Why a blockbuster superconductivity claim met a wall of scepticism

Rumours were flying in Las Vegas, Nevada, ahead of a session of the American Physical Society meeting on 7 March. The buzz was that Ranga Dias, a physicist at the University of Rochester in New York, would be making a stunning announcement. When Dias’s 3 p.m. session started, so many scientists and journalists were packed into room 212 that meeting organizers had to usher latecomers into an overflow site to watch his presentation on video.

Dias didn’t disappoint. He and his colleagues had created a material — lutetium mixed with nitrogen and hydrogen, or LuNH — that was a superconductor at room temperature, he announced.

That claim, which the scientists published in Nature the next day¹, would have been historic — if true. Superconductors conduct electricity with zero resistance, meaning that no energy is lost as heat. But they usually work only at very low temperatures, well below −100 °C, so need expensive refrigeration. This limits their use to niche applications, such as magnetic resonance imaging scans and quantum computing. A superconducting material that needs no cooling could potentially transform electricity generation and transmission, transportation and a slew of other applications.

Dias’s claim was remarkable not only for the material’s balmy operating temperature of 21 °C (294 K), but also because it required comparatively modest pressures. Other teams working with hydrogen compounds, called hydrides, have observed superconductivity at high temperatures, but had to squeeze their samples to hundreds of gigapascals (GPa) — millions of times more than atmospheric pressure. Dias, by contrast, said that his hydride needed just 1 GPa (10,000 times atmospheric pressure): still impractical for real-world applications, but a striking advance. A patent application for LuNH, released in April, goes further, claiming superconductivity at room temperature and pressure.

“This was potentially a Nobel prize discovery,” says Lilia Boeri, a theoretical condensed-matter physicist at the Sapienza University of Rome. However, Boeri and many other physicists aren’t enthusiastic about the claim. Of 21 specialists outside Dias’s group who talked to Nature’s news team (which is editorially independent of the journal), only two, both of whom had professional connections to Dias, were positive about his work. Numerous other groups in the United States and China have tried and failed to replicate the result. Only one, in Chicago, has declared partial success, but with not enough evidence yet to be convincing. And in September, Nature added a note to the paper alerting readers that “the reliability of data presented in this manuscript is currently in question”.

Claims of room-temperature superconductors pop up regularly. This July, for instance, South Korean researchers reported superconductivity at more than 100 °C and at ambient pressure, in a material they called LK-99. The announcement went viral, but other researchers swiftly showed that the material wasn’t a superconductor after all. Dias and his colleagues, however, have claimed room-temperature superconductivity multiple times. They reported it in a different hydride in 2020, another result that they published in Nature, but which was retracted last year² (see ‘Seeking ambient superconductors’).

The scepticism over Dias’s claims stems in part from specific questions about his methods. But it’s also because of wider concerns about his track record. Some researchers have accused Dias and a co-author, Ashkan Salamat, at the University of Nevada in Las Vegas, of fraud over the retracted 2020 paper². This year, Dias was also accused of plagiarism in his PhD thesis, and has had a second paper retracted, from the journal Physical Review Letters (PRL), leading to a further erosion of trust. After that retraction in August, his university said it would open an investigation into his work.

“The data, on the surface, is quite compelling,” says James Hamlin, a physicist at the University of Florida in Gainesville, of the LuNH work. “But knowing the history, it’s hard to put much stock in it.”

Dias, who responded to questions from Nature’s news team by e-mail, stands by his results and insists that some of his detractors are attacking him unfairly. “They’re interested in destroying me,” he says.

Hydride surprise

Although sensational, Dias’s claims didn’t come out of the blue. For the better part of a decade, physicists have been crushing hydrides between the tips of diamonds in high-pressure devices called anvil cells, then cooling these crystalline materials while keeping them under pressure. When the temperature drops low enough — at a point called the critical temperature (T_c) — the hydrides suddenly lose all electrical resistance.

Theorists think that they know how this works. Superconductivity occurs when electrons couple up into ‘Cooper pairs’ and swoop through a material without bumping into any atomic nuclei. This frictionless motion ceases above a certain temperature because thermal fluctuations break up the pairs. Vibrations in a crystal lattice help the Cooper pairs to form — which is where the hydrides come in. A crystalline lattice made of hydrogen’s very light atoms could vibrate quicker and create more strongly bound electron pairs than would a lattice of any other element, allowing the Cooper pairs to withstand higher temperatures.

