Chemistry matters. Join us to get the news you need.

If you have an ACS member number, please enter it here so we can link this account to your membership. (optional)

ACS values your privacy. By submitting your information, you are gaining access to C&EN and subscribing to our weekly newsletter. We use the information you provide to make your reading experience better, and we will never sell your data to third party members.



Chemists retract 15-year-old paper and publish a revised version

After discovering they misidentified the molecule central to the study, researchers carry out a structural do over

by Stephen K. Ritter
September 7, 2017 | APPEARED IN VOLUME 95, ISSUE 36

In what is being called an honest mistake, a research group has reported that the structure of the headline compound from a paper it published in 2002 is incorrect. To set the record straight, the team has retracted the paper and at the same time published a new paper reporting the actual structure.

This strategy offers chemists a palatable way to correct the literature when a pervasive error is found and the error is not the result of any ethical misconduct. Other journals, such as the Journal of the American Medical Association, have handled similar situations by retracting and replacing papers, making the original versions available in supporting information sections with errors highlighted.

The original paper published by medicinal inorganic chemist Chris Orvig and his group at the University of British Columbia reported the synthesis of a new tetraazamacrocycle ligand (Inorg. Chem. 2002, DOI: 10.1021/ic010716a). The researchers made the compound to see how replacing two methylene groups in a macrocyclic ring with phosphinate groups affected the stability of complexes formed with lanthanide ions. Such complexes are of interest as contrast agents for magnetic resonance imaging.

Orvig says all the characterization data were consistent with the ligand having a dimeric structure. But the team had a hard time growing crystals of the ligand and couldn’t confirm its structure by X-ray analysis. When graduate student David M. Weekes was recently revisiting the ligand for a new project, he was able to get a crystal structure and found the data didn’t add up, literally. Weekes discovered the molecule is half the size expected—it is monomeric, not dimeric—and forms a 2:1 complex with lanthanides instead of a 1:1 complex as originally thought.

An incorrect structure might normally be handled by issuing a correction statement appended to the paper pointing out the error. But in this case, Orvig says, the incorrect assumption of the ligand structure nullified the conclusions. Orvig and the journal’s editors agreed the best solution would be to retract the paper (Inorg. Chem. 2017, DOI: 10.1021/acs.inorgchem.7b01932). But they also agreed that Orvig’s group could publish a new paper along with the retraction reporting the correct structure (Inorg. Chem. 2017, DOI: 10.1021/acs.inorgchem.7b01117). The event was first reported by Retraction Watch, a blog that aims to increase the visibility of the retraction process for scientific papers.

Orvig says it was important and the right thing to do to correct the literature, even if a little embarrassing, given that some researchers might have already tried to make the ligand and failed, and others might continue to do so. “This is the way science is supposed to work—correcting itself,” Orvig says.

Retracting a paper oftentimes is associated with misconduct—fabrication, plagiarism, or rigged peer review. Some researchers have suggested that papers like Orvig’s should be given a status different than retraction: For example, “withdrawal” could be used for an honest mistake, and “retraction” could be reserved for misconduct.

Inorganic Chemistry editor-in-chief William B. Tolman of the University of Minnesota applauds Orvig for coming forward to self-correct the mistake, when others might not have done so. “This is precisely why the retraction mechanism exists—to correct the published record,” Tolman says. “Orvig realized the error, did a bunch of work to reexamine it, and then pushed to have it corrected in the literature despite it being years later.” Tolman adds that he would hope researchers view this case as one of many data points in the larger story of how science corrects itself.



This article has been sent to the following recipient:

Richard (September 11, 2017 7:39 PM)
“This is the way science is supposed to work—correcting itself,” I completely disagree. This is how publishing is supposed to work, giving the chance to verify results from other researchers. Science is supposed to develop our understanding of the world and it does so by publishing results that are accurate. I do not wish to criticize Orvig, but the scientific community. There are just a few journals which actually reproduce an experiment in some of their editors labs before accepting the manuscript. And this is exactly what should have been done here, 15 years earlier. Experiments have to be reproducible. And submissions have to be specific, with an Editorial Board that takes its responsibility for accurate science seriously and has the equipment or contacts to do so. Especially with such LMW compounds.
Robert Buntrock (September 12, 2017 4:58 PM)
In reply to Richard, a noble proposition but to invoke the old fable, "Who's going to bell the cat?". Reproducing experiments, on commission, is a volunteer function. Same applies to refereeing. The only publication I know of that uses checkers to rerun the preps submitted is Organic Synthesis. The checkers must support the efforts, both for materials and labor, on their own grants. For doing that one gets a preview of the prep and a "publication". Do you do this work of reproducing?
A. Chandrasekaran (September 12, 2017 10:17 PM)
How about any mass spectrum? That would have been much faster than growing a crystal; it would take 15 minutes, not 15 years! In case of such symmetric molecules, MS or crystal structure must be considered as a requirement (at least in the future).
Steve Ritter (September 12, 2017 11:34 PM)
The researchers had mass spec, elemental analysis, and other data that suggested the dimeric structure they were expecting was correct. In the new paper, the crystal structure and mass spec helped determine the monomeric structure. The new paper provides the details on how the researchers misidentified the structure the first time.
Yahia Hamada (September 13, 2017 2:23 PM)
Why researchers still correlate the X-ray diffraction pictures of a single crystal in the solid-state of complexes to the species that were supposed and presumed to be found in solution?

