IBM researchers said they have made a breakthrough in chip development that could lead to processors that are smaller but more powerful than current offerings.
Researchers used a technique called molecular self-assembly to create important parts of a semiconductor memory device, IBM reported in a paper presented last week at the IEEE International Electron Devices Meeting in Washington.
According to researchers at IBMs Yorktown Heights, N.Y., research lab, the self-assembly technique takes advantage of a reliable way that certain types of polymer molecules come together and organize themselves. The result of that tendency, the researchers said, are patterns that can be used to create device features that are smaller, denser and more uniform than techniques currently used, such as lithography.
Chip makers will continue to be able to use lithography for many more years to create smaller, faster chips, but those enhancements will also increase the cost and complexity of the technique, an IBM spokesman said.
Molecular self-assembly, an approach based on nanotechnology, will give processor manufacturers another method to shrink the chips while boosting the performance. IBM said the new technique is compatible with existing chip-making tools, enabling manufacturers to implement the molecular self-assembly without greatly increasing costs by having to retool machines and assume risks that come with major changes in processes. The result could be more powerful processors for everything from computers to wireless devices.
IBM researchers expect molecular self-assembly to be used in pilot programs in three to five years.
In creating the crucial parts of the semiconductor memory device using the technique, researchers were able to create a dense silicon nanocrystal array, the basis for a variant of conventional flash memory.
Rob Enderle, principal analyst with The Enderle Group, said IBMs announcement is a step in the right direction for nanotechnology. The key for IBM and other vendors will be developing a process that can produce the billions of chips that are needed every year.
"While some of them have been able to build little itty-bitty test machines ... nobody has come up with a way to address the scalability issue," said Enderle, in San Jose, Calif. "They have to find a way to ramp and produce lots and lots of these things. But this is a good step."