3D NAND: Who Will Make It and When?

(The following is an update of a post that originally ran on 13 December 2013. It was republished in 2024 as a part of a series on The Memory Guy blog to honor the 3D NAND inventors who have received the 2024 FMS Lifetime Achievement Award.)

This series, originally run in 2013, has looked at 3D NAND technology in a good deal of technical depth. At that time, the last question to be answered centered around the players and the timing that the technology would make its way to the market. By that time, a lot had been said about the technology and its necessity, but it was unclear that all of the players would be making 3D NAND, and when the big transition would occur.

Now, at the end of 2024, it’s uncommon even to encounter planar NAND since 3D NAND has come to dominate the NAND business. This made me wrestle with what to do with this closing post of the series. Eventually, I decided to keep it close to its original form, since it provides a nice picture of the NAND business in transition.

At the date of the post’s original publication (13 December 2013) it provided each company’s then-current status, to the best of The Memory Guy’s understanding. Readers may also want to refer back to the earlier posts in this series, as well as to a June 2013 Nikkei TechON article that gives a good review of the 3D NAND alternatives that were presented at various technical conferences before the publication of the article.

Let’s start with Samsung, the largest NAND flash producer. Just prior to the Denali/Cadence Memcon conference in August 2013 Samsung announced mass production of its 3D “V-NAND” product (for “Vertical NAND”) and at the 2013 Flash Memory Summit that same month, the company introduced enterprise SSDs based on the technology.

In comparison, at an investor conference only a few months earlier, in May 2013, SanDisk, now a part of Western Digital, plotted the path for its next two generations of planar NAND (dubbed 1Ynm and 1Znm) and showed charts to explain that the company did not expect for 3D NAND to be cost-effective until those two generations had run their course. SanDisk presented a timeline with a pilot run of 3D NAND in late 2015 becoming production in 2016.

Clearly SanDisk and Samsung had very different 3D NAND introduction schedules. So did the rest of the NAND flash cosmos.

(Please note that Toshiba, now Kioxia, and SanDisk, now WDC, introduce the same products at the same time by nature of their joint venture. The same could be said at that time of Intel, now SK hynix’ Solidigm, and Micron, a JV that was later dissolved. For brevity’s sake, this post will only name SanDisk and Micron when speaking of these JVs.)

So, Samsung said that it would begin 3D NAND production in 2013, and SanDisk in 2016. What of the two others, SK hynix and Micron?

At the August 2013 Flash Memory Summit SK hynix announced that it would sample 3D NAND by the end of the year. (This post’s graphic shows side-by-side cross section illustrations of the SK hynix 3D NAND design and an electron micrograph of that device.) At about the same time, in a CNET interview, Micron’s then-CEO Mark Durcan explained that the company planned to sample 3D NAND in the first quarter of 2014, but indicated that the company’s 3D timing would be determined by strategy, and not technology.

A month earlier, in July, Micron announced that it was sampling 16nm planar NAND. SK hynix also announced full production of a 16nm planar part in November 2013.

It is generally understood that planar NAND can’t be produced below about 20nm without using a High-k gate dielectric. High-k materials are pretty tricky to put into production: Micron introduced High-k at the 20nm process node, but delayed ramping production to this node for roughly a year, presumably because of difficulty getting the process to beat the cost of the company’s 25nm process.

In this light, the SK hynix 16nm announcement was a little surprising. SK hynix had not by that time mentioned any use of High-k dielectrics, but had said that it was using an “air gap,” another technology that was new at that time and that could therefore be challenging to bring to mass production.

Samsung’s strategy appeared to be to skip the headache of ramping a High-k planar NAND and to confront the inevitable headache of 3D early on. Even so, since Samsung’s TCAT technology uses a High-k gate dielectric, the High-k headache was still something that Samsung needed to resolve.

