It is a debatable point and may ultimately be proven to be false, but it should and likely will govern how nanotechnology research is conducted and as a result we will surely have better work as a result.

Dexter Johnson, Spectrum journalist and judge

To quote Dexter Johnson (Spectrum journalist and judge), "The main thrust of Kelly's argument is one that is not altogether that radical, which is that you may be able to fabricate one-off structures that have dimensions below 3nm but you won't be able to duplicate that in a full-scale manufacturing process. It is a debatable point and may ultimately be proven to be false, but it should and likely will govern how nanotechnology research is conducted and as a result we will surely have better work as a result."

Professor Kelly summarises his argument below.

"If I give you a large piece of carpet, a 1.5m steel ruler and some carpet shears, and ask you to cut out 100 tiles each a metre square, you could do a pretty good job if you were careful: the tiles could be stacked, and you could observe how reproducible the tiles are. If, with the same tools, I asked you to cut out 100 pieces the size and shape of the chip on your credit card, you would struggle to achieve reproducibility. The same type of argument applies to the manufacture of nanostructures, making ever smaller objects with the same fabrication tools. It is easy to make one, but extremely difficult to make thousands all the same, especially if you wanted an ordered array of such nanostructures.

"In a paper*, 'Intrinsic Top Down Unmanufacturability', published in April last year (after three years of trying and 10 prior rejections, not for being wrong, trivial or derivative, but for containing a profoundly negative result that would taint the journal if it were to publish it), I showed that an array of 3nm diameter pillars on a 6nm pitch is intrinsically unmanufacturable by the known top-down methods of fabrication. With 80 atoms per plane in the pillar, the statistics of small numbers predicts a standard deviation of 12%, and given that modern microelectronic manufacture works at a standard deviation for feature sizes of less than 2%, nothing useful will come of the attempts to manufacture. Even if one could make an array atom-by-atom, the nodes are too small to address, store and read information. All this has serious implications for the ultimate practicality of work on quantum dots, quantum wires, nanotubes, graphene and much else besides. Why waste money researching something that can be proven unmanufacturable? The better strategy would be to invent new methods of manufacture that circumvented the conditions of the theorem. In case you are wondering, I think that the bottom up method of fabrication will have different problems but reaching the same manufacurability impasse for a 3nm array on a 6nm pitch.

"The paper went viral on the internet, thanks to publicity from the Institute of Physics. Within a month there were 1400 pages reproducing the title of the paper, up to 2700 at the end of the year. No-one has come up with a counter-example as yet. I am pleased that the IEEE Spectrum has endorsed my views on more than one occasion. I have also given a 30 minute radio interview to a science radio programme in the US, and was asked to lead off an evening debate on the topic at the 2011 IEEE NANO meeting in Portland Oregon last summer. I have also just received an EPSRC research grant to pursue the ramifications, as I think that the practical limit for manufacture is not at 3nm but 7nm."

*M J Kelly, Intrinsic top-down unmanufacturability, Nanotechnology 22 (2011) 245303 (3pp) doi:10.1088/0957-4484/22/24/245303

The Second Annual Nanoclast Awards on the http://spectrum.ieee.org website.