Intel opens D1X-Mod3 Fab extension; Intel moves production of 18A to H2’2024

Intel has been undergoing a major period of product expansion for the company over the past few years. Recent announcements of new facilities in Ohio and Germany have attracted much attention – especially given their emphasis on Intel’s foundry services projects – and Intel has been working for a long time to expand existing facilities for their own use. The development of the company’s next generation EUV and gate-all-round-style transistors (ribbonfed) requires not only developing and refining technology, but also more space. That’s a lot.

To that end, Intel is hosting a grand opening ceremony today for the Mod3 expansion of the D1X Mod3 in Oregon. The first expansion announced in 2019 was the third model (module) and second expansion for Intel’s flagship Dave Fab, built after the initial build of the D1X in 2010. And in keeping with the tradition of Intel Fab releases and enhancements, the company created an event for anything, including bringing out the governor of Oregon to show off their $ 3 billion investment.

But the hype aside, the latest mod for Fab is very important to Intel: it not only adds another 270,000 square feet of clean room space to the facility – expanding the D1X by about 20% – but this is the only Fab module. The high number hole (high NA) that Intel starts with in its 18A process is large enough to support the EUV tool. ASML’s upcoming TWINSCAN EXE: 5200 EUV tool is designed to be even more powerful, but it’s slightly larger than the NXE 3000 series EUV tools used by Intel for their first generation EUV processes (Intel 4 / Intel 3). It is so large that the ceiling of the D1X is too low to fit the engine, regardless of its weight-bearing floor.

Size Comparison: ASML Normal & High NA EUV Machines

As a result, Mod3 is built without a small area to fit this massive machine. Intel did not expect to deliver the machine for another two years, but they had to start production many years ago to get to this point.

Meanwhile, although it has been announced that the D1X-Mod3 will only officially open today, Intel has been moving important tools to the Mod3 since last August. As a result, today’s opening is a formal launch for Modi because some parts of it are already set up (if not already in use). However, even with that launch, according to Intel, the company expects the devices to move for another year, especially as they bring in the remaining, lower priority tools.

Incidentally, our own doctor. Ian Cutress, while touring the facility late last year, had the opportunity to see the D1X in all its glory. At the time, Intel was already in the final stages of completing the Mod3 expansion, and introduced EUV engines as part of the development of Intel’s first EUV nodes, the Intel 4 and Intel 3 process nodes. So for more information on D1X and what’s going on there, see that article.

Sequence of EUV engines in D1X

Finally, with the Mod3 expansion open, Intel is taking advantage of the opportunity to rename the 450-acre campus where D1X sits today. Intel’s Ronler Acres campus has been the focus of Intel’s Fab R&D initiatives for decades, and includes D1X as well as Intel’s older D1 development fabs such as D1B, D1C and D1D. Therefore, in response to all the important R&D happening on the site, Intel is renaming co-founder Gordon Moore, one of the key players behind the development of Intel’s early technologies. The newly renamed campus will now go Gordon Moore Park on Ronler AcresOr Gordon Moore Park In short. Although there have been many (many) things named after Moore over the years, from laws and buildings to awards and medals, this is the biggest thing Moore has ever named (yet), and this is the first time the entire campus has been named. After the light.

Intel Road Map Update: Intel 18A H2 moved to 2024

Intel also used their latest press release to brief reporters on the D1X-Mod3 launch and to keep everyone informed of the latest updates on Intel’s Development Roadmap. Strictly speaking, nothing new here – all of which was first announced at Intel’s 2022 investor meeting in February. However, this is the first time Intel has involved more technical reporters than investors in the current state of its development efforts.

The big news here is that Intel is systematically moving the start date for production to the Intel 18A tip. Intel’s second generation “angstrom” node was first expected in 2025; But now the company is raising it in half a year i.e. in the second half of 2024.

As a result, Intel’s road map now looks like this:

As the company prepares for its first EUV process, the Intel 4, later this year, Intel’s roadmap begins to be very concise, starting in the second half of 2023. Intel 3 will go into production in the second half of that year. Improved EUV process. Meanwhile, after 6 months, the Intel 20A goes into production. 20A is Intel’s first “angstrom” terminal, which includes their gate-all-round-style “ribbonfed” finfets and PowerVias.

But, if all goes according to plan, 20A will be a relatively short term terminal due to the movement of 18A. Intel’s second generation angstrom node, which includes an updated ribbon design and other improvements to Intel’s GAA manufacturing technology. As the 18A is the farthest point in Intel’s production roadmap, the company is relatively reluctant to bring anything new to the 18A, but Intel plans to re – establish unquestioned leadership in the chip – making industry.

According to Intel, the 18A development is moving well, and the company’s R&D operations are currently at or near all their development milestones, giving the company confidence that it will be able to produce products based on the process node by 2024 instead of 2024. As originally planned.

One consequence of bringing in the 18A, however, is that Intel is now definitely going into early production of the 18A without all their high-end NA engines. The process of launching high-end NA engines is at 18A, but the TWINSCAN EXE 5200 is not expected to last until 2025, meaning that Intel will now have to use the existing 3000 series engines to kickstart 18A production. Until this latest development, Intel visualized that it was built on high NA engines and 18A waist, so whether it was always true or not is not clear now.

It remains to be seen, in turn, what 18A means for production. The biggest advantage of high NA engines was the performance, as Intel could use their native (non-HA) engines up to 18A, allowing Intel to process scales with smaller (or no) multi-format due to the high accuracy of high NA. Apparently, they will be able to produce Intel 18A in 2024, perhaps even in decent sizes, but they will not be able to go to large-scale production on the Intel scale until the first high-end NA engine is available. In 2025.

Also, as always, it should be noted that Intel’s production roadmap dates are not the start-up dates for a new process node production, but the technology-based hardware shelves. Although the 18A will start as planned on the H2’24 now, it may take a few months in 2025 for the first products to reach customers’ hands, especially if Intel launches that window later. All of this, given the large size of these publishing windows and Intel’s own history, is a chance bet as Intel rarely introduces new products / technologies at the beginning of a launch window.

Finally, in Intel’s development briefing, it was confirmed that Intel is using a completely internal “test risk reduction” node as part of their development process for PowerVia technology. The purpose of the test node is to cut off the full risk of 20A by allowing Intel to create and test PowerVias separately from ribbons. In this case, the test node uses Intel’s well – established FinFET technology on the front – end, while the backend uses the trial version of PowerVia. No such tip has been announced for ribbonfed, but even if one is not, debugging the first-generation PowerVia 20A with ribbonfettes is still a simplification of the process, as it allows Intel to run both components semi-independently. , And learn from both in the process.

This is a significant change from how Intel has developed key new production nodes in the past, and they agree at first. Intel’s 10nm issues were largely caused by the simultaneous integration of several technological changes, and there is a very drastic reduction in feature sizes. Dividing these things into smaller, more frequent production node updates is one way Intel can reduce this risk in the future. Now with an internal testing node for PowerVia development, they aim to do even more risk reduction by releasing both RibbonFETs and PowerVia together in the first half of 2024 as part of the Intel 20A.

Leave a Comment