Research integrity

Bad Choices in Dresden

"There is no reason for an investigation into scientific misconduct and therefore it will not take place."

Angewandte Chemie (translated Applied Chemistry) is an elite journal in the field, issued by the German Chemical Society. If you publish there, your academic career is made.

A Chinese scientist, trained in Germany at the Institute of Polymer Research Dresden of the prestigious Leibniz Society, published in Angewandte Chemie. And then the paper was right away withdrawn because of data manipulation. PubPeer sleuths found more. Will the German institute ever open an investigation?

The Chinese scientist is Xiaoling Liu, now associate professor at the College of Polymer Science and Engineering of the Sichuan University in China, and Secretary of the Blood Purification Branch of the Chinese Society of Biomaterials. Her research is described to be on “design and application of medial lung materials for extracorporeal membrane oxygenation (ECMO)“. Basically if Dr Liu’s biomaterials are faulty, due to, say, research misconduct, people will die.

This was Dr Liu’s retracted paper:

Xiaoling Liu, Dongxu Zhou, Yunbo Feng, Jing Gou, Chenxi Li, Chao He, Weifeng Zhao, Shudong Sun, Changsheng Zhao, Dietmar Appelhans , Brigitte Voit Quantitative Synthesis of Temperature‐responsive Polymersomes by Multiblock Polymerization Angewandte Chemie (International ed. in English) (2019) doi: 10.1002/anie.201910138

A problem was found in August 2019:

On Fig 4 two polymersomes look exactly the same pixel-wise.”


Others confirmed the concerns about a duplicated nanoparticle, but Liu disagreed and replied on PubPeer, with a picture:

First, It is possible to obtain the same shape of polymersomes for the cryo-TEM image, which is caused by the shape of the polymersome itself, the water environment inside and outside the capsule and the preparation process of sample. In addition, when we zoom in on these two polymersomes (as shown below), there are still many differences, such as color depth.”

In July 2021, the paper was retracted, or rather withdrawn, with a notice:

“This Communication, published online on 14 August 2019 in Wiley Online Library (onlinelibrary.wiley.com), has been withdrawn by agreement between the corresponding authors, the journal’s Executive Committee, and Wiley-VCH GmbH, Weinheim. The withdrawal has been agreed upon following the explanation by the authors that problems with a microscopy image presented in Figure 4 of the article cannot be fully clarified.”

What the notice forgot to mention, was another problem present in that Figure 4:

Orchestes Quercus: “the PS150 microscopy image is presented in ‘Functional Cellular Mimics for the Spatiotemporal Control of Multiple Enzymatic Cascade Reactions’ / https://doi.org/10.1002/anie.201708826 as something else: ‘ Cryo-TEM images of Ada-Psome photo-crosslinked for 20 min at pH 8’. ).

The original paper is gone completely, you can thank Sci-Hub for storing a copy. So I contacted the last author Brigitte Voit, scientific director of the Leibniz Institute of Polymer Research (IPF) and professor at the Technische Universität (TU) Dresden. We communicated in German, so I present Voit’s replies in translation. She confirmed that her lab “was involved in this publication” and explained:

There is a difference between withdrawal and retraction. The article was still at the Accepted Article stage, i.e. the article had not yet been edited and the version of record had not yet been published. In this case, according to the publisher’s policy, there is a withdrawal and the article is indeed deleted. A retraction is carried out when the version of record is published. In this case, the article remains online and is provided with a retraction note.”

Liu did her PhD at IPF in Dresden under Voit in 2017, stayed for some months as postdoc and in 2018 returned to the same Sichuan University where she studied for MSc, to become professor. Her PhD thesis was titled “Polymeric Multicompartmentalized Systems Mimicking Artificial Cells for Controllable Multiple Enzymatic Cascade Reactions” and was supervised by Voit, together a certain TU Dresden professor Xinliang Feng, who is however never mentioned in the thesis detailed acknowledgements, unlike Liu’s “co-supervisor” Dietmar Appelhans and a lot of other colleagues at the Institute of Polymer Research in Dresden.