SK hynix had then decided to introduce a very aggressive planar technology and to work on an early introduction of 3D at the same time. Although that seemed to be the most difficult path, SK hynix is pretty advanced when it comes to technology, even though few in the industry give the company credit for that distinction. The company fared no worse than its competitors in the long run by choosing this harder path.

Micron tends to be quieter than other firms about new technologies, seldom disclosing much of anything prior to sampling. The company shared very little about its 3D technology at technology conferences, but did show a cross section of a Micron 3D NAND chip during a 2011 Flash Memory Summit keynote, and revealed work on a 256Gb 3D NAND at the 2013 Flash Memory Summit. (The graphic for the first post in this series is based on the electron micrograph of Micron’s 3D NAND in the 2011 keynote.) Micron was openly telling everyone that it was developing the technology, but neither Micron nor its partner Intel, disclosed much about it until they were ready to ship.

This was consistent with the Micron/Intel approach to new technology introductions. While Samsung often boasts that a technology is available well in advance of its actual appearance, Micron, Intel, and their IMFT joint venture rarely discussed a new technology until it is already sampling.

When the chip was finally introduced, we learned that these two companies used a floating gate approach, rather than the charge trap that others employed. SK hynix had demonstrated a floating gate 3D NAND structure at an industry conference, but chose to use a charge trap when it finally went into production.

It’s very difficult to get a charge trap to work the first time, and this was a challenge for all the 3D NAND makers who took this path (Toshiba, SanDisk, Samsung, and eventually SK hynix), but a floating gate, while much better understood, was very difficult to make vertically, and this presented a different challenge to Micron and Intel. In the end, Micron abandoned the floating gate, and SK hynix appears ready to do the same at the the 3D NAND fab in Dalian China that was a part of its acquisition of Intel’s NAND flash & SSD businesses. As of the end of 2024, this was the last facility to still produce floating gate 3D NAND.

At the original 2013 publication of this post, it appeared that Samsung and Hynix would ship their brand of 3D product to some extent in 2013, Micron in 2014, and SanDisk in 2016, but as of mid-December 2013 there was not yet any evidence of customer sampling. In retrospect, the industry struggled to enter mass production for quite some time, and Samsung, determined to be the first, finally started a serious migration from planar to 3D NAND in 2016.

In 2013 there were other reasons to doubt that 3D NAND was ready for production. A late-2013 IEEE article posed several discouraging points, leading to the conclusion that 3D NAND in its current form might never become cost competitive against more lithographically-intensive planar designs.

In the end, the industry’s overall 3D NAND status looked pretty discouraging in late 2013. Some players had announced mass production but were not yet sampling, others who had set a future schedule to sample but nothing more, and still others who said that they would not produce until 3D NAND costs could be brought below those of planar technologies, all while an IEEE article pointed to a future that might not even include any of the then-current versions of 3D NAND!

The Memory Guy ended the series by asking: “Will 3D NAND actually materialize? It’s hard to tell.” At the end of 2013 that was certainly the industry’s status.

It’s pretty amazing how much progress the industry has made since that time!

Epilogue

If you’re like me, you have read in this series about macaroni channels and cheese charge traps, as well as layer cakes, and you may be starting to feel hungry for junk food. Maybe we should take a break and go get something to eat! It’s been a long and technical series of posts. Thanks for bearing with The Memory Guy as I explored the technology.

I would especially like to thank my reviewers, twelve wonderful people who generously gave me their time and suggestions to help make this series as clear and accurate as possible, and I must also thank the technical geniuses who helped me to understand 3D NAND technology by explaining arcane concepts at a level low enough for me to grasp.

This post is the last of a series called What is 3D NAND? Why do we need it? How do they make it? that was originally published in weekly segments during the fourth quarter of 2013 on The Memory Guy blog and was updated in 2024. The different sections are listed below, with a hot link to each section.

Click on any of the above links to learn more about 3D NAND technology.

3D NAND: Who Will Make It and When?

Epilogue

4 thoughts on “3D NAND: Who Will Make It and When?”

Leave a Reply Cancel reply