The Sciences

New Challenge to IISc Superconductivity Claim Spotlights Data Anomaly

A potentially big discovery, close scrutiny by the scientific community, doubts and questions, and validation of original data till the pall of confusion clears…

After an initial period of silence, the claim by two scientists of the Indian Institute of Science (IISc), Bengaluru, of having demonstrated superconductivity at room temperature has come under doubt after a physicist spotted some anomalous data in their paper.

Anshu Pandey and Dev Kumar Thapa had reported last week that they had discovered room-temperature superconductivity, a phenomenon scientists have been searching for many decades, in a composite nanostructure material made of silver and gold.

On August 9, Brian Skinner, a physicist at the Massachusetts Institute of Technology, Boston, uploaded a critique of the work to the arXiv preprints repository, where the IISc preprint was also posted. Skinner pointed out that data in the original paper for two independent sets of measurements, obtained under different experimental conditions, show identical background noise patterns. This is considered highly improbable because such noise is supposed to be random and unpredictable.

Pandey and Thapa had reported their investigation of the nanostructure, in which they claimed to have observed two important features that characterise the onset of superconductivity in a material: the sharp drop of electrical resistance towards zero and a similarly sudden fall in the magnetic susceptibility of the material. Both these phenomena occurred at around the same temperature of 235 K (-38.15º C), indicating a superconducting transition at this high temperature in this unsuspecting material.

The best way to keep news free and independent is to crowdfund it. Just Rs 7 per day from readers like you will keep The Wire going. To support The Wireclick here.

Magnetic susceptibility is “a measure of whether a material is attracted into or repelled out of a magnetic field”, to quote the author from The Wire‘s previous report introducing the finding.

The image below shows the Pandey-Thapa data on the variation of magnetic susceptibility with temperature under different applied external magnetic fields: 0.01 tesla (T) to 5 T. Magnifying the bottom part of the plot, Skinner noticed a peculiarly identical wavy pattern in the data with 0.1 T and 1 T (green and blue dots), except for a constant vertical offset between the two sets. This would suggest similar background noise in these two apparently independent measurements.

An image from Skinner’s preprint paper showing the similar noise patters from two different plots (overlapped as blue and green dots for comparison). Source: arXiv:1808.02929v1

An image from Skinner’s preprint paper showing the similar noise patters from two different plots (overlapped as blue and green dots for comparison). Source: arXiv:1808.02929v1

Further, this curious correlation in the two sets of data seems to be absent at temperatures above 225 K or so, Skinner noted. While the noise pattern in the red curve (0.01 T) is not clearly discernible, patterns in the data at 3 T and 5 T are distinctly different. “This unusual feature of repeated noise patterns in the magnetic susceptibility,” Skinner wrote in his arXiv comment, “has, to my knowledge, no precedent in superconducting literature, and no obvious theoretical explanation”. He also found it surprising that, while there is a relatively large amount of noise in the lower plateau, there is less noise in the rest of the graph.

But the real sting came not from this purely academic observation but a comment in a Twitter thread he published on August 10, wherein he refers to the infamous Schön scandal. Jan Hendrik Schön was a German semiconductor and nanotechnology physicist at Bell Labs who was later found to have fabricated the data in numerous papers. What had given him away was a similar observation of repeated noise patterns in a number of independent measurements, leading to the discovery of massive data fraud. As many as 28 of his papers had to be retracted from the reputed journals they had been published in.

To a reader, this reference obliquely suggests the possibility that Pandey and Thapa fabricated their data. This is somewhat unfortunate because one does not have the full details of the experimental conditions and the data the duo obtained.

Skinner told The Wire, “I am not an expert in superconductivity, and I am not an experimentalist. So it would be irresponsible for me to comment much beyond what I wrote in that short arXiv note… I would highly encourage you to get an opinion from someone more qualified to comment on these measurements than I am.”

The Wire had already done so. The unnamed experimentalist quoted in the previous report in The Wire made one further remark in the context of Skinner’s note: “I am also not quite comfortable with the susceptibility data. The observed diamagnetic transition is too clean and sharp for a granular superconductor.”

Another highly experienced experimentalist, though not engaged in superconductivity research but who also wished to remain anonymous, made the following observation countering Skinner’s note: “Usually, instruments are set to select the best dynamic range to suit the signal level. Hence, scatter [in the data] will look different in different parts of the graph. [This is] perfectly reasonable and not a mystery. Unless one knows the configuration of data acquisition, and the evolution of noise as a function of temperature, say with a blank sample, the measurements cannot be doubted, a priori. Only after access to the table of data, a more sound judgement can be made.”

He also added that “noise is often periodic, 50 Hz being one of the prominent culprits.” This is a common frequency of the power supply via alternating current. “If you turn on an instrument at its lowest (most sensitive) range, the noise is likely to be periodic due to pickup from ambient sources, and is very likely to be identical from one run to another.”

Skinner responded that he couldn’t see “how these effects could give two curves – 0.1 T and 1 T – with the exact same pattern of noise, while leaving two other curves of the same family – 3 T and 5 T – with completely uncorrelated noise.”

The physicist had also clarified in an email that he was not making an accusation or calling their measurements fraudulent, and that he was only remarking that the correlation was very strange and needed to be understood further. “I have encouraged the authors to make a short post on the arXiv where they detail exactly what their measurement procedure was and show their susceptibility data in a more thorough way,” Skinner said. ”In that way perhaps, we can get to the bottom of this mystery and any cloud of suspicion (one hopes) can clear.”

Good journalism costs money. We’ve made it easier for you to fund The Wire. Now, you can support us by giving as little as Rs 200 every month. More details here.

According to his latest tweets, Pandey and Thapa had responded to his suggestion that was along the lines of – per Skinner – “Thanks for pointing this out! We hadn’t noticed this peculiar noise correlation. We don’t know its origin yet.”

It seems the authors have not backed down from their original claims, and that they are now focused on providing information that could help validate their data. It seems they have also agreed to post new data or a response to Skinner’s note only once they have performed the validation, according to Skinner.

What one is witnessing here is science at work: a potentially big discovery, close scrutiny by the scientific community, doubts and questions, and validation of original data till the pall of confusion is cleared.

Of course, even after that, the scientific community at large will be able to accept the discovery as such only after other experimentalists have been able to reproduce the observation and determine that the original finding is indeed true.

R. Ramachandran is a science writer.