I also pointed out to Voit a possible problem with another paper from her lab:

Xiaoling Liu, Xueyi Wang, Brigitte Voit , Dietmar Appelhans Control of Nanoparticle Release by Membrane Composition for Dual‐Responsive Nanocapsules Chemistry – A European Journal (2019) doi: 10.1002/chem.201903459 

Acanthobothrium Oceanharvestae: “Figure S3 of this paper appears to overlay with Figure 3a of her previous paper (https://doi.org/10.1002/advs.201600308).” “the same issue also appears in figure 3b, which looks quite similar to Figure 7a (https://doi.org/10.1021/acs.macromol.7b02347)

All papers feature Liu, Voit, Apperhans, plus one extra co-author with a Chinese name, a different one in each case. Voit explained to me that the data reuse was intentional:

The paper shows a comparison of 2 systems, the previously published and a new one. Unavoidably and completely correctly, a reference is made to the results of the previous manuscript or the results are reproduced (which must be then very similar) and shown in the SI [Supplemental Information, -LS]. Criticism should actually consider the context of the manuscript.”

There are different schools of thought on data reuse. All three papers are after all presented as original research, and the figures do not mention that the presented data was previously published as original research elsewhere. So one school of thought says any data reuse must be avoided, for example by repeating the experiment (also to ensure reproducibility), and if for some reason a reuse is unavoidable, it must be clearly acknowledged and justified in the paper. Another school of thought thinks all this is for losers, and the career trajectories of this school of thought representatives kind of proves them right.

Leibniz Institute of Polymer Research Dresden, old building (Source)

The withdrawn Angewandte Chemie paper used to list only the Sichuan affiliation for Liu. Maybe the problematic research was done there? I asked Liu’s mentor Voit, who explained:

The work was submitted a long time after Ms. Liu found her new place of work in China, and some experiments were also completed there.
As I said: The manuscript was withdrawn in agreement with all authors, as the problems with Fig. 4 could not be fully clarified. There is simply nothing more to say
.”

Also Figure 4, with its photoshopped cryo-TEM images of nanoparticles, was it made in China then? The withdrawn Angewandte Chemie paper has no declarations on authors’ contributions, but I found this in the above discussed study Liu et al Chemistry 2019: “We thank […] Dr. P. Formanek for cryo-TEM investigations.“, this is also the IFM expert whom Liu’s PhD thesis acknowledges.

Which suggests, regardless of where and when the picture in Figure 4 was falsified, that it was most likely originally taken in Dresden, in Voit’s institute. The guidelines of the German Research Council (DFG), which apply to all German research institutions, mandate a storage of raw data for at least 10 years after publication. Also the Leibniz Society, to which the Institute of Polymer Research belongs, mandates 10 years for raw data storage. In this case therefore, until 2029. Voit, whose task as scientistic director of her Leibniz institute is to implement the DFG and Leibniz guidelines, remained silent on this topic ever since.

Then, a more recent study was criticised for data reuse, incidentally from the now withdrawn Angewandte Chemie paper.

Xiaoling Liu, Yunbo Feng , Lunqiang Jin , Xueyi Wang , Xiang Zhang , Yi Xie , Changsheng Zhao , Dietmar Appelhans, Brigitte Voit Rapid synthesis of PEGylated multiblock polymers by sequence-controlled polymerization in H2O Polymer Chemistry (2020) doi: 10.1039/c9py01202h 

Elisabeth Bik: “A reader pointed out that Figure 7C of this paper (right) looks remarkably similar to Figure 3B of Xiaoling Liu et al., Angew. Chem. Int. Ed (2019), DOI: 10.1002/anie.201910138 (RETRACTED) – https://pubpeer.com/publications/081CB94A0F970314B7DC59D3188E6F (right). The compound appear to be different.

Liu protested on PubPeer that the graphs merely look similar:

The specific data involved in the two graphs are different. And the different data can be identified by the scale lines on the vertical axis of the graphs, which are also reflected in the tables in the Supporting Information (Table S5 and Table S8 respectively).

You see, the Y-axis is different, which somehow proves the graphs cannot be same even if they clearly do look identical (except for two data points). Voit’s only comment was:

the statement was clearly not agreed with me“.

Well, the data in the highlighted figure was apparently obtained in Dresden also, as Liu et al 2020 mentions: “The authors would like to thank Dr. H. Komber for samples of NMR“. Voit could easily check the raw data.

