Reliability of NAND Flash Arrays: A Review of What the 2-D–to–3-D Transition Meant

Abstract

This paper reviews what changed in the reliability of NAND Flash memory arrays after the paradigm shift in technology evolution determined by the transition from 2-D to 3-D integration schemes. Starting from a quick glance at the fundamentals of raw array reliability, the reasons for its worsening with the evolution of 2-D technologies will be discussed, focusing on the physical phenomena which contributed more to that outcome. By exploring the dependence of the magnitude of these phenomena on cell and array parameters, the abrupt improvements achieved from the 3-D transition in terms of raw array reliability will then be explained, highlighting also that these improvements were turned into new opportunities for the technology. Finally, the physical issues specific to 3-D arrays will be addressed, providing a glimpse of the challenges that the NAND Flash technology will have to face from the standpoint of array reliability in the near future. Index Terms-3-D NAND Flash arrays, Flash memories, semiconductor device modeling, semiconductor device reliability. I. INTRODUCTIONT HE NAND Flash technology has become, today, the undisputed leader in the nonvolatile memory market, largely overcoming the hard-disk drive (HDD) technology in terms of revenues [1]. This outcome has been determined by the capability of the NAND Flash solution to address quite a variety of applications better than any other storage technology, thanks to successful tradeoffs among cost, performance, and reliability. At the heart of all that there is, of course, the possibility to steadily increase the integration density of the NAND array and, in turn, the memory capacity per chip. This is clearly proved in Fig. 1 in terms of gross bit storage density (GBSD), i.e., the ratio between the storage capacity and the total chip area, of the NAND Flash chips presented at the IEEE International Solid-State Circuits Conference (IEEE ISSCC) since 2001 (see [2] for further details on the analysis methodology).Up to ∼2015, the GBSD increase was achieved, first of all, through a constant pace miniaturization of the memory cells

On DRAM Rowhammer and the Physics of Insecurity

“…For advanced DRAM processes where the SN capacitance has shrunk below 10 fF [24], this would require less than 5000 electrons. In the latter case, each electron would result in an SN potential shift of about 20 μV making it one of the most sensitive electrometers in existence except for the case of a NAND Flash cell transistor [57]. Second, the retrograde p-well presents a reflecting electric field to any stray electrons within the p-well.…”

Error Suppression of Last-Programmed Word-Line for Real Usage of 3D-NAND Flash Memory

“…Therefore, it is more serious problem that not VTH-up error by capacitive coupling but VTHdown error by lateral charge migration as data-retentions time increases. The impact of capacitive coupling is smaller than that of lateral charge migration because this paper researches 3D charge-trap NAND flash memory [12][13]. Fig.…”

Random Telegraph Noise in 3D NAND Flash Memories

“…The rest of the cells are contacted by planar wordlines that span over an entire block of the array. One of the advantages of this structure is that the large increase in density allowed by the exploitation of the third dimension makes it possible to relieve some of the pressure on channel length scaling and its many drawbacks from the viewpoint of process complexity and reliability, well known in planar devices [ 52 , 53 ]: cell length in 3D NAND is around 25–30 nm [ 54 ], with the additional advantage of becoming less dependent on the availability of advanced lithography tools. A second advantage of this solution lies in its manufacturing process: memory cells are not patterned individually, but they are formed all at once as cylindrical holes are cut through the stacked wordlines, creating the strings.…”