Science has retracted a paper it published in July by a group of authors in China over concerns about two images in the article — problems the researchers have attributed to chaos in their group due to the COVID-19 pandemic.
The paper, “Proton transport enabled by a field-induced metallic state in a semiconductor heterostructure,” was a collaboration by researchers from various institutions in the country, as well as one in the United Kingdom. The senior author was H.B. Song, of the University of Geosciences in Wuhan.
Shortly after publication, Fang Zhouz, a science sleuth in China, posted concerns about the article on his blog — which drew the attention of Elisabeth Bik. Bik eventually amplified those concerns on PubPeer, where she noted that:
Figure S10 of this paper (left) appears to look very similar to Figure 8b of a previous paper with some shared authors (right), i.e. Rong Xu et al. IJ Hydrog Energy (2017), doi: 10.1016/j.ijhydene.2017.05.218.
Red boxes show the areas that look similar between the two figures. Yellow markings are part of the original figures.
The two photos appear to be showing a different composite.
Could the authors please clarify if these two images are supposed to show the same structure? Maybe I am misinterpreting the labels.
Turns out, she wasn’t, and they couldn’t.
In early September, Science issued an expression of concern for the paper and said it was investigating the issue. Now, barely a month later, the journal has retracted the work:
The supplementary material (SM) for our Report “Proton transport enabled by a field-induced metallic state in a semiconductor heterostructure” (1) contained mistakes. Fig. S10 was incorrect. The image had been previously saved in an experimental database. The figure should have shown a schematic illustration. This mistake occurred when computer data was transferred from one student to another. Fig. S9B also raised questions. The image appears to be a copy of figure 8B in a 2018 paper in the International Journal of Hydrogen Energy (2). The tests producing the data for these two figures were performed during the same period of time. Because of the similarity between the IV-IP measurements of SN-120 and the NaxCoO2/CeO2 (x = 0.55) measurements, these data were wrongly saved in the same folder. Therefore, the final plotted fig. S9B in Science erroneously contains data from the figure in the 2018 paper. We repeated the experiment to replicate the data and found that the new outputs of two NaxCoO2/CeO2 (x < 0.6) cells are above 550 mW cm–2 at 520°C, consistent with the data in fig. S9.
These mistakes were related to recent location and personnel changes. Between June and September 2019, all laboratories of the Faculty of Materials and Chemistry, including ours, had to move to a new campus 30 km away. During the same time, author L. Liu graduated, but the ordinary data transfer to successors was temporarily discontinued. The manuscript was finalized between mid-December 2019 and March 2020, during the unprecedented lockdown of Wuhan city for coronavirus disease 2019 (COVID-19). All authors who prepared the main manuscript and SM data were isolated in different cities and countries, without access to lab computers or database resources, which led to miscommunications.
We take full responsibility for the accidental mistakes in the SM. We stand by our experimental design, theoretical calculations, main data, and analysis of low-temperature proton ceramic fuel cells. However, given the mistakes in the SM, we have decided to retract the Report. All authors except for B. Zhu, M. Huang, M. Akbar, and J. S. Kim approved this Retraction.
In an email, Zhu told us that the four authors who dispute the retraction believe their research is “self-explanatory”:
the findings addressed in the Science article have been proved their reproducibility by independent research groups (further details available upon request), and this is one of the reasons why some other authors disagree to the retraction; 2) Our entire article has indeed genuine and not influence to mis-lead readers of the article. The supplementary materials were included to help the readers’ understanding only. I would like to suggest to read this article without referring to the relevant SI materials to anyone as the article itself is a self-explanatory.
Also, Zhu noted, Science sometimes corrects, rather than retracts, articles for problems — so why not this one?
we found a very similar case from Peking university’s paper on Science which has been published with corrigendum, https://science.sciencemag.org/content/369/6508/eabe5323 from the first published article：https://science.sciencemag.org/content/354/6316/1170 . This told us the retraction is not only the way to our Science article.
Bik praised Science for its “great response” in the case, starting with the rapid publication of the expression of concern:
this was a great response from the journal, a fast way of flagging the paper while still giving the authors a chance to reply and set the record straight.
Now that the paper has also been retracted, I am even more impressed with the way the journal handled this case. There were only 57 days between the PubPeer post on August 12 and this retraction. That is a very rapid handling of this case, and a good outcome. Too many papers with severe concerns raised by other scientists are still out there and too many journals and publishers are unwilling to take action. That could lead to other scientists unsuccessfully trying to reproduce experiments that contain errors or that never happened. Journals that act rapidly and flag, correct, or retract such papers show they care about science. I wish every case on PubPeer would be handled with such speed and vigor.
Bik added that the authors’ excuse, while plausible, casts doubt on the validity not just of the figures but the entire study:
They explain that there had been miscommunications due to different campus locations, moves, and the coronavirus lockdown. Although understandable, such miscommunications cast doubts about the trustworthiness of other figures and graphs too. So the decision to retract is a good one, and I applaud the authors for doing the right thing. I hope these errors will encourage them to keep better track of lab notes and filenames, and that the journal will consider allowing the authors to resubmit their revised work in the future.
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].