IBM Towards 1nm chips

IBM researchers presented a series of innovations at this year's IEDM conference [1] that demonstrate a future beyond nanosheet devices and copper interconnects, laying the groundwork for semiconductors with nodes above 1 nm in the near future.

01IBM announces two major breakthroughs

For many years, the pace of semiconductor innovation has been driven by Moore's Law, which states that approximately every two years, the number of transistors on a microchip will double. In recent years, this prediction has slowed as we have hit the physical limits of the materials used to develop chips. 2021 saw the release of the world's first chip with a 2 nm node at IBM Research [2], the world's smallest chip. Prior to this, there has been a flurry of innovations in recent years that have continued to shrink the node of the chip. While the industry is still a few years away from reaping the greatest benefits from the 2nm breakthrough, IBM Research has been focusing on something beyond the horizon.

Now, IBM has identified two major breakthroughs: those that will lead the way in designing computer chip nodes, targeting 1 nanometer and beyond. Both have been presented at this year's IEEE International Electron Devices Meeting (IEDM) in San Francisco.

02"Interconnect 3.0": A new era beyond aluminum and copper

In computer chips, the wiring between components in a semiconductor is called an interconnect. It is the way in which current flows between individual transistors, between memory, processing units and any other component in the chip: the more efficient the interconnect is in allowing this transfer, the more efficient the chip will be. For decades, the most advanced interconnects between chips were made of aluminum.

All the way back in 1997, IBM announced [3] that by using copper instead of aluminum as interconnects, it could make microchips smaller and faster. Copper wires conduct electricity with 40% less resistance than aluminum wires, which means about 15% faster processing speed. Over the past few decades, this seismic shift has led to copper becoming the industry standard for interconnects.

But, as was the case with silicon, IBM is gradually reaching the physical limits of what copper wire can do. On the road to 1 nanometer and beyond, IBM believes the effectiveness of microcopper wires is beginning to wane. As a result, IBM researchers have been looking for something after copper, and the answer may be found in the metal ruthenium.

Copper interconnects have always required a barrier liner material to form a proper wiring structure. As devices shrink, the space and liner material available for copper wiring becomes smaller and smaller. With ruthenium, it is possible to scale to nodes beyond 1 nm and still be an efficient conductor without the need for a liner, which helps save space. Ruthenium formed by subtractive plate-making methods can also potentially be used in a new type of interconnect integration scheme called top-through-hole integration. In this scheme, the interconnect vias are formed on top of the conductor rather than underneath it, allowing the formation of continuous conductors and self-aligning vias for the most critical interconnect layers. In addition, the embedded air gap formed by this top through-hole integration robustly leads to a reduction in interconnect parasitic capacitance, which will also help increase the speed and reduce power consumption of the chip.

The team at IBM has been studying the potential of ruthenium for more than three years, and it is strongly believed that this precious metal is a serious contender to replace copper. The researchers used dual patterning with extreme ultraviolet lithography (EUV) to create test structures on their existing machine in Albany. This makes the current generation of EUV machines available today a breakthrough that will extend to the next generation of high-NA EUV machines. It is called "Interconnect 3.0" to reflect the new era beyond aluminum and copper.

Over the next few years, the researchers plan to refine their testing to the point where they can produce fully viable chips. But they believe the path to "Interconnect 3.0" is clear with the use of ruthenium at the 1nm node and beyond.

03Release of VTFET, the chip of choice for the post-nanometer era

At last year's IEDM conference, IBM announced VTFET [4], a new approach to designing semiconductors. With VTFET, transistor elements are stacked vertically, rather than horizontally, as has been the standard for designing chips since the computer era. This dramatically increases the number of transistors that can fit on a single chip, just as cities made up of skyscrapers are far more densely populated than row houses in the suburbs.

At this year's IEDM conference, the group announced that they have achieved 90 percent of the device performance of their technology goals on the best chips demonstrated in actual silicon hardware. The group's research shows that it is possible to scale VTFET designs well beyond the performance of state-of-the-art 2nm node nanosheet designs.

While nanosheet technology for 2nm chips is still many years away: most companies have not even released commercially viable 2nm chips yet, IBM Research is always focused on what's next.

VTFET designs represent a giant leap toward building the next generation of transistors that will enable the trend toward smaller, more powerful and more energy-efficient devices in the years to come. Now, it is clear from the latest silicon hardware results that VTFETs have the performance capabilities to support these claims. In 2017, IBM Research stated that nanosheet device architectures will be the next device architecture needed beyond FinFET to mass produce smaller, more efficient devices now that the industry has adopted the 3nm and 2nm nodes.

IBM believes that VTFETs are a viable option for future generations of innovative chip designs in the post-nanowafer era.

"Combining the space and efficiency gains of VTFETs with the potential of subtractive ruthenium lines, interconnected by EUV double patterning, there is a long road to smaller, more efficient devices at the 1nm and beyond nodes."

Reference links：

[1]https://research.ibm.com/blog/1nm-chips-vtfet-ruthenium

[2]https://research.ibm.com/blog/2-nm-chip

[3]https://www.ibm.com/ibm/history/ibm100/us/en/icons/copperchip/[4]https://research.ibm.com/blog/vtfet-semiconductor-architecture