Good news: there are far fewer bigoted extremists out there when it comes to FD-SOI vs. FinFETs. People want the best technology for their application. It’s that simple. That’s a key piece of news from the updated survey by Dan Hutcheson, CEO of VLSI Research, which he presented in the afternoon session of the SOI Consortium’s 2018 SOI Symposium in Silicon Valley
The afternoon then featured presentations by foundry partners, which I’ll cover here.
Also in the afternoon were presentations by wafer-maker Simgui, some innovative start-ups leveraging FD-SOI for custom SoCs and the final panel discussion. I’ll cover those in Part 3 of this series.
BTW, if somehow you missed my coverage of the morning sessions about very cool new products and projects from NXP, Sony, Audi, Airbus and Andes Technology, be sure to click here to read it.
The presentations are starting to be posted on the SOI Consortium Events page – but some won’t be. Either way, I’ll cover them here.
A couple years ago at the annual SOI Symposium in Silicon Valley, Dan Hutcheson presented results of a survey he did (ASN covered it – you can still read about it here). At the 2018 event, he presented an update, which is now posted. You can get it here.
The FD-SOI roadmap and IP availability are no longer issues for decision makers, he found. The 14nm branch – do you go FinFET or FD-SOI? – is gone. “Fins and FD are complementary,” he observed. Most people said they’d consider using both and running two roadmaps, choosing whichever technology is appropriate to a given design.
From a transistor viewpoint, the top reasons to choose FD-SOI is that it’s better for analog and has lower leakage/parastics. It’s perceived as better for complex, high mixed-signal SoCs, and especially for RF and sensor integration. In fact, people see RF as the new mixed-signal, wherein FD-SOI is uniquely positioned for 5G and mmWave.
From a business viewpoint, FD-SOI is perceived to have real advantages. In particular, FD-SOI wins when it comes to keeping down design costs, manufacturing costs and time-to-market. IoT is still the hottest target market for FD-SOI, to which he adds high growth expected in automotive and medical.
With 20 tape-outs in 2018, Samsung is seeing an acceleration in its FD-SOI business. “The trend is healthy,” said Hong Hoa, SVP of the company’s foundry business. FD-SOI, he continued, is on a “differentiation path.”
Samsung’s 28nm FD-SOI process, called 28FDS is at full maturity with very strong yields. They’re seeing more customers and a wider range of applications. The design infrastructure, silicon-verified IP and methodologies are also all mature. They have optimal implementation and verification guidelines for body bias design, a body bias memory usage guide, and a body bias generator integration guide. The process supports Grade 1 automotive, and will be qualified for Grade 2 in a few weeks.
FD-SOI, Hoa reminded the audience, offers superior RF performance compared to both planar bulk and 14nm FinFET. The Samsung strategy is to first provide a base for for the FD-SOI process, then add RF and eMRAM. The base for 28nm was done in 2016; they added RF in 2017 and eMRAM this year.
The Samsung platform for IoT applications integrates both RF and eMRAM to support multi-function needs in a single platform. Lead customers are already working with eMRAM in their designs, he added. (BTW, Samsung has a really nice video explaining their eMRAM offering – you can see it on YouTube here.)
The basic PDK for the Samsung 18nm FD-SOI process (18FDS) will be available in September 2018, with full production slated for fall of 2019. It will deliver a 24% increase in performance, a 38% decrease in power, and a 35% decrease in area for logic. RF for the 18FDSplatform will be ready by the end of this year, and eMRAM beginning in 2019.
With design wins from 36 customers underway, 12 of which are taping out in 22FDX (GF’s 22nm FD-SOI process) this year, the market has validated FDX for differentiation, said GF SVP Dr. Bami Bastani. And indeed, designers are using it for a wide array of applications across North America, Europe, Asia/Pacific and Japan.
Customers in the North America are designing in 22FDX for NB-IoT, industrial, RF/analog, mobile, network switches and cryptocurrency applications. In Europe, it’s more or less the same plus automotive/mmWave, optical transmission, wireless BTS and AI/ML. In Asia Pacific/Japan the mix is similar to Europe.
Bastani sees the three big enablers as the the strengths of the roadmap, the ecosystem and multi-sourcing from Dresden and Chengdu (where they’re already equipping the cleanrooms). He also tipped his hat in acknowledgment to the partnership with FD-SOI wafer supplier Soitec, noting that they have gone the extra mile to match GF’s requirements.
So that was the first part of a great afternoon. As mentioned above, my next post (part 3) will cover a very informative presentation by wafer-maker Simgui on the markets in China, plus talks by some innovative start-ups leveraging FD-SOI for custom SoCs and the final panel discussion.