Now back in China, Liu continues producing excellent research. Like this, spotted by Cheshire aka Actinopolyspora Biskrensis:

Haifeng Ji , Hao Xu , Lunqiang Jin , Xin Song , Chao He , Xiaoling Liu, Lian Xiong , Weifeng Zhao, Changsheng Zhao Surface engineering of low-fouling and hemocompatible polyethersulfone membranes via in-situ ring-opening reaction Journal of Membrane Science (2019) doi: 10.1016/j.memsci.2019.03.082

Figures 3 and 6 appear to have an image in common (red boxes), although it seems to be described differently.

This one is the best:

Lunqiang Jin, Zhenqiang Shi , Xiang Zhang , Xiaoling Liu, Huiling Li , Jingxia Wang , Feng Liang , Weifeng Zhao , Changsheng Zhao Intelligent antibacterial surface based on ionic liquid molecular brushes for bacterial killing and release Journal of Materials Chemistry B (2019) doi: 10.1039/c9tb01199d 

The last author of both these above papers is Liu’s Sichuan colleague and another biomaterials researcher Changsheng Zhao, who also happens to be co-author on the withdrawn Angewandte Chemie paper and on the criticised Liu et al 2020 of Voit’s. As reminder, these people work on medical applications, research fraud will swiftly translate from bench to bedside into dead patients.

Voit however early on made clear to me her position on her PhD mentee Liu:

She did an excellent job. There is no reason for an investigation into scientific misconduct and therefore it will not take place.

As reminder, Voit is the scientific director of her Leibniz institute, and thus theoretically the one in charge of opening or not opening such investigations. But not just this. If I were, hypothetically, to write to the Leibniz society to complain that the scientific director of IPF, Prof. Dr. Voit, decided regarding her own papers that “There is no reason for an investigation into scientific misconduct and therefore it will not take place” I would have to address my complaint to one of the so-called “central ombudspersons on the Leibniz Ombuds Committee“, whose tasks are defined:

“The central Ombuds Committee investigates accusations of scientific misconduct levelled against current and former employees of member institutions and, where necessary, sets up a committee of enquiry in accordance with the relevant guidelines. The central ombudspersons also advise the member institutions and their ombudspersons on all issues to safeguard scientific integrity.”

Can you guess who the central ombudsperson of the Leibniz Society is?

Prof Dr Prof. Brigitte Voit (IPF, Section D)“.

Which ends the story. Anyway, maybe Professor Voit needs to reconsider her choice of collaborators.

Bibekananda De , Brigitte Voit , Niranjan Karak Transparent luminescent hyperbranched epoxy/carbon oxide dot nanocomposites with outstanding toughness and ductility ACS Applied Materials & Interfaces (2013) doi: 10.1021/am402415g 

Sicarius Tropicus: “Is it me or does there appear to “squares” of a different TEM pattern on top of the background image in this figure? Can the authors help me out on this? The scalebar is from elsewhere as well.
“Figure 2. TEM images of (a) ECD0.5 [overview picture and individual integrated carbon oxide nanodot with internal structure (inset)]
Cheshire: “Returning to Figure 2a after 2 years.

Voit’s brief reply was:

All experiments and Analyses were done in the group of Prof. Karak

Professor Niranjan Karak of Tezpur University in India is indeed a special case, with 22 papers on PubPeer. I would even suspect his moustache to be fake and photoshopped. Some warning examples, luckily without any coauthors from Germany:

Problem is, if Voit doesn’t use her high standing to retract at least that one fraudulent Karak paper she co-authored, nobody will. Certainly not the Editor-in-Chief of that ACS journal and University of Texas San Antonio professor, Kirk Schanze, whose ACS Applied Materials & Interfaces specialises on covering up fraud by the most toxic fraudsters from India and China by issuing “conclusions not affected” corrigenda equivalent to science sabotage.

But Voit made clear she wishes to have nothing to do with that Karak paper she is co-author of. I forwarded the concerns to the institute’s Ombudsman, Klaus Werner Stöckelhuber. No reply so far, maybe Dr Stöckelhuber is still conferring with the Leibniz Society’s central ombudsperson? Remember what Prof Voit said regarding her PhD student?

There is no reason for an investigation into scientific misconduct and therefore it will not take place.

To be fair, it may be a culture clash and research misconduct investigations are kind of inadmissible in Dresden. Last time I tried to report a rather serious case to TU Dresden, its Ombudsman turned around and accused me of research misconduct.