Stunning room-temperature-superconductor claim is retracted

The catch is that huge pressures — around 400–500 GPa — are needed to break up hydrogen’s molecular structure and create a metallic lattice. But in 2004, physicist Neil Ashcroft, then at Cornell University in Ithaca, New York, argued that alloys made from hydrogen plus a heavier element might become metallic at lower pressures than would hydrogen alone. The idea, Boeri says, is that hydrogen atoms would create a cage around each bigger atom, yielding a dense and stable lattice.

Experimentalists have since been chasing this lead. In 2015, Mikhail Eremets at the Max Planck Institute for Chemistry in Mainz, Germany, and his colleagues reported³ a T_c of −70 °C (203 K) in hydrogen sulfide at about 145 GPa. Dozens of other superconducting hydrides, some with higher T_cs, followed. By 2019, Russell Hemley at the University of Illinois, Chicago, and his colleagues reached −13 °C (260 K) in lanthanum hydride⁴ at 180–200 GPa.

Then, in October 2020, Dias’s group announced in its Nature paper² that it had reached room temperature — about 15 °C (288 K) — by squeezing a mixture of hydrogen, sulfur and carbon (CSH) to around 267 GPa. The feat made media headlines. But scientists were puzzled; unlike previous reports of superconductivity in hydrides, the finding had not been predicted by theory. Plus, others couldn’t repeat the result. Eremets says that he spent “more than half a year of intense efforts and resources” trying to replicate it; he says the synthesis protocol wasn’t clear in the paper and Dias didn’t help to clarify it.

Concerns over background

Other researchers criticized Dias’s paper directly. Physicists Dirk van der Marel at the University of Geneva in Switzerland and Jorge Hirsch at the University of California, San Diego, as well as Hamlin, took aim at its results on magnetic susceptibility, a measure of how much a material is magnetized when exposed to an external magnetic field⁵.

Such measurements are needed because plummeting electrical resistance isn’t enough to prove that a material exhibits superconductivity. Among other issues, a sudden drop in resistance could simply indicate that a material has transformed from an insulator in to a good conductor, but not a superconductor.

Scientists can remove such doubts by measuring a second hallmark of superconductivity: the Meissner effect, which involves a material repelling a magnetic field. This effect is hard to demonstrate directly in high-pressure experiments, so researchers make do with an indirect measure: a drop in magnetic susceptibility at the material’s T_c. But even this measurement is complicated by the need to filter out large, everyday magnetic fluctuations in the laboratory.

LK-99 isn’t a superconductor — how science sleuths solved the mystery

Dias’s team said it had filtered out this background by measuring magnetic susceptibility versus temperature at a lower pressure (when the material didn’t superconduct), and then subtracting that from its high-pressure measurements. Hirsch wasn’t convinced that the team had subtracted background data in this way and asked them to provide the raw, unfiltered data. Dias and one of his co-authors, Salamat, eventually did so in a preprint⁶ in late 2021. But, two months later, in response to critical comments from Hirsch and van der Marel about that work⁵, they uploaded another preprint, in which they said that they had subtracted something different from their raw measurements to generate the results in their 2020 Nature paper: what they described as a “user-defined background”⁷.

That was a step they had not spelled out in their original report, prompting Nature to retract the paper in September 2022. “These processing issues undermine confidence in the published magnetic-susceptibility data as a whole,” the journal editors said in a statement accompanying the retraction, which the paper’s authors disagreed with.

Hirsch and van der Marel said that Nature’s retraction was not strong enough. They examined correlations between the data in the 2020 paper and the raw data released in 2021. And they argued that Dias and Salamat had produced the raw data afterwards by adding a background signal to the final data — rather than, as the scientists had claimed, subtracting a measured background from the raw data⁸.

“Stating that something is measured when it is not measured is fraud,” Hirsch says. Van der Marel agrees: “I wouldn’t know how else to describe this other than fraud,” he says.

