This is the second post in a series on Non-Volatile Memory Trends. In the previous post, we prepared a foundation by explaining the basics of Non-Volatile Memory and asked an important question: will Moore’s law continue to hold true for Flash Memory areal densities for another decade? One cannot be sure, but certain innovations seem to be giving it a lease of life for the moment. We will examine these innovations in the remaining posts of this series, starting with 3D NAND.
The Limits of Current Flash Technologies
In the previous post, we discussed how SLC, MLC, TLC, and QLC Flash technologies achieve scale over their predecessor by encoding more bits in a cell. However, the context for all these technologies is Planar (Two-Dimensional) NAND. By planar, we mean the cells are arranged in a single two-dimensional physical layer. With such a geometry, we can either increase capacity by packing more bits into a cell, or by packing more cells in the same physical space. We have already seen the limitations of packing more bits into a cell in the previous post. Similarly, the second option of packing more cells is also problematic. To pack more cells, we need to decrease the size of an individual cell; however, the feature size of 2D Flash has shrunk significantly over the years from 1500 nm in 1990 to about 15 nm today. Consequently, the number of electrons used per bit have reduced from about 105 to under 100. It’s virtually impossible to reduce the number further because electrons are comparatively large, fuzzy, quantum-mechanical particles that get restless when confined to very small regions, making them susceptible to leakage and disturbance from nearby cells.
2D Flash technology seems to have reached a point where Moore’s law can no longer apply unless we use the physical medium in a fresh and innovative way.
3D NAND, a technology being pioneered by several manufacturers like Toshiba, Hynix, and Samsung, overcomes limitations of 2D NAND by stacking cells vertically one on top of another to achieve higher areal density. It’s similar to how metropolises like New York accommodate more people by growing vertically rather than reducing the size of apartments. 3D NAND technology, which is also often referred to as V-NAND (Vertical NAND), can presently create 24 to 32 vertical layers, yielding significantly higher areal densities. However, 3D NAND is more than just a change in geometry. It also employs structural as well as material related innovations.
In terms of material, traditional (2D) Flash uses floating-gate MOSFET technology to hold charge, while 3D NAND uses Charge Trap Flash (CTF) technology.
Charge Trap Flash (CTF) is a semiconductor memory technology used in creating non-volatile NOR and NAND flash memory. The technology differs from the more conventional floating-gate MOSFET technology in that it uses a silicon nitride film to store electrons rather than the doped polycrystalline silicon typical of a floating gate structure… — Wikipedia
Structurally, 3D NAND wraps each planar charge trap cell into a cylindrical form.
An individual memory cell is made up of one planar polysilicon layer containing a hole filled by multiple concentric vertical cylinders. The hole’s polysilicon surface acts as the gate electrode. The outermost silicon dioxide cylinder acts as the gate dielectric, enclosing a silicon nitride cylinder that stores charge, in turn enclosing a silicon dioxide cylinder as the tunnel dielectric that surrounds a central rod of conducting polysilicon which acts as the conducting channel. — Wikipedia
This combination of geometrical, structural and material innovations in 3D NAND have resulted in significantly higher areal densities (about 10x as compared to planar NAND). Since some designs can also pack as much as 3 bits per cell pushing the total capacity even higher, 3D NAND design can lead to 2 – 4 TB on a thumb drive, 15 TB on a 2.5 inch form factor device, and 60 TB in a 3.5 inch device. Additionally, increased performance is not the only benefit of these innovations. They also give reduced cross-cell interference, better performance, and reduced power consumption.
So is 3D NAND the future of Flash? Intel and Micron are both betting big on the technology. 3D Xpoint SSDs are claimed to be 10x denser and 1000 times faster than traditional SSD devices; although, in reality, these speeds have not yet been achieved due to interface limitations. These companies are also claiming that these devices will be durable enough to make them last forever (at least from a practical standpoint). However, not everyone is as optimistic. Increasing costs of manufacturing have put a huge question mark on the viability of Flash technology itself. It’s hard to say if the future of storage lies in 3D NAND or in competing technologies such as MRAM, PRAM, DWM, and Atomic Memory. However, according to the MediaSilo blog, the immediate future does seem to have 3D NAND written in it:
For now, and in the coming years, it is safe to say that we can expect to see 3D NAND SSDs being used to store our growing iTunes libraries and photo galleries, as well as our HD and increasing UHD footage.
In the next post, we will discuss Phase Change Memory; the technology which unlike Flash is claimed to be immune to memory wear.