Intel helps build low-power transistor

As the search for a silicon transistor replacement rolls on, Intel and a European company called Qinetiq have created a transistor that consumes far less power and performs better than the transistors inserted into today's chips.

Following two years of joint research, the companies announced this week that they created a "quantum well" transistor out of indium antimonide developed by Qinetiq. Indium antimonide is a so-called III-V compound, meaning that its two elements come from III and V columns in the periodic table of the elements. Some communication chips already incorporate III-V compounds.

A quantum well transistor differs substantially from existing transistors in that data values are determined by the state of electrons, rather than the electrons' transmission.

The companies said that the experimental transistors could run on one-tenth the power of current transistors and provide the same performance--or provide three times the performance using the same energy consumption.

"The experimental results of our joint research with Qinetiq demonstrate that indium antimonide is a promising material for potential integration in future transistors," Ken David, director of components research for Intel's Technology and Manufacturing Group, said in a prepared statement.

Silicon can only shrink so much
Universities and chip companies are putting increasing energy into finding a replacement for silicon transistors, the tiny on/off switches inside computer chips that create the 1s and 0s of computing. Chipmakers have steadily shrunk the size of silicon transistors over the past three decades, which in turn has paved the way for cheaper and faster cameras, computers and phones.

The end for shrinking silicon transistors, however, is getting closer. Some transistor components are only a few atoms wide and thus can only be shrunk slightly more than they already have been. Smaller transistors also leak electricity, which creates heat and battery problems.

Researchers say hybrid chips that combine silicon and some as-yet alternative material will begin to appear in about 10 years, while chips made almost completely of the new materials may start to show up in the 2020s.

When these alternatives will become real, and what they will look like, is the subject of huge debate. Hewlett-Packard researchers, who are promoting a radical device called a "crossbar" to replace transistors, assert that hybrid chips may begin to appear around 2011. Others, such as Sunlin Chou, general manager of the Technology and Manufacturing Group at Intel, believe that by that point, the industry may only have a good grasp of the likely alternatives.

Nonetheless, Intel is exploring all the options. Like other companies, it is experimenting with the possibility of using nanowires and nanotubes in transistors, Chou said in a press conference earlier this week.

Other promising ideas include spintronics, the science of creating 1s and 0s by controlling the spin of electrons. Another possibility is the use of "phase change" devices, in which data is recorded by changing the physical state of a medium.