Santa Barbara (CA) - Scientists at UC-Santa Barbara have engineered a new nano-construction process capable of assembling integrated circuits. Five companies helped fund the research, including giants Intel and IBM. A patent was also filed for their discovery. According to the researchers, this technology could be introduced into mainstream semiconductor manufacturing as early as 2011.
BCP
The new process is called Block Co-polymer Lithography. At the most fundamental levels it employs self-assembling chemical patterns to create desired structures - just like proteins in the body. The created features are between 5nm and 20nm, significantly smaller than the features created using today's 45nm lithography, which for gate lengths, for example, are typically around 35nm.
The process was developed by Craig Hawker, a materials professor and director of the Materials Research Laboratory (MRL) at UC-SB, along with two other professors, Glenn Fredrickson and Edward J. Kramer. Hawker describes the process with an analogy to salad dressing. Oil and water do not like to be together. In salad dressing they separate into layers very quickly. Shaking up the bottle allows much smaller droplets of each to be in much closer proximity to the other for short periods of time.
With BCP, a similar two-part process is used. Whereas the two separate processes do not like to co-exist, a method has been created which allows them to exist long enough and closely enough to do their thing which, in this case, is being used to create semiconductor cicuits.
Hawker is basically addressing the future problems of scalability.
History faces the future
A fundamental concept observable over the years is something now called Moore's Law. Moore's Law states that the number of transistors which can be inexpensively placed and mass produced on an integrated circuit doubles every 18-24 months. This voracious reality was discovered by Intel fellow, Gordon Moore, who was instrumental in early semiconductors.
His observed trend has been proven since the original 2,300 transistors in the world's first microprocessor - the Intel 4004 4-bit CPU which was first used in early, simple, handheld calculators. The trend continues today through Intel's immense Itanium 2 and its 800+ million transistors, as well as Intel's Core architecture. Competing companies have followed the trend as well, such as AMD and their AMD64 architecture.
In the past, key new releases followed specific trends relating to the reduction of lithography used in manufacturing. The 4004 was manufactured using a 10,000nm process. To inject some other dates, in 1989 the 80486 was introduced using an 800nm process. It followed the trend. The Pentium 4 was introduced in 2001 with 180nm process. And in 2008 we have 45nm still following the trend, but there is an end in sight. This pace cannot keep up for much longer, and this is where the new technology solutions come into play.
Every leading indicator suggests that by the year 2020 we will have reached the 16nm node, marking the end of the line for conventional lithography. While new products will continue to be developed, and a potential doubling of transistors can continue with manufacturing "tricks," something more is needed.
Square, not hexagonal
One of the problems with previous attempts at self-assembling molecules was that they naturally come together in particular ways. In many cases they would resemble a hexagonal bee-hive array, often assembling at undesirable places making them essentially useless.
In addition, since modern CPU and semicon designers deal with squares, hard edges and straight lines, hexagonal structures do not naturally translate to the toolset base already created - which is huge, literally tens of billions of dollars.
Regarding his new assembly technique Hawker said, "...we've actually shown that by changing the structure of the molecules, and using two self-assembling procedures at the same time, we're actually able to get square arrays for the first time. So now you can start to marry the old technology with the new technology for the fabrication of microprocessors."
Old can meet new
According to Hawker, this self-assembly method could be created in a type of drop-in manner. This means the multi-billion dollar investments in fabrication facilities by companies like Intel and IBM will not have to be scrapped. The same design tools and essential processes will still be in tact. Only the way the circuits themselves are physically formed will be different.
"This [new discovery] allows them to introduce a new technology using current tools in the same fabrication plants. So they don't have to make huge up front investments to bring this to manufacturing." Hawker said, "That's a key feature."
BCP
The new process is called Block Co-polymer Lithography. At the most fundamental levels it employs self-assembling chemical patterns to create desired structures - just like proteins in the body. The created features are between 5nm and 20nm, significantly smaller than the features created using today's 45nm lithography, which for gate lengths, for example, are typically around 35nm.
The process was developed by Craig Hawker, a materials professor and director of the Materials Research Laboratory (MRL) at UC-SB, along with two other professors, Glenn Fredrickson and Edward J. Kramer. Hawker describes the process with an analogy to salad dressing. Oil and water do not like to be together. In salad dressing they separate into layers very quickly. Shaking up the bottle allows much smaller droplets of each to be in much closer proximity to the other for short periods of time.
With BCP, a similar two-part process is used. Whereas the two separate processes do not like to co-exist, a method has been created which allows them to exist long enough and closely enough to do their thing which, in this case, is being used to create semiconductor cicuits.
Hawker is basically addressing the future problems of scalability.
History faces the future
A fundamental concept observable over the years is something now called Moore's Law. Moore's Law states that the number of transistors which can be inexpensively placed and mass produced on an integrated circuit doubles every 18-24 months. This voracious reality was discovered by Intel fellow, Gordon Moore, who was instrumental in early semiconductors.
His observed trend has been proven since the original 2,300 transistors in the world's first microprocessor - the Intel 4004 4-bit CPU which was first used in early, simple, handheld calculators. The trend continues today through Intel's immense Itanium 2 and its 800+ million transistors, as well as Intel's Core architecture. Competing companies have followed the trend as well, such as AMD and their AMD64 architecture.
In the past, key new releases followed specific trends relating to the reduction of lithography used in manufacturing. The 4004 was manufactured using a 10,000nm process. To inject some other dates, in 1989 the 80486 was introduced using an 800nm process. It followed the trend. The Pentium 4 was introduced in 2001 with 180nm process. And in 2008 we have 45nm still following the trend, but there is an end in sight. This pace cannot keep up for much longer, and this is where the new technology solutions come into play.
Every leading indicator suggests that by the year 2020 we will have reached the 16nm node, marking the end of the line for conventional lithography. While new products will continue to be developed, and a potential doubling of transistors can continue with manufacturing "tricks," something more is needed.
Square, not hexagonal
One of the problems with previous attempts at self-assembling molecules was that they naturally come together in particular ways. In many cases they would resemble a hexagonal bee-hive array, often assembling at undesirable places making them essentially useless.
In addition, since modern CPU and semicon designers deal with squares, hard edges and straight lines, hexagonal structures do not naturally translate to the toolset base already created - which is huge, literally tens of billions of dollars.
Regarding his new assembly technique Hawker said, "...we've actually shown that by changing the structure of the molecules, and using two self-assembling procedures at the same time, we're actually able to get square arrays for the first time. So now you can start to marry the old technology with the new technology for the fabrication of microprocessors."
Old can meet new
According to Hawker, this self-assembly method could be created in a type of drop-in manner. This means the multi-billion dollar investments in fabrication facilities by companies like Intel and IBM will not have to be scrapped. The same design tools and essential processes will still be in tact. Only the way the circuits themselves are physically formed will be different.
"This [new discovery] allows them to introduce a new technology using current tools in the same fabrication plants. So they don't have to make huge up front investments to bring this to manufacturing." Hawker said, "That's a key feature."