When asked to respond to these allegations for this article, Dias did not provide specific rebuttals. He said only that “Dr. Hirsch’s objections have been answered”, and did not respond to requests for clarification; Salamat did not reply to requests for comment. “Our work and our evidence speak for itself,” Dias added, noting that others have now observed superconductivity in CSH. This February, Hemley and his colleagues at the University of Illinois, with Dias’s group and others in Nevada, published a preprint that reports superconductivity in CSH samples⁹. Their materials had more carbon than those described in the retracted 2020 paper, yielding superconductivity at a slightly lower T_c of up to −13 °C (260 K) but at a lower pressure, of 133 GPa.

The researchers say that they rebuilt their apparatus and added more magnetic coils, allowing them to boost the samples’ magnetic-susceptibility signal so that they no longer needed to remove background.

An even bolder claim

Despite the retraction of the previous Nature paper, Dias’s team was in buoyant mood when announcing its achievement in March. The LuNH result marked “the dawn of ambient superconductivity and applied technologies”, a university press release said.

Dias and his colleagues write in their paper that they loaded tiny foil pieces of lutetium into an anvil cell, before adding gas with a 99:1 mixture of hydrogen to nitrogen, compressing the mixture to 2 GPa and heating it in an oven overnight at 65 °C. When the material was restored to ambient conditions, it looked blue. With the pressure raised to 0.3 GPa it turned pink, and at around 3 GPa it took on a red colour.

The researchers made more than 100 such samples. Holding each one at a certain pressure, they warmed them up and cooled them down while measuring one of four properties: electrical resistance; magnetic susceptibility using both alternating and direct current; or heat capacity. Although most samples did not show signs of superconductivity, those that did yielded a specific T_c that depended on the pressure that they were subjected to. A plot of these temperatures against pressure produced a dome-shaped curve that shows LuNH’s maximum T_c as 21 °C (294 K) at around 1 GPa — two orders of magnitude lower than any previous result.

Isaac Silvera at Harvard University in Cambridge, Massachusetts, who supervised Dias’s postdoctoral research, says that the results look solid. “The fact that they see the dome is very good evidence,” he says.

But Silvera’s view is not typical. Two other scientists who spoke to Nature’s news team were on the fence, including physicist David Ceperley at the University of Illinois Urbana-Champaign, who wrote a News & Views article in Nature commenting on the findings of Dias’s team and who says he is “willing to give them a chance”. But all others were critical, including News & Views co-author ChangQing Jin at the Chinese Academy of Sciences’ institute of physics in Beijing. “If those measurements are true, then it should be a superconductor,” he says of LuNH. “But I really doubt where they get their measurements from.”

How would room-temperature superconductors change science?

Critiques started appearing as soon as the Nature paper was published. One major line of criticism is that the Rochester team didn’t provide enough evidence to show that resistance does go to zero in its material. Dias and his colleagues state in the paper that they removed “small residual resistance” from some of their electrical measurements, but critics argue that it should not be necessary to remove background for these types of measurements, given clean readings of both a sample’s current and voltage. The problem with removing a background, says Sven Friedemann, a physicist at the University of Bristol, UK, is that it implies that the raw data do not go to zero — and therefore don’t show superconductivity.

Dias and his co-authors don’t say in the paper whether they removed a background from their main chart that shows a drop in resistance. In an appendix, however, they supply further evidence for superconductivity in a chart from a separate experiment, involving measuring how the T_c changes when magnetic fields of varying strengths are applied. In that case, they do state that they removed a background, and also provide the raw data. Some researchers have used those data to plot a fresh graph without the background subtracted. That shows resistance dropping fairly gradually with temperature, and remaining significantly above zero, says Peter Armitage, a physicist at Johns Hopkins University in Baltimore, Maryland, who posted his result on the online journal club PubPeer in March (go.nature.com/3osq6u8). He is not the only researcher to point this out. This, says Armitage, implies no superconductivity.

In response, Dias told Nature’s news team that the researchers did not subtract a background signal for any chart of electrical resistance except that in the appendix. (Other co-authors of the paper, and of the retracted 2020 paper, did not reply to requests for comment.)

Partial replication?

Among the researchers who haven’t criticized Dias’s results is Hemley, who has a long track record in the field of high-pressure experiments. After their work on CSH, Hemley and his colleagues uploaded another arXiv preprint, in June, reporting observations of superconductivity in LuNH, using samples from the Rochester team¹⁰. The researchers measured only electrical resistance (which is not enough to prove superconductivity) and generated just four data points, with one reaching a T_c of 3 °C (276 K). But they argue that their temperature–pressure curve is similar to that of the Rochester group’s. Hemley acknowledges that this constitutes only partial confirmation, but says that magnetic-susceptibility measurements are “in progress”.

