Reproducibility is one of the great strengths of the hard sciences. Put simply, it is a system of self-enquiry anchored by precise measurements that yield quantitative results within a shared model of understanding. Transparency between researchers ensures that other scientists can read a study, understand their methods, reanalyze their data, and conduct replication studies. So when research doesn’t live up to these standards, scientists become suspicious, which is why recent claimed breakthroughs in high-temperature superconductors start raising red flags.
We should point out in advance that the investigation is still ongoing. Scientific articles have been withdrawn, inconsistencies and errors have been acknowledged, and allegations of plagiarism have been raised. We don’t know the whole truth yet, but the process works and we will find it eventually. We should soon know if there are acceptable explanations for apparent anomalies, or if a broader swath of untruths is being purged from the scientific literature.
The holy grail of superconductors
In 2020, Professor Ranga Dias’ research group published an article in the prestigious journal Nature. In it, they claimed to have found a new material that is superconducting at around room temperature. In the 120 years since superconductivity (the ability to perfectly transmit electricity without resistance) was first observed, every known superconductor required a low temperature to function. Mercury was the first, and it only becomes superconducting when it is colder than five Kelvin above absolute zero. (5K is -451°F.)
During the 20th century researchers found superconductors that could withstand increasing temperatures: 10 K, 25 K and up to 150 K (-190 °F). Progress stalled before a superconductor was found that could withstand room temperatures of around 295 K (70 °F). Dias’ work was a breakthrough.
The discovery came with a caveat. While the new material – carbonaceous sulfur hydride (CSH) – remained superconducting at room temperature, it only did so under enormous pressures of more than two million atmospheres. (To put that in perspective, the air pressure around us is about one atmosphere.) Days ago, the Dias group announced another room-temperature superconductor: lutetium hydride doped with nitrogen (NLH). It is said to be superconducting at room temperature under a pressure of only ten thousand atmospheres. In theory, this is another big step towards a practical superconductor. But the announcement was met with more skepticism than applause.
Shortly after 2020 Nature Papers, independent expert physicists led by Professor Jorge Hirsch set out to test Dias’ work themselves. (For brevity, we’ll call them “replicators.”) They performed a thorough reanalysis, focusing primarily on the data in the paper but eventually also looking at other related work by the same authors. Their results uncovered startling anomalies in the data supposedly proving superconductivity.
The details are complex, but essentially, the replicators discovered that the mathematical patterns (signal vs. noise) in the data could not be obtained as Dias’ team claimed. In a written response, Dias modified the description of his analysis technique. The replicators tried that too, also without success. Thus, even under the changed analysis, the reported signal was still anomalous.
Unfortunately, the replicators came to the conclusion that it was a third analysis technique was consistent with the published data: it could have been created by combining a noiseless pure signal with some kind of artificial noise, and then even more noise could be added. (Here’s the full technical explanation.) But there’s a problem with this “technique”: It’s simple data fabrication.
Things are beginning to unravel
The Replicators found more warning signs. Hirsch looked up a 2009 paper on superconductivity led by one of Dias’ collaborators in 2020 Nature Paper. Hirsch requested the original data (which was not met), but it inspired one of the 2009 paper’s co-authors to verify the data himself. Physicist James Hamlin says he expected to exonerate the work he was involved in, but came to believe the data was actually tampered with.
Subscribe for counterintuitive, surprising, and impactful stories delivered to your inbox every Thursday
As a result, the paper was soon withdrawn, with official statements that “some if not all” of the results were due to “a subtle experimental artifact”, that some of the data “were not accurately reported” and that the most recent measurements found “none Evidence” for the discovery claimed in the newspaper.
Hamlin looked further and found additional problems in Dias’ work. He independently reported anomalies in 2020 Nature paper which has also been formally withdrawn. Hamlin claimed, and others confirmed, that several parts of Dias’s dissertation appear to be a plagiarism of Hamlin’s own dissertation. He alerted the authors of another of Dias’ major studies that it also appeared to contain plagiarized material. One of the authors of that study looked at the evidence, couldn’t find a plausible explanation, and then requested that it be withdrawn. It is currently under investigation by the publisher.
Finally, experimental replication studies of Dias’ CSH superconductor work — asking not whether the reported data were anomalous, but whether other scientists could independently synthesize a similar sample and reproduce the superconductivity experiment — have failed. Theorists weren’t lucky either.
A great mess
As of March 2023, the superconductor discoveries – if we can actually use that word – will be verified by other scientists. So far, two papers have been withdrawn and a third is under investigation. Further allegations of plagiarism seem convincing, but are rejected. Replication experiments failed. Regardless of the final verdict, there is a silver lining: the introspective and self-critical nature of the scholarly inquiry works as intended.