Another ‘Majorana’ particle paper retracted, this time from Science

Nearly a year after marking a paper on the elusive “Majorana” particle with an expression of concern, and almost three years after publishing a critique of its reproducibility, Science has retracted the article due to “serious irregularities and discrepancies” in the data. 

A few papers about Majorana particles, which would be useful in quantum computing if scientists could indeed produce and detect them, have been retracted, flagged with expressions of concern, or otherwise proven difficult to reproduce. 

The latest article to be retracted, “Chiral Majorana fermion modes in a quantum anomalous Hall insula- tor–superconductor structure,” has been cited more than 400 times since it was published in 2017, according to Clarivate’s Web of Science. About 10 percent of those citations have come since Science’s editors published their expression of concern last December. 

The retraction notice states: 

On 21 July 2017, Science published the Report “Chiral Majorana fermion modes in a quantum anomalous Hall insulator–superconductor structure” by Q. L. He et al. (1). Readers who failed to reproduce the findings requested raw data files from the authors, which they provided. Subsequently, the provenance of the raw data came into question; additionally, an analysis of the raw and published data revealed serious irregularities and discrepancies. These issues have caused the editors at Science to lose all confidence in the conclusions of the paper, and we are therefore proceeding with an Editorial Retraction. Authors A. L. Stern, J. Wang, and B. Lian agree with this decision. Authors Q. L. He, L. Pan, X. Che, G. Yin, E. S. Choi, K. Murata, X. Kou, Z. Chen, T. Nie, Q. Shao, Y. Fan, K. Liu, J. Xia, and K. L. Wang disagree with this decision. Authors E. C. Burks and Q. Zhou did not respond. Author S.C. Zhang is deceased.

This Retraction replaces the Editorial Expression of Concern posted on 16 December 2021 (2).

We reached out to Holden Thorp, editor-in-chief of Science, with some questions to flesh out the retraction notice, and received this response from a spokesperson: 

Science Editorial reached their decision after consulting with a number of experts, which they did under agreement of confidentiality. After careful and exhaustive deliberation, the editors concluded that there was no scientifically plausible explanation for the irregularities in the files the authors originally provided when asked for raw data.

None of the corresponding authors of the paper – including last author Kang L. Wang of UCLA and first author Qing Lin He, formerly a postdoc at UCLA and now an assistant professor at Peking University in China – or any of the authors who agreed with the retraction decision, responded to our requests for comment. 

Questions about the retracted article became public in January of 2020, when Science published a paper by a team of scientists from Pennsylvania State University and the University of Wurzburg that found its claims unreproducible. 

We asked a few of the authors if they could fill in the gaps in the retraction notice. Laurens Molenkamp, one of the scientists from the University of Wurzburg, told us: 

​​Yes, you guessed right, we were involved in the retraction process. [Three authors from Penn State, Cui-zu Chang, Nitin Samarth, and Moses Chan, told us they were not involved in the retraction.]

You should realize that the experiment in the paper, which was suggested by theorists, looked very improbable to succeed to any experimentalist in the field, like us. Reading the actual paper directly led to us having serious scientific questions and we contacted the authors about the data. After a while, authors did supply us with a data file, but analyzing this data file made us even more worried about the figures in the publication. Since some authors did not want to involve [us] in further discussions, we warned Science about the inconsistencies between the data file and the actual figures, as well as between the data authors chose to present and the remaining data available in the file.

Eventually this led to Science suggesting us to write the paper where we show that the original results cannot be reproduced.   

What then happened is the retractions of other Majorana papers (rather different experiments by other groups on localized Majorana states, not the ‘flying’ ones in the retracted Science paper). Since the scientific sloppiness in those papers actually was less severe than what we found in the data we had available, we went back to Science. This then led to the Expression of Concern and now to the retraction.

Two experts in the field who have found problems in other Majorana papers, Sergey Frolov of the University of Pittsburgh and Vincent Mourik of Forschungszentrum Jülich in Germany, in an email to us described the issue with the retracted paper more bluntly: 

These data were made up. People who found this did an amazing job. They found so many problems and artefacts, it is absolutely brilliant. 

For the papers that Frolov and Mourik found problems with, they said: 

the key issue was extreme data selection. Whether the data are made up, like in this Chiral Majorana paper, or the conclusions are made up but illustrated by cherrypicked data makes no difference. Cherrypicking to convince the editor/referee/reader of a false conclusion is just as bad, and can be harder to prove. We hope that Science editors understand this.

Frolov and Mourik also said that they “applaud the editors of Science” for retracting the paper: 

Editors are afraid to take this step, but they really should not be. The law is on their side, especially in situations like this, where it is painfully obvious the paper is invalid.

At Science they appear to be re-imagining their approach to the situations where they cannot get the university to run a proper investigation. See recent editorial by the Editor-in-Chief, which is a step in the right direction, and all journals should pay attention to this. [RW: We have some questions about how the model in the editorial would work]. Science also published, on two occasions, negative reproduction studies of Majorana experiments, including an experiment where a group of researchers failed to reproduce the conclusions of this now retracted paper. It deserves separate praise that Science does this. In contrast, Nature journals offer publication of reproduction studies in lesser-impact journals within their vast family. This discourages reproduction work and reduces its impact.

Research into Majorana particles may be troubled, they said, because of its complexity: 

Majorana physics is a complicated and at the same time fascinating topic in modern science, which requires cross-cutting expertise. It is therefore especially susceptible to unreliable claims, because not many experts can fully evaluate the results. Sadly, we see a lot of attrition in terms of claims not holding, and it is not only bad science but also unreliable research. The topic of Majorana is not unique in suffering through this dynamics, and some fields mature past it, so we are hoping this is what will happen here.

Molenkamp shared Frolov and Mourik’s concerns about the field: 

And yes, I am worried about the impact these retractions may have in my research field. I have worked on topological physics for a long time now, basically from the beginning, and it is not nice to see the reputation of the field damaged by the type of grave issues displayed by the now retracted papers.

That being I said, I do believe topological superconductivity and Majorana states will be harnessed, just not with the approaches followed by the now retracted papers.

Like Retraction Watch? You can make a tax-deductible contribution to support our work, follow us on Twitter, like us on Facebook, add us to your RSS reader, or subscribe to our daily digest. If you find a retraction that’s not in our database, you can let us know here. For comments or feedback, email us at [email protected].