A key trick, is to add thiol side chains to a polymer, which binds the tubes to gold electrodes.
“The results look deceptively simple: a self-assembled transistor with nearly 100% purity and very high electron mobility, but it took ten years to get there,” said the University of Groningen, whose Professor Maria Antonietta Loi designed the polymers.
Carbon nanotubes are not all the same, and have properties ranging from semiconductor to semi-metallic to metallic.
Making them always produces a mix, but only semiconductor tubes can be used to fabricate transistors.
In a solution of mixed tubes, Loi’s polymers are tuned to selectively wrap themselves around specific carbon nanotubes – semiconducting types, for example.
Adding thiols – a type of suphur-containing molecule – then makes the selected polymer attach the nanaotube to the gold.
There is a ‘but’ – the thiols reduce selectivity.
“We had the idea of using polymers with thiol side chains some time ago”, said Loi.
IBM, according to Groningen, also patented the concept.
“But there was a big problem in the IBM work: the polymers with thiols also attached to metallic nanotubes, and included them in the transistors, which ruined them,” said Loi.
Groningen scientists have now worked with IBM Zurich, and the University of Wuppertal, to get a polymer which remains selective while retaining the ability to attach to gold.
With scientists at Wuppertal, Loi reduced the thiol content of the polymers.
“What we have now shown is that this concept of bottom-up assembly works: by using polymers with a low concentration of thiols, we can selectively bring semiconducting nanotubes from a solution onto a circuit,” she said.
The sulphur-gold bond is strong, fixing the nanotubes firmly enough to survive sonication in organic solvents.
To make a transistor, metallic patterns are deposited on a carrier, which is then dipped into a solution of carbon nanotubes.
Careful electrode spacing achieves proper alignment.
“The tubes are some 500nm long, and we placed the electrodes for the transistors at intervals of 300 nanometres. The next transistor is over 500 nanometres away,” said Loi.
Resulting FETs have mobilities up to 24cm2/Vs.
This spacing limits the density of the transistors, but Loi is confident that this could be increased. “It is difficult to predict whether the industry will develop this idea, but we are working on improvements, and this will eventually bring the idea closer to the market,” said Loi.
‘On-chip chemical self-assembly of semiconducting single-walled carbon nanotubes (SWNTs): towards robust and scale invariant SWNTs transistors‘ is published in Advanced Materials.