Update: The Leibniz Central Ombudsperson Prof Voit replied to my notification of suspected research misconduct:

I duly receive your email and your notification. But since I am obviously biased, as you yourself have noticed, I cannot follow this up myself, and I ask my colleagues to do so. There are internal investigations at the IPF.

Incidentally, there is a reason why the Leibniz Society has a Central Ombudsman Board since 2020. Some years ago, there was a high-profile research misconduct and animal abuse affair featuring the (dethroned) director of the Fritz-Lippman-Institute on Aging in Jena, K Lenhard Rudolph. Read about it here and here.

Update 13.10.2021

I informed the Leibniz Society about further irregularities in the papers from the Voit lab and the PhD thesis of Xiaoling Liu, defended at IPF under Voit’s supervision.

A “Reader” commented:

XiaoLing’s thesis and related articles also show a surprising overlap between zeta potential measurements on differently coated particles, see Fig. 7.9 and Fig. 9.8.

The figures show zeta potential versus layer number whilst coating silica particles with different coatings. Unexpectedly, the two figures share three points (layers 4, 5, and 8 encircled below). It seems very coincidental that identical potential values are obtained for identical layer numbers in unrelated experiments.

The following issue is present in both the PhD thesis and in this paper:

Xiaoling Liu, Petr Formanek, Brigitte Voit, Dietmar Appelhans Functional Cellular Mimics for the Spatiotemporal Control of Multiple Enzymatic Cascade Reactions Angewandte Chemie (2017) doi: 10.1002/ange.201708826

Orchestes Quercus: “Figure 1(b) and Figure 1(c) contain an area of overlap. The panels should show confocal microscopy images of nanoparticles in different states of assembly (coating) as indicated by the insets in each panels. It is my understanding that the two coating steps between (b) and (c) cannot take place on a microscope slide, especially not without displacing the particles.”

This statement just arrived from the head of the Department Integrity at the Leibniz Association (highlight mine):

Dear Mr. Schneider,

The Leibniz Ombudsman Committee received via the Scientific Director of the IPF Dresden your allegations against various publications with the participation of Leibniz scientists, as well as the additional information you supplied directly. We will investigate this named notification of scientific misconduct in accordance with the applicable rules of the Leibniz Association. I would also like to inform you that Prof. Voit has relinquished her position on the Leibniz Ombuds Committee until further notice.


Get For Better Science delivered to your inbox.

One-Time
Monthly

Make a one-time donation

Make a monthly donation

Choose an amount

€5.00
€10.00
€20.00
€5.00
€10.00
€20.00

Or enter a custom amount


Your contribution is appreciated.

Your contribution is appreciated.

DonateDonate monthly

11 comments on “Bad Choices in Dresden

  1. Great reporting Leonid.

    Liked by 2 people

  2. The zeta potential “data re-use” that Acanthobothrium Oceanharvestae noticed is troublesome. The systems are different, one has a 12 layer coating, the other only a 10 layer coating. Similar (the same?) values could be expected for the first 9 layers as these are the same for the two systems. Values are exactly identical, except for layer 7(!). The zeta potential of the 10th terminal layer should be similar to the 12th terminal layer of the other system. It is reported to be the exactly the same.

    In all, the 10 layer coating figure has one datapoint that differs from the 12 layer coating figure. And it it is not for the expected layer.

    Liked by 1 person

  3. XiaoLing’s thesis and related articles also show a surprising overlap between zeta potential measurements on differently coated particles, see Fig. 7.9 and Fig. 9.8.

    The figures show zeta potential versus layer number whilst coating silica particles with different coatings. Unexpectedly, the two figures share three points (layers 4, 5, and 8 encircled below). It seems very coincidental that identical potential values are obtained for identical layer numbers in unrelated experiments.

    Like

  4. (image 2 of the above post)

    Like

  5. https://doi.org/10.1002/ange.201708826 and thesis.

    Figure 1(b) and Figure 1(c) contain an area of overlap. The panels should show confocal microscopy images of nanoparticles in different states of assembly (coating) as indicated by the insets in each panels.

    It is my understanding that the two coating steps between (b) and (c) cannot take place on a microscope slide, especially not without displacing the particles.

    Like

  6. Below a quantification of the ‘zeta potential’ match between unrelated coating experiments as published in the thesis, Angewandte Chemie (https://doi.org/10.1002/ange.201708826, see comment above), and Advanced science news (https://doi.org/10.1002/advs.201600308).

