A few weeks ago, we covered foundry plans for the next several years for TSMC, Samsung, GlobalFoundries, and Intel.
Since then, Samsung has released a major update on its own roadmap plans and projects. The company plans to launch a variety of new process capabilities, including new technology nodes, the introduction of extreme ultraviolet lithography (EUV), and its own fully depleted SOI (FDSOI) technology. To date, GlobalFoundries has been the foundry toying with their FDSOI credentials, so it's interesting to see Samsung break into this space.
Let's start with an overview of Samsung's roadmap before diving into the details of its technology ramps or how we expect these initiatives to play out in various markets. Unfortunately, for reasons known only to him, Samsung distributed some perfectly nice slides, then refused to let us post most of them, despite being allowed to talk about the data they contain and in some cases recreate them on our own. images. Since I have the rough image design aptitude of Koko the gorilla, I apologize for what you are about to endure.
In its presentation, Samsung pointed out that risk production is a goal for a whole year, and that the left and right edges of the roadmap blocks (I can't show you on the slides) don't really mean anything. That's significant, because in some cases, the left edge of a block is read as the beginning of a process, while the right edge means that the product or process is ready for completion. Different foundries mean different things, sometimes, when they use the same phrases. The time span between risk production, production volume, and actual product launch will vary depending on both the product in question and the vagaries of the acceleration process.
When Intel refers to entering volume production on a given node, it generally releases parts within 1-2 months. When TSMC, Samsung, or GlobalFoundries announce a production volume, the chips coming off the line still have to enter various devices, which then go through a manufacturer-specific evaluation and testing process. This is why Samsung could claim that it entered volume production of 10nm SoCs on October 17, 2016, but the first device in those SoCs was the Samsung Galaxy S8, which was launched on April 21. The six-month gap between those two events is typical, and the longest gaps are not unknown.
From now on, I will describe what the company told us in general terms, rather than backing it up with pictures. The aforementioned 14nm process node LPU is Samsung's third generation of 14nm technology, and the company told us that it expects the LPU node to offer improved performance of up to 15 percent compared to 14nm process lines. previous. This type of incremental scaling is a relatively recent innovation in the foundry business. In the past, companies like Samsung, TSMC and GlobalFoundries offered "all" the benefits of a given node up front, with small improvements in the coming years. Now the gap between early and late production at the same node is somewhat larger.
Regarding EUV, we have talked before about the difference between EUV and existing Argon Fluoride (ArF) lithography. Samsung's 8nm node will scale non-EUV production one last time, with strictly incremental performance improvements above 10nm. Beyond that point, Samsung believes that the mask count and costs will simply be too high to justify the non-EUV technology, while the overall performance of any future non-EUV nodes would be worse than the company's planned 8nm. Obviously, if EUV is not launched in time, foundries will try to find ways to further extend existing technology. But we've already been stretching EUV-free lithography for at least a decade, while embracing immersion and multi-pattern lithography to compensate for continued EUV development difficulties.
The goal of offering new FD-SOI progression paths is to provide customers who may not have the funds to invest in state-of-the-art process nodes a way to continue expanding and improving their existing product lines. If you are a current 28nm bulk silicon customer who doesn't want to do a complete chip redesign, 28nm FDS (Fully Depleted SOI) offers performance and performance advantages. FinFET will be the performance option of choice for customers who can justify a complete architecture change, while 18FDS will provide 28nm and 28FDS customers with significant performance improvements.
GlobalFoundries have also talked about their investments in FD-SOI, and it's interesting to see Samsung taking over on this as well. The company hopes to bring eMRAM and RF silicon to market in FD-SOI and cites cost and performance improvements even for RF parts (RF manufacturing is quite different from traditional silicon manufacturing and the profits from newer ones). compute nodes are traditionally much smaller).
Finally, there are Samsung's advances in packaging manufacturing. The company wants to advance the use of new techniques to minimize thickness when fanning out, while eliminating the interposer entirely for HBM and HBM style memory technologies. I-Cube (Interposer Cube) is Samsung's proper name for its 2.5D interposer technology. The goal is to completely eliminate the need for a silicon interleaver and increase bandwidth while reducing costs. We know that AMD is also acquiring HBM2 from Samsung, so they are clearly pushing these technologies into mass production.
Overall, Samsung's roadmap is extremely ambitious and reflects a company that wants to challenge TSMC in multiple product markets. Samsung is looking to position itself to guard against potential unknowns in the market (such as the ramp time for EUV), while also advancing in spaces like FD-SOI, where previously GlobalFoundries appeared to be the only company carrying that particular technology forward.
When Samsung took the lead in 14nm, companies like TSMC were quick to argue that this was a fluke and that they would resume their historic leadership position in new nodes shortly thereafter. It is now clear that Samsung does not intend to allow its 14nm success to be a one-off proposition. It remains to be seen how your competitors in the foundry will respond to this challenge. But the semiconductor industry has been the best for its recent challenges to TSMC, and that's a trend we hope will continue.
Date update on 2021-05-31. Date published on 2021-05-31. Category: Computer class Author: Oscar olg Fuente: hielosmendez