Chris Orvig is a very honest in his reporting. That is how honest people report their new findings if they see that it differs from what they have reported before. I do have some examples that i will be happy to share with C&EN-News.

Prof. Hamada
Steve Ritter (September 14, 2017 12:12 PM)
Please do by email at C&EN is interested in following up with a broader look at corrections and retractions.
A. Buck (September 13, 2017 2:32 PM)
I commend the researchers for correcting the error. It can be embarrassing, but being so forthcoming elevates this to and admirable act. I would point out, however, that there are more physical organic tests that can be done on guest/host interactions or ligand association ratios, such as this, to identify the ligand to ion ratio. The main test I think of is a Job's Plot. I would encourage anyone doing these types of tests to consider a Job's Plot as a 2:1 ligand to La-ion ratio would have been another cue to reconcile in the original manuscript.
Benjamin Davis (September 13, 2017 4:09 PM)
qNMR could also rapidly discern between mono, di, and tetrameric species. Pick a standard, make up a 10 mM sol, do a T1 experiment and you're nearly done.
Benjamin Davis (September 13, 2017 4:33 PM)
sorry, retract my comment! mental error on my part!
Steve (September 13, 2017 5:06 PM)
I wonder how a monomeric ligand would exhibit a dimeric ligand MASS peak?
A. Chandrasekaran (September 13, 2017 10:00 PM)
It is somewhat common to see a (2M+H) peak in MS, but usually much weaker than parent peak. But, if we are specifically looking for this dimer as parent peak, then we may consider the monomer peak as a breakdown product. Surely the MS would have raised doubts that warranted confirmation by some other method, but may be got overlooked by all.
A. Chandrasekaran (September 13, 2017 10:36 PM)
Just now I had the chance to look at the MS data (I did not think it will be freely available!). The strongest peak at mass 265.1 had been interpreted as (M-2H)2- of dimer; this same peak could also be interpreted as (M-H)- of monomer. Di-anion coming as major peak, with the "parent" mono-anion coming only at 8%, could be the red flag. I think mono-anion peak will always be much stronger than the di-anion (that was the case in ESI-MS of our di-phosphates).
Wim (September 14, 2017 1:20 AM)
If a (M-2)2- dianion is assumed, than one should see C13 isotopes at 1/2 mass units apart at (M-2)/2 , not 1 mass unit. This is a simple and fast way to distinguish between monomers and dimers, just look for the C13 isotope of the assumed product.
A. Chandrasekaran (September 15, 2017 12:41 AM)
Sounds great! Thanks for that tip.
Ron Kluger (September 13, 2017 7:21 PM)
If the error led others astray or upset a principle of chemistry, then it would have been short-lived. If others found the compound useful for a purpose that would require the dimeric structure, they would have discovered the error. Once it's out there, the world of chemistry becomes the fact-checker. If no one caught it, how did that happen? There is little value in being the person who corrects someone else's mistake but if we do we should appreciate that we could be making the next mistake ourselves. Be sympathetic, even if you wasted your time and resources. And if you correct your own error, tell C&E News about it.
Edgar Mueller (September 14, 2017 10:29 AM)
Please allow me a little remark on this story.
The ligand in question can be formed in a "Mannich-type" reaction between
formaldehyde, a secondary amine, and hypophosphorous acid (H3PO2).
Such reactions are reversible in the presence of water and Lewis-acid
catalysts or H+.
It is thus possible that there is an equilibrium between the monomer and the
dimer, and that the monomer crystallized more easily out of the equilibrium
mixture. But this is not sufficient evidence for affirming that all of its complexes
must contain it in the monomeric form, too. I dare suppose that adding Ni(2+) to
the monomeric ligand will push it into the dimeric form, because it prefers nitrogen
ligands over phosphinate and carboxylate oxygen. For other metal ions, such
as the lanthanide ions, there might be equilibria between both forms, as they like more carboxylate and phosphinate oxygen ligands than nitrogen ligands.
A. Chandrasekaran (September 15, 2017 1:03 AM)
We have observed such disparity between solid and solution states; the solid had only one structure (based on X-ray and solid state 31P NMR that used over 100 mg sample) whereas the solution had equilibrium with tri- and hexa- valent phosphorus. There also the ligand was a Mannich base (phenol derived).
If it was a static/dynamic mixture, the NMR/VT-NMR would have shown some evidence. Unlike crystallography, which uses micrograms sample, NMR, especially 31P NMR uses several milligrams. Normally it will represent the sample better. About 10 mg seems to have been used in the original NMR study. 1H and 31P NMR data seems normal.
Paul (October 2, 2017 10:41 PM)
RE:Oct.2,2017 issue Reactions section.
The published comments/letters from Richard and Robert Buntrock,although somewhat thoughtful,contain terribly awful syntax effectively reducing their impact. In fact, it is an embarrassment to any educated individual. Maybe you should contact them before publication with suggested edifications. After all, this is supposed to be a professional trade journal.
Steve Ritter (October 5, 2017 12:48 PM)
After reading this story, Sibrina Collins, an inorganic chemist and head of the STEM Center at Lawrence Technological University (, developed a “learning objective” for students to explore the chemistry presented in the original paper and the new paper, discuss the kinds of analytical data that are important for chemists to ensure their conclusions are accurate, and to better understand the publishing and peer review process. The objective is a free resource--anyone can create an account and access the material at:

Leave A Comment

*Required to comment