Hemley’s preprint has made some researchers more positive about Dias’s work, with Friedemann describing the replication effort as “a first step in the right direction” to resolving the controversy about whether this material is a near-room-temperature superconductor. But there are doubts here, too. Alexander Drozdov, a colleague of Eremets at the Max Planck Institute for Chemistry, says that the claimed superconductivity could be because of bad electrical contacts, and notes that Hemley’s results show rises in resistance below the temperature at which the preprint finds that superconductivity appears. Physicist Douglas Natelson, at Rice University in Houston, Texas, agrees. “This looks exactly like what would be seen if one of the contacts was bad,” he wrote in a blog post in June.

Hemley, however, says that these effects are caused by variations in crystal composition or structure across the sample. His team’s measurements show that the impure LuNH material contains both non-superconducting and superconducting phases, and that in such cases, high electrical current density can quench (that is, eliminate) superconductivity in regions of the sample, hence the variations in resistance, he says. He adds that similar effects were seen in heterogeneous samples of the lanthanum hydride that his own team discovered, which others subsequently confirmed to be superconducting.

No one else has come close to reproducing the findings of Dias’s team — a task made harder because only Hemley is known to have received original samples. Dias told Nature’s news team that he had given samples to other researchers who, in his opinion, “are serious about the science and trustworthy”; beyond Hemley, he didn’t say who those people were. A group led by physicist Hai-Hu Wen at Nanjing University in China reported in Nature in May that it saw no signs of superconductivity in LuNH it made (although it saw similar colour changes)¹¹. Dias says that the Nanjing group didn’t have adequate samples or do enough testing. Two other Chinese groups similarly got null results^12,13.

A fourth group in China says that it found the real reason for the sudden drop in resistance in LuNH¹⁴. Ho-kwang Mao and his colleagues at the Center for High Pressure Science and Technology Advanced Research in Shanghai reported drops in resistance at −23 °C (250 K) in their samples of LuNH. They argue that this is probably because of the material transitioning from a poor conductor above the T_c to a better conductor below it (although still inferior to pure lutetium metal). They didn’t see any evidence of zero resistance. Dias, in response, says that the experimental conditions were different in the two cases. Mao’s group made and measured its samples at 10 GPa, whereas Dias says that his team looked for superconductivity from roughly 0.3 to 3 GPa. Mao says that his team has since looked across a range of pressures and still finds the same behaviour.

Other physicists have also obtained negative findings. Superconductivity researcher Paul Chu at the University of Houston in Texas says that he and his colleagues saw the colour changes but not superconductivity, albeit without putting nitrogen into the mix; he says he also knows of two other US groups who’ve tried and failed to reproduce the work.

Researchers from four other groups, meanwhile, told Nature’s news team that they had abandoned their own attempts to replicate the work or hadn’t even tried. Eremets said that he wasted time on the CSH work, so didn’t bother with LuNH. “I just ignored it,” he says.

Differences also remain on the theory side. Several groups have simulated many different arrangements of lutetium, nitrogen and hydrogen atoms to see whether any might yield superconductivity at high temperatures. But most have come up empty-handed, Boeri says, including one that she led. Hemley’s group, in a preprint published on arXiv in May, says that it found a class of structures — LuH₃ doped with nitrogen — that could work, allowing a large number of electrons to form Cooper pairs¹⁵. But Boeri argues that the group has not provided convincing evidence that such structures can form, and also hasn’t shown that the structures would vibrate quickly enough for room-temperature superconductivity.

Choice to publish

With all the question marks hanging over Dias’s research, many scientists are highly critical of the way that Nature handled the LuNH paper and want to know on what basis the journal decided to publish it. Armitage says that it took him only two minutes to find what he describes as “disqualifying problems” — among other things, the form of the raw resistance data in the supplementary figure. Boeri says that Nature should have required the authors to publish the precise chemical composition of their samples so that others had the recipe. And Hirsch says that editors should have first resolved the question about the provenance of the raw data in the retracted 2020 Nature article before even considering the 2023 paper.

