It was an explosive claim: the discovery of a superconducting material that can carry electricity with virtually no resistance in normal, ambient conditions. The purported finding — announced by two Indian physicists in a preprint1 last month — sparked a rush of replication efforts. But independent researchers have grown increasingly sceptical as they have dissected the claim, in a process that played out mostly on social media.
“All these researchers who normally do not discuss on a single platform have come together and discussed this,” says Pratap Raychaudhuri, who studies low-temperature physics at the Tata Institute of Fundamental Research in Mumbai, India. He led a discussion of the results on Facebook. “I think the self-correcting mechanism of science — the ruthless scrutiny of the community — has worked extremely well,” he says.
Raychaudhuri also says that the episode is evidence of the value of posting preprints before publication and having an open discussion about them. “I think this is possibly going to set a very good precedent,” he says.
In the original preprint, posted1 to the arXiv server on 23 July, Dev Kumar Thapa and Anshu Pandey of the Indian Institute of Science in Bangalore (IISc) described a material made from gold and silver that became superconducting at a balmy –37 ˚C, and at normal ambient pressure. Most superconducting materials identified so far work only at much lower temperatures, often close to absolute zero. The highest seen yet is –70 ˚C, reported2 in 2015 — and that compound is superconducting only at extremely high pressures. (Just this week, the same laboratory posted3 a preprint on the arXiv describing a new record, –58˚C, for superconductivity at high pressure, but that result has not yet been confirmed.)
“It was a remarkable claim, so there was lots of interest,” says Raychaudhuri. Several laboratories quickly leapt into action to try to replicate the results. But their efforts were frustrated because the preprint did not provide the details needed to manufacture the gold–silver material, and because Thapa and Pandey declined requests to share their samples. “After two weeks or so, the community started to get more impatient,” Raychaudhuri says.
Thapa and Pandey told Nature’s news team that they would not comment on the details of their research while their paper is under review at a journal. Pandey said that they are having their results validated by independent experts, and that they will announce the results of the validation in the appropriate forum as soon as possible. “Without validation,” says Pandey, “the synthesis and device-fabrication details are speculative and will add to further confusion.”
Brian Skinner, a theoretical physicist at the Massachusetts Institute of Technology (MIT) in Cambridge, began studying the preprint soon after it came out — and eventually chronicled his findings in a widely shared Twitter thread. Although superconductors are not his specialty, the excitement surrounding the paper piqued his curiosity. He noticed that one of the preprint’s figures contained curves of data points that were surprisingly free of random background noise at relatively warm temperatures, but became noisier below the temperature at which the material transitioned to a superconducting state. “Usually, they look smooth on both sides, or dirty on both,” Skinner says.
When he zoomed into the picture, Skinner was even more surprised: the graphic included several data sets in which the experiment was run in slightly different conditions, and the patterns of noise seemed very similar for each run. But noise is, by nature, random. He went to discuss his observation with a fellow theorist at MIT who is an expert on superconductors. “I went into his office and closed the door.” The colleague agreed with him, and in the following days, Skinner had conversations with many other researchers — including experimentalists.
Repeated patterns of noise alone do not necessarily mean that the data are faulty or intentionally fabricated, Skinner says, but he still wanted the broader community to know about his concerns. So on 8 August, Skinner submitted a two-page response4 to the preprint on the arXiv. The post, which he mentioned on Twitter, prompted a viral response, with more than 3,600 shares and countless online mentions. The attention left him disoriented, he says. “It’s the standard story of someone being overwhelmed by sudden internet popularity.”
A separate group — led by experimentalist Mingda Li in MIT’s nuclear science and engineering department — that had also been attempting to replicate the results took note of Skinner’s post, and Li became concerned. “Fluctuations really shouldn’t be that identical,” he says. He called a meeting of his group, and together, they decided to call off the replication attempts. Pushan Ayyub, another physicist at the Tata Institute, says that his lab has not completely stopped trying to replicate the results, but “the pace is no longer frenetic”.
Meanwhile, Raychaudhuri and others took to Facebook. On his page, Raychaudhuri gave a possible explanation for the repeating patterns of the data points discovered by Skinner, and pointed out that to get to the bottom of the story, the authors would need to share their own data. “For the sake of healthy academic discourse, it is of paramount importance that Thapa and Pandey come out openly with their data and their samples,” he wrote.
The story also briefly took a “Da Vinci Code” turn, as Raychaudhuri puts it, when he received an e-mail supposedly from an eminent Indian physicist, who tried to dissuade him from criticizing the work of Thapa and Pandey. Raychaudhuri then realized that the e-mail had come from an anonymous but cunning impersonator, who appeared to have set up an e-mail account in the physicist’s name. The identity of that person remains unknown.
Although Raychaudhuri is not convinced by the claims, he sees a silver lining in the affair: it has provided an opportunity to show science in action. “There is an interest in superconductivity, and one can harness it to get more people interested,” he says.
As for the claims, says Li, “if the authors don’t provide any new experimental measurements, this will gradually go away.”
Nature 560, 539-540 (2018)