News & Analysis
Comment
Barry.Pangrle_#1
From the source below, 2017-2019 for 7 nm seems pretty close to my previous ...
wilber_xbox
I will go with the extension of the CMOS such as nanowire rather than completely ...
CICC talk: Graphene could succeed CMOS
Rick Merritt
9/16/2011 11:49 AM EDT
SAN JOSE, Calif. – CMOS semiconductor technology could run out of gas at about 7 nm in 2024, and graphene is the leading candidate to replace it, according to a keynote address to be delivered at next week's Custom Integrated Circuits Conference (CICC) here.
"Graphene has shown a lot of promise to eventually replace silicon microchips, but in my opinion we won't see it in use until after silicon reaches its limits early in the next decade," said James D. Meindl, professor of electrical and computer engineering at Georgia Institute of Technology and founding director of its Nanotechnology Research Center that has been studying graphene for five years.
By 2024, silicon MOSFETs will hit walls in how short engineers can make a channel and gate and how thin a gate insulator they can support, Meindl said, citing projections of the International Technology Road Map for Semiconductors.
Graphene faces plenty of challenges before it can become a successor to CMOS. "We have to make multiple billons of transistors in a sheet of graphene, but we've made less than a handful of transistors so far," said Meindl.
The new material was discovered by researchers at the University of Manchester in 2004 in work that won a Nobel Prize for finding a way to create a single layer of carbon atoms. "No one thought you could do that, but now that is just the starting bell for what can be done with a single layer of carbon atoms in perfectly ordered hexagonal cells," he said.
To date, researchers have uncovered at least two techniques for making graphene. They have also created some "rather crude" working transistors in the material.
The graphene transistors have better electrical and thermal conductivity and current carrying capabilities than copper interconnects. They also are very attractive as a way to make MEMs, Meindl said.
"The most impressive graphene transistors described to date have been RF transistors," such as an amplifier for a 500 GHz analog signal, said Meindl. "Graphene switches are more challenging to make for many reasons including their leakage current," he added.
Meidl's lab is working on ways to make 15nm wide ribbons of grapheme that could be building blocks for graphene switches that are as fast and power efficient as silicon. The chief challenge is making the ribbons without damage at the edges that degrades the positive characteristics of the material.
So far nearly 700 researchers from nearly every department in the Georgia Tech college of engineering have visited Meindl's lab to explore graphene. "The interest in our technology is as broad as engineering and virtually as broad as physical science," he said.
The CICC program includes a broad range of papers covering topics ranging from wired, wirelsss and optical communications to clocks. PLLs, ADCs and power components. Special sessions will address 3-D chip stacking and biomedical technologies.
"Graphene has shown a lot of promise to eventually replace silicon microchips, but in my opinion we won't see it in use until after silicon reaches its limits early in the next decade," said James D. Meindl, professor of electrical and computer engineering at Georgia Institute of Technology and founding director of its Nanotechnology Research Center that has been studying graphene for five years.
By 2024, silicon MOSFETs will hit walls in how short engineers can make a channel and gate and how thin a gate insulator they can support, Meindl said, citing projections of the International Technology Road Map for Semiconductors.
Graphene faces plenty of challenges before it can become a successor to CMOS. "We have to make multiple billons of transistors in a sheet of graphene, but we've made less than a handful of transistors so far," said Meindl.The new material was discovered by researchers at the University of Manchester in 2004 in work that won a Nobel Prize for finding a way to create a single layer of carbon atoms. "No one thought you could do that, but now that is just the starting bell for what can be done with a single layer of carbon atoms in perfectly ordered hexagonal cells," he said.
To date, researchers have uncovered at least two techniques for making graphene. They have also created some "rather crude" working transistors in the material.
The graphene transistors have better electrical and thermal conductivity and current carrying capabilities than copper interconnects. They also are very attractive as a way to make MEMs, Meindl said.
"The most impressive graphene transistors described to date have been RF transistors," such as an amplifier for a 500 GHz analog signal, said Meindl. "Graphene switches are more challenging to make for many reasons including their leakage current," he added.
Meidl's lab is working on ways to make 15nm wide ribbons of grapheme that could be building blocks for graphene switches that are as fast and power efficient as silicon. The chief challenge is making the ribbons without damage at the edges that degrades the positive characteristics of the material.
So far nearly 700 researchers from nearly every department in the Georgia Tech college of engineering have visited Meindl's lab to explore graphene. "The interest in our technology is as broad as engineering and virtually as broad as physical science," he said.
The CICC program includes a broad range of papers covering topics ranging from wired, wirelsss and optical communications to clocks. PLLs, ADCs and power components. Special sessions will address 3-D chip stacking and biomedical technologies.
|
|
|
|
|
Navigate to related information
p_g
9/16/2011 11:18 PM EDT
Interesting that we will be at 7nm by 2024. What fascinates me most is that we are following Moore law till the physical limitation of semiconductor.
Carbon nanotube/graphene switches, don't know which one will be the next technology but curiously watching it.
Sign in to Reply
GREAT-Terry
9/17/2011 3:57 AM EDT
2024 is not very far away from now! If graphene is going to be the next technology in electronics, we need to start more education now. At least engineering students have to learn more about this new material to be well prepare for the next generation electronics!
Sign in to Reply
skal_jp
9/18/2011 7:41 PM EDT
It will be interesting to see the impact on design flows. I think digital won't change that much (will graphene use GDS?), but as for analog flow it might be different. Will good old Spice stay in line?
The period with CMOS and graphene (or any other successor) co-existing might be tricky. How about mixed technology, like it was for bipolar and CMOS?
Sign in to Reply
resistion
9/18/2011 9:28 PM EDT
Graphene doesn't have much of a bandgap. Until it does, it won't replace any semiconductors.
Sign in to Reply
resistion
9/18/2011 9:53 PM EDT
2024 is not 7 nm, it should be ~ 2 nm or several graphene lattice constants. 7 nm is already ~28 graphene lattice constants.
Sign in to Reply
JMUNN
9/19/2011 3:25 PM EDT
It sounds like we are at a realm where microbes and enzymes could do the heavy lifting instead of bulk processes like diffusion and ion implantation.
Sign in to Reply
Barry.Pangrle_#1
9/20/2011 9:43 PM EDT
If Intel is shipping 22 nm in 2012 and on pace to deliver 14 nm in 2014, at that pace it would be 10 nm in 2016 and 7 nm in 2018. Is there some reason to believe that suddenly the pace is going to fall off dramatically between now and 2018?
Sign in to Reply
seaEE
9/20/2011 10:48 PM EDT
At some point we may find out whether Moore's law is process dependent, or if it is an equation that describes the capabilities of human endeavor.
Sign in to Reply
prabhakar_deosthali
9/25/2011 7:56 AM EDT
what is the vision for the associated process and test equipment availability for something with geometry of less than 7nm.
Sign in to Reply
wilber_xbox
9/25/2011 9:11 AM EDT
I will go with the extension of the CMOS such as nanowire rather than completely different carbon based devices due to ease of integration with the current technology unless proven otherwise.
Sign in to Reply
Barry.Pangrle_#1
9/29/2011 7:33 PM EDT
From the source below, 2017-2019 for 7 nm seems pretty close to my previous comment's 2018 timeline. Again, what were the reasons for pushing 7 nm out to 2024?
http://www.electroiq.com/blogs/electroiq_blog/2011/09/the-first-450mm-clubhouse.html
"And note that a 2015 450mm pilot line coincides with what is expected to be the 10nm node at leading edge, and 2017-2019 could be 7nm or even 5nm."
Sign in to Reply