The editor’s note added to the 2023 paper on 1 September, saying that the reliability of data are in question, adds that “appropriate editorial action will be taken once this matter is resolved.” Karl Ziemelis, Nature’s chief applied- and physical-sciences editor, based in London, says that he and his colleagues are “assessing concerns” about the paper, and adds: “Owing to the confidentiality of the peer-review process we cannot discuss specific details of what transpired.”

As for the 2020 paper, Ziemelis explains that they decided not to look into the origin of the data once they had established problems with the data processing and then retracted the research. “Our broader investigation of that work ceased at that point,” he says. Ziemelis adds that “all submitted manuscripts are considered independently on the basis of the quality and timeliness of their science”.

Other condensed-matter physicists, including Siddharth Saxena at the University of Cambridge, UK, and Marie-Aude Measson at the Néel Institute in Grenoble, France, want to see what peer reviewers said about the 2023 paper. Nature allows authors to opt out of having peer-review reports published; separately, referees can choose whether to reveal their names. For the Dias paper, neither review reports nor referee names were released — a situation that has occurred with about 10% of the papers that the journal has published so far in 2023, according to an analysis by Nature’s news team.

Wider allegations

Beyond the superconductor claims, researchers have found problems in some of Dias’s other work. This year, Hamlin alleged that large swathes of Dias’s 2013 PhD thesis, at Washington State University (WSU) in Pullman, were copied from his own; Hamlin says that Dias took material from other papers, too. A WSU spokesperson said that the university is aware of concerns but would not say whether it had started an investigation. Dias acknowledges that he failed to provide “explicit attribution” for some aspects of the work, and says that he will remove portions of the thesis “and/or properly attribute the needed sources”.

‘A very disturbing picture’: another retraction imminent for controversial physicist

The problems don’t end there. This August, Dias, Salamat and their colleagues incurred a second retraction for their work, after Hamlin spotted a plot in Dias’s PhD thesis that looked very similar to a PRL paper that Dias had co-authored in 2021 on an unrelated material. PRL commissioned an investigation by four independent referees, whose findings “back up the allegations of data fabrication/falsification convincingly”, the journal’s editors wrote in documents about the investigation obtained by Nature’s news team. The final retraction statement did not mention data fabrication, however¹⁶. All of the authors except Dias signed the retraction; the statement adds that Dias stands by the data in the plot and does not agree with the retraction.

Dias says he remains “certain” that there was no scientific misconduct with the paper, and that he was looking to “resolve the several questions that exist regarding the data and the figures in the paper”. (Other co-authors declined to comment or did not reply).

After the retraction, the University of Rochester said that it had “a comprehensive investigation under way into the questions raised about the integrity of all data at issue in this and other studies”, according to spokesperson Sara Miller. Previously, the university had carried out two inquiries regarding Dias’s 2020 paper, which, in May 2022, concluded that “there was no evidence that supported the concerns”, Miller said (she did not say who conducted the inquiries or what concerns they investigated). The investigation is being carried out by outside specialists and will involve “the collection of information from many sources”, Miller added.

Commercial interests

Another tension comes from the commercial interests of the University of Rochester and the researchers. In 2020, Dias and Salamat founded a firm called Unearthly Materials, also based in Rochester, to commercialize room-temperature superconductors. Dias says that he and his colleagues have since raised US$16.5 million, but he won’t say who the investors are.

And in July 2022, the University of Rochester applied for a patent on the LuNH work, listing Dias as the sole inventor. This became publicly known only when the patent application was published in April. The document provides a different recipe from that in the Nature paper, stating that the samples were heated to 200–400 °C rather than just 65 °C, for instance. Although it has a temperature–pressure plot that looks similar to the paper’s, it also claims that two samples achieved room-temperature superconductivity at ambient pressure. Friedemann says that many of the uncertainties around the original work remain for this even more ambitious claim — including whether the researchers subtracted a background from their resistance measurements. Dias, again, says that they did not.

Beyond specific details on the papers or the patent, Friedemann says that other scientists are becoming increasingly wary of research from Dias owing to the two retractions.

In principle, he says, any set of results should be taken at face value, with scientists being “only interested in what nature does”. But in practice, he adds, researchers have to weigh up how best to use their funding and time. The controversy, he says, “makes it difficult to strike the right balance between devoting the efforts room-temperature superconductivity deserves and the risk of going down a dead end”.