In the furore that followed last year's revelation that Jan Hendrik Schön had fabricated data in a string of high-profile papers, one question remained unanswered: was there a nugget of truth in any of his reported findings? Two teams have now attempted to replicate some of Schön's key results, and find that little of use can be salvaged from them.

The two groups recreated studies by Schön, who was based at Bell Laboratories in Murray Hill, New Jersey, on molecular field-effect transistors. The papers that they revisited featured examples of 7 of the 16 different types of data falsification or fabrication of which Schön was found guilty last October1.

Field-effect transistors — electronic devices that can switch and amplify currents — are the workhorses of silicon circuits. Schön claimed to have observed transistor-like behaviour in a layer of an organic compound, just one molecule thick, sandwiched between two metal electrodes. This suggested that the size of transistors, and thus of integrated circuits, could be hugely reduced.

Teams of physicists at the Delft University of Technology in the Netherlands and IBM's Thomas J. Watson Research Center in Yorktown Heights, New York, have now investigated devices based on Schön's design2,3. Both groups followed the methods in his papers: the electrodes were made from thin layers of gold, between which rod-like organic molecules capped with gold-binding chemical groups provided a conducting pathway.

Neither group obtained the results recorded by Schön. Jeong-O. Lee and colleagues at Delft, for example, say that their devices were unreliable from the outset. Of more than 1,000 prepared, many were short-circuited — the current simply flowed between the two electrodes, heedless of the intervening layer of molecules. Only in 5–16% of devices did the current seem to flow through the molecular layer. Even then, the devices quickly degraded.

Coming from two of the world's most experienced molecular-electronics labs, the results imply that the approach is fundamentally flawed. But the new studies have highlighted requirements for making similar molecular transistors, such as better methods for purifying the components. “There is no scientific reason to lose faith in the tremendous promise of the field,” insists Cherie Kagan, a member of the IBM group.