Source: Content compiled from "
IEEE
", thank you.
Back in 2014, as the industry grappled with the imminent end of Moore’s Law, IBM embarked on an ambitious, $3 billion project called “7nm and Beyond.” The bold goal of the five-year research project was to examine how computing technology would continue to evolve in the future as the physics of reducing chip size conspired to combat it.
Six years later, Moore’s Law is no longer the same law it once was. Gordon Moore’s observation (and later the industry’s) that the number of transistors on a chip would double roughly every two years now seems a thing of the past. However, we still need innovation in computing, and the “7nm and Beyond” project has helped meet that ongoing need.
“The work of finding new device architectures to enable device scaling, and new materials to enable performance differentiation will never end,” said Huiming Bu, director of IBM’s Advanced Logic and Memory Technology Research, Semiconductor and AI Hardware Division.
While the chip industry may not be as constrained by Moore's Law as it was in the past, the "7nm and Beyond" project has already led to important innovations, even as some chipmakers seem to be expressing frustration over much of the past year.
One example of this frustration was GlobalFoundries’ decision two years ago to halt development of its 7-nanometer chips.
Back in 2015, a year into the "7nm and Beyond" project, IBM announced its first 7nm test chip, with extreme ultraviolet lithography (EUV), provided by ASML, as a key technology. Despite the growing pains of using EUV, with the result that the richest chipmakers were the only ones to continue with its scaling, it has since become a key enabling technology not only for the 7nm node, but also for nodes 5nm and beyond, Huiming Bu said.
“In the 2014-2015 time window, there were big doubts in the industry about the actual feasibility of EUV technology,” Bu said. “That’s not a problem now. EUV is now a mainstream enabler. At that time, we delivered the first 7nm technology based on EUV, which helped build confidence and momentum for EUV manufacturing in our industry.”
Of course, EUV has enabled the 7nm node, but IBM is aiming to go beyond that. IBM believes that the fundamental element of chips that will enable scaling beyond FinFETs will be nanosheet transistors, which some even believe could be the final step in Moore's Law.
Nanosheets appear to be an alternative to the FinFET architecture and are expected to enable the transition from the 7nm and 5nm nodes to the 3nm node. In the architecture of nanosheet field-effect transistors, current flows through multiple silicon stacks that are completely surrounded by the transistor gate. This design greatly reduces the amount of current that can leak in the off state, allowing more current to be used to drive the device when the switch is open.
“In 2017, the industry was still questioning what the new device structure after FinFET would be,” Bu said. “Three years later, the industry is chasing nanosheets and thinking they will be the next device structure after FinFET.”
There have been some key advances in transistors and switches, but the “7nm and Beyond” project is also giving some important insights into how the wiring on top of all those transistors and switches will evolve into the future.
“One of our innovations was to make the copper lines as large as possible,” said IBM Fellow Daniel Edelstein. “As always, the hardest part is just patterning these extremely tiny and tall trenches and filling them with copper without defects,” Edelstein further noted.
Despite the challenges of using copper, Edelstein doesn’t see the industry moving away from copper to more exotic materials in the near future. “For the products that are being manufactured today, copper is certainly not going to be the end of the world,” Edelstein said.
He added: “Several companies have indicated that they intend to continue using it. So I can’t tell you exactly when it will break. But we have seen so-called resistance crossovers that have been pushed out into the future.”
While chip size, architecture, and materials drive many of the innovations in the "7nm and Beyond" project, both Edelstein and Bu noted that artificial intelligence (AI) also plays a key role in how they're heading into the future of computing.
“With the advent of AI, brain-inspired computing, and other types of non-digital computing, we started developing other devices at the research level, especially emerging memory devices,” Edelstein said.
Edelstein is referring to emerging memory devices such as phase-change memories (or “memristors,” as some others refer to them), which are thought of as analog computing devices.
The emergence of these new memory devices has provided a renaissance in thinking about potential applications beyond traditional data storage. Researchers are envisioning new roles for the 30-year-old magnetoresistive random access memory (MRAM), which IBM has been working on since MRAM debuted.
“MRAM has finally had enough breakthroughs that it can not only be manufactured, but it also meets the kinds of requirements needed to be competitive with SRAM for system cache, which is ultimately the holy grail,” Edelstein said.
Last year, chip equipment maker Applied Materials Inc. made available to customers the tools to make that change, providing evidence of its potential to embed MRAM and other nonvolatile memories, including RRAM and phase-change memory, directly into processors.
Bu said IBM will continue to pursue new devices, new materials and new computing architectures to achieve better power performance. He also believes that the need to integrate various components into an overall computing system is beginning to drive a whole new world of heterogeneous integration.
Bu added: "Building these heterogeneous architecture systems will be key to future computing. This is a new innovation strategy driven by the needs of AI."
★
Click [Read original text] at the end of the article to view
the original link of this article
.
*Disclaimer: This article is originally written by the author. The content of the article is the author's personal opinion. Semiconductor Industry Observer reprints it only to convey a different point of view. It does not mean that Semiconductor Industry Observer agrees or supports this point of view. If you have any objections, please contact Semiconductor Industry Observer.
Today is the 2371st issue of content shared by "Semiconductor Industry Observer" for you, welcome to follow.
Semiconductor Industry Observation
"
The first vertical media in semiconductor industry
"
Real-time professional original depth
Scan the QR code
, reply to the keywords below, and read more
NVIDIA｜SMIC｜CPU
｜Wafer
｜
FPGA
｜
5G｜Google｜RF
Reply
Submit your article
and read "How to become a member of "Semiconductor Industry Observer""
Reply
Search
and you can easily find other articles that interest you!
Click to read the original text
to view the original link of this article!
京公网安备 11010802033920号