Currently the most used storage memory is not hard drives but what we call non-volatile RAM, which has the particularity of not losing data when they are turned off. However, for some time now we have seen how two different approaches are used. On the one hand the planar NAND Flash memory or 2D NAND and on the other the 3D NAND. How do they stack up when it comes to scaling in storage?
What is 3D NAND memory?
3D NAND memory has long been used in several flash memory storage chips, this type of non-volatile memory is based on stacking several chips vertically making use of paths through silicon that cross the entire stack and allow communication of the different layers with the interface.
The fact of being able to vertically stack flash memory chips means that storage grows not only in the classic two dimensions or matrix of all memories but also in three dimensions, allowing an increase in storage capacity that would otherwise be only possible using more advanced manufacturing nodes and therefore make use of smaller transistors.
To make the silicon paths pass vertically, the 3D NAND memory chips, unlike the standard NAND ones, have holes throughout the surface of the chip, which completely go through the different layers of flash memory chips that make up the stack.
3D NAND memory cells are different from 2D
A memory cell is the structure in the form of transistors in which we store one bit of memory in a conventional or non-volatile RAM. In the case of 3D NAND, its structure is different from the standard NAND, since it has a cylinder-shaped structure with a hole in the central part that is where the silicon pathway passes.
Unlike 2D NAND, which increases capacity by using narrower transistors, 3D NAND makes use of less advanced manufacturing processes. This has the advantage that flash memory factories do not have to be redone with new machinery to manufacture smaller transistors, but can scale by simply adding more and more layers to the structure, thereby increasing the memory density per flash memory chip. .
However, the number of layers that a 3D NAND memory stack has is limited and although it has advantages in the cost of the manufacturing process, it also represents an extra cost in other aspects with respect to increasing the storage capacity of flash memory.
How do they stack up when it comes to storage?
With all this we come to a key question: How do both types of NAND Flash memory compare in terms of capacity evolution? The fact that 2D NAND memory uses more advanced manufacturing nodes makes a direct comparison between the two types of memory difficult in terms of storage capacity.
The 3D NAND memory has a peculiarity and it is that not all the chips in the stack are the same, since as the silicon path goes through the different layers it becomes narrower compared to the previous one. This leads us to the fact that depending on the size of the base chip, the scaling will be different and the stack will have a smaller chip each time.
The graph above these lines indicates the size, in nanometers, of the transistors of a 2D NAND memory to have a density equivalent to a 50 nm 3D NAND memory of a certain number of layers. The number goes down due to the fact that the vertical axis marks the size of the NAND transistors in two dimensions, while the horizontal axis marks the number of layers of an equivalent 3D NAND memory in terms of storage capacity.
As can be seen, 2D NAND memory using 10 nm transistors would be necessary to have a storage similar to a 128 layer 3D NAND using 50 nm transistors.