Three-dimensional technologies will smooth the way for the next frontier in flash memory as production of 3D NAND accelerates much more quickly than initially anticipated, according to a Flash Dynamics brief from HIS.
Global share from 3D NAND of overall flash memory shipments is projected to increase marginally to 0,4% by year-end, but the years ahead will propel the market in leaps and bounds.
For instance, a solid jump to 5,2% is projected in 2014, and then 3-D NAND makes a spectacular surge in production to attain 30,2% of total flash memory shipments in 2015.
From there, 3D NAND enlarges its market to 49,8% – virtually half of the entire flash memory market. And by 2017, NAND produced via 3D methods will account for the majority of total flash memory shipments, equivalent to 65,2%.
While the NAND industry is currently in the process of moving to the next lithography, there is widespread agreement that just one or two generations may be left before planar NAND hits its theoretical limit. As lithographies go down further, performance and reliability may become too degraded for NAND to be used in any but the very lowest-cost consumer segment.
As a result, the most promising next-technology replacement will move focus away from miniaturization, which has traditionally concentrated on shrinking NAND cells in the xy dimensions.
Instead, the shift will be toward increasing density via layering. This will be the most cost-effective way of pushing NAND to the next level because most of the same equipment can be used, minimising expenses while maximising returns on investment.
Already Samsung and SK Hynix, the biggest players in the global memory trade, have announced their initiatives in 3D NAND, made public earlier in August during the Flash Memory Summit in Santa Clara.
Samsung says commercial production of V-NAND, the company’s name for its 3D flash memory products, started during the second quarter.
The end product will be a V-NAND-based solid-state drive, in 480Gb and 960Gb densities, initially targeting the enterprise market. V-NAND, Samsung says, will be more reliable than 1x-nanometer NAND, consume less power and also deliver higher performance in sequential as well as random writes.
But the cost differential between a V-NAND solid-state drive and one powered by traditional flash memory will be quite large, IHS expects, which would explain why Samsung is aiming the product first at the enterprise market.
Samsung archrival SK Hynix is also planning to manufacture 3D NAND; its initial 3D product will be very similar to Samsung’s V-NAND, to be available in 128Gb capacity. But this is part of a two-pronged approach, as the company is also sampling a 16-nanometer product.
While both Samsung and SK Hynix have previously mentioned internal development of 3D NAND, the timeline for production has moved faster than expected, and the quickened pace is much more accelerated than many in the industry had anticipated.
Other memory manufacturers, however, have decided to continue with planar NAND for at least one more generation, pushing any 3D plans to a later date. In this group are makers like SanDisk of California, Idaho-based Micron Technology and Toshiba of Japan.
All told, initial production of 3D NAND will be limited, and failure analysis will be difficult because of the multilevel structure of the device. Still, an initial ramp-up of higher-performing products into the enterprise segment will enable suppliers to generate margins and allow processes to mature, IHS believes, even though it may be some time before 3D contributes meaningfully to overall industry bit growth.
At any rate, the 3D race for NAND has already begun, solidifying the timeline for the new technology. NAND suppliers that have yet to address the change are likely to feel pressure to innovate as a result.