-volatile memories are introduced and summarize, which provide new opportunities for further scaling and cost reduction in quantom/nano/bio technology era.