    The thesis’ PDF contained vector graphics allowing one to extract potential values with a very high precision (+-5 uV). Calculating the potential difference between a layer in Advanced science news (rows) and Angewandte chemie (columns) one can see that the matches (difference 10x ‘worse’ than the worst diagonal match (7 uV). Note that layer 0 can/should match, this is the zeta potential of an uncoated silica particle and is presumably measured once.

    This level of coincidence is somewhat troubling and adds to a similar coincidence observed in https://pubpeer.com/publications/6B1A49ED40F53DD7058C7B288D026A#2.

    Like

  7. The author of the thesis seems to have copy-pasted a significant amount of text from Sect. 7.3 to Sect. 9.3, see below. These sections also contain the zeta potential figures. One could surmise that also the zeta potential figure was copied and that this is the source of the unexpected overlap. This figure then went on to the Angewandte Chemie paper.

    The text itself is sometimes an ill fit with the figures. The recurring sentence “The initial zeta potential of the silica surface changed from -45 mV to -2 mV after the adsorption of cationic PAH polymer. This value is still negative, since …” seems to miss that in Sect. 9.3 the corresponding PAH datapoint is slightly positive. And in Sect. 7.3 it sits at -5 mV… The text and figure of Sect. 7.3 went into the Advanced Science news article.

    Sect. 7.3

    To prove the successful formation of multilayers on silica particles, the variation in the zeta potential of silica particles coated with the bilayer PAH/PNMB11 was investigated as a function of the number of layers (Figure 7.9).

    The initial zeta potential of the silica surface changed from -45 mV to -2 mV after the adsorption of cationic PAH polymer. This value is still negative, since the particle surface was not completely covered by PAH polymer. This means that the adsorbed PAH had not enough positive charge to offset the negative charge of bare silica particles. However, the increase of about 40 mV as compared with the bare particles was sufficient to start the fabrication of the desired hollow capsules.
    […] .
    This negative zeta potential obtained enabled us further to carry out subsequent adsorption of the positively charged PAH. Such alternating reversal in the sign of the zeta potential is characteristic of LbL assembly of polyelectrolyte multilayers on particles surface, suggesting a successful stepwise layer growth of PAH and PNMB11.

    Sect. 9.3

    To prove the successful formation of multilayers on silica particles, the variation in the zeta potential of silica particles coated with the PAH/PMA(β-CD)/Ada polymersomes/PMA(βCD)/[PAH/PNMD)]3/PAH/PMA(PEG) multilayers was investigated as a function of the number of layers (Figure 9.8).

    The initial zeta potential of the silica surface changed from -45 mV to -2 mV after the adsorption of cationic PAH polymer. This value is still negative, since the particle surface was not completely covered by PAH polymer. This means that the adsorbed PAH had not enough positive charge to offset the negative charge of bare silica particles. However, the increase of about 40 mV as compared with the bare particles was sufficient to continue the subsequent adsorption.
    […]
    This negative zeta potential obtained enabled us further to carry out subsequent adsorption of the positively charged PAH. Thus, the alternating reversal in the sign of the zeta potential is characteristic of LbL assembly of polyelectrolyte multilayers on particles surface, suggesting a successful stepwise layer growth of PAH/PMA(β-CD)/ Ada-polymersomes/PMA(β-CD)/ [PAH/PNMD)]3/ PAH/PMA(PEG) multilayers.

    Like

  8. The Leibniz Association informs me:
    “In accordance with the applicable procedural requirements of the Leibniz Guideline on Good Scientific Practice, we hereby inform you that as a result of the preliminary review by the ombudsman, the Presidium of the Leibniz Association will propose the establishment of a investigative committee.”

    Like

    • Probably sounds even more stilted in German. Why not, “we’re looking into it, thanks.”

      Like

      • actually, they said that the preliminary analysis determined the allegations to be convincing, that’s why an official investigation will open.
        At the end, someone might lose her PhD degree, and someone else might get a written reprimand which could lead in a voluntary administrative demotion.

        Like

  9. The pathological practice of falsifying data in order to quickly publish high-impact-factor academic papers and gain promotion in academic careers must be severely punished. Otherwise, it will greatly dampen the enthusiasm of hard-working, pragmatic researchers who may not have published well.

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: