Written November 21, 2021. Contact me@spxtr.net with any questions or comments.
One of the most important unsolved problems in physics is room-temperature superconductivity. At ambient pressure, the highest measured superconducting critical temperature is -140°C in a class of materials known as cuprates. For context, nitrogen liquifies at -200°C, and carbon dioxide solidifies into dry ice at -80°C.
To prove superconductivity in a new material, scientists should demonstrate vanishing electrical resistivity and the Meissner effect upon cooling below a critical temperature. In practice, it can be difficult to perform certain measurements on certain samples, so either corners are cut, or other measurements are substituted. We must then consider heuristics to decide if a claim of superconductivity is credible or not while we wait for additional measurements. In some cases it is clear one way or the other, but in one recent case it is not at all clear.
In 2018, a group of researchers reported measurements of room-temperature superconductivity at ambient pressure in a material consisting of silver nanoparticles embedded in a gold matrix.
At the time, I evaluated the paper based on these heuristics:
I decided that most likely the authors either fabricated the data, or else their measurements were not performed carefully. In either case, it was not worth my effort to attempt to replicate their work. Shortly thereafter, Brian Skinner pushed a comment to arXiv, noting that there was a repeating pattern in the noise in one of the figures. It ended with a brutal sentence:
This unusual feature of repeated noise in the magnetic susceptibility has, to my knowledge, no precedent in the superconducting literature, and no obvious theoretical explanation.
As the superconductivity has never been reproduced and the authors refused to share their samples with other groups for independent verification, the claim was nearly certainly bogus.
Also in 2018, Pablo Jarillo-Herrero's group at MIT published a report of superconductivity in bilayer graphene with a 1.1° twist angle between the layers. The low critical temperature of around 2 degrees above absolute zero made the system seem useless, practically speaking, and it was. However, it was a big deal to physicists, because it appeared to be "unconventional" superconductivity, with features that were remarkably reminiscent of the cuprates. Graphene was signficantly easier to work with than cuprates, so the hope was (and still is) to use graphene to further our understanding of superconductivity in general, and someday use this understanding to make room-temperature superconductors. That may sound far-fetched, but it's where we are.
I evaluated it as follows:
I immediately started attempting to replicate and extend this work, and I succeeded shortly thereafter, as did numerous other groups. It is now the hottest field in condensed matter physics.
At high pressures (over 100 GPa, 1 atmosphere is around 100,000 Pa), it is suspected that certain hydrogen-rich materials superconduct at temperatures even higher than the cuprates. In October 2020, Snider et al. published a landmark paper in Nature: they finally observed superconductivity above 0°C in a material containing carbon, sulfur, and hydrogen. This was practically useless, as it required over 200 GPa of pressure. Still, a Big Deal.
My evaluation was:
I did not suspect for a moment that there was any foul play going on, because I did not look very closely at the results.
Jorge E. Hirsch is a physicist probably most well-known for inventing the h-index. He's also a conspiracy theorist who had some absurd ideas about Iran, nuclear weapons, and avian flu. See his Wikipedia page if you're interested. In 2020, when the Nature paper was published, Hirsch followed it with a "matters arising" article arguing that the measurements were likely not of a true superconductor. This was because of a particular feature of the data: the width of the superconducting transition should in theory have been much higher than observed, and it should have increased in a magnetic field.
In October, 2021, Hirsch strengthened his claim: not only were the measurements wrong, but they were likely fabricated. In other words, the authors committed blatant scientific fraud.
The paper was fun to read, featuring many wtf and smh moments. There were roughly three important points.
Although the latter two points were damning, I found the former particularly sad. Scientists are supposed to openly share all information that is required to evaluate their research's validity. It is frustratingly common for them to refuse to do so, and I have personally experienced this on multiple occasions.
The authors of the Nature paper claimed that their data was available upon reasonable request:
Data availability: The data supporting the findings of this study are available within the article and its Supplementary Information files, and from the corresponding author upon reasonable request.
The main measurements in the paper were of resistivity vs temperature and magnetic susceptibility vs temperature. What was shown in the paper was not the raw numbers that came out of their instruments, but was instead background-subtracted. They measured the sample first in some parameter regime without the superconducting transition (the "background"), then again with the transition (the "raw data"), and then they subtracted the two measurements to make their figures. This background subtraction could have a significant impact on the results, and so Hirsch asked the authors for the raw data. He was refused, on the grounds that the authors may wish to patent the results in the future. Completely independently, Evgueni Talantsev's request for the same data was refused on the same grounds.
This is obviously not acceptable. In the first place, the data for such an important result should be available in a public repository for anyone to inspect, without requiring approval of the author. In the second place, "we may wish to file a patent in the future based on the data" should never be an acceptable excuse for refusing to allow others to evaluate your scientific claims. This excuse is fully general and can be used in nearly any situation that an author does not wish to share their data with a peer.
The only morally acceptable excuse for refusing to share raw data is that it contains PII of research subjects.
Nature ostensibly agrees with me, as on their website they say
A condition of publication in a Nature Portfolio journal is that authors are required to make materials, data, code, and associated protocols promptly available to readers without undue qualifications.
However, when contacted by Hirsch, Nature refused to enforce this rule. Instead, they agreed with the authors that the possibility of future patents voided the requirement to share data.
As did NSF, the funding body for the work, who made the same empty claim as Nature:
Investigators are expected to share with other researchers, at no more than incremental cost and within a reasonable time, the primary data, samples, physical collections and other supporting materials created or gathered in the course of work under NSF grants. Grantees are expected to encourage and facilitate such sharing.
The NSF refused to take any action against the authors. Hirsch concludes,
... there should be procedures and regulations in place that make it difficult for scientists to commit undetected fraud. Such provisions exist on paper, but from what I have experienced and discussed in this paper it should be evident that they are not enforced, neither by the most high profile journals nor by the preeminent research funding agency. So they are worthless in their current form.
With this, I agree 100%.
I have already mentioned that Hirsch has held absurd conspiracy theories about politics. He also strongly believes that BCS theory is incorrect, and he has his own pet theory of superconductivity to replace it. BCS theory is one of the most successful theories in all of physics. It correctly predicts many properties of many materials, and is based on sound fundamental principles. It is BCS theory that predicts that hydrogen-rich compounds will superconduct at high pressures and temperatures. Thus, Hirsch has an ulterior motive for proving this paper wrong: it furthers his personal academic agenda. This is therefore a case where the heuristics point both ways. Hirsch cannot be trusted, but neither can the authors of the Nature paper. One would hope that Nature or the NSF would step in and force an independent review of the data, as was done for the recent Majorana retraction, but so far this has not happened.
Ultimately, we need to wait for some other group to reproduce the results. In 2021, a preprint was released by an independent group arguing for clear evidence of the Meissner effect in a similar hydrogen-rich system. Hirsch has already published a response arguing that the results are misinterpreted.
Time will hopefully tell who is right. I suspect (70% probability-ish) that Hirsch is correct.
All content for this site is CC-BY-SA.