Authors retract highly cited Nature quantum dot letter after discovering error

Authors have retracted a highly cited Nature letter that purported to discover a much sought-after, stable light source from quantum dots, after they realized the light was actually coming from another source: the glass the dots were affixed to.

When the paper “Non-blinking semiconductor nanocrystals” was published in 2009, it received some media coverage, such as in Chemistry World. That’s partly because very small sources of “non-blinking” light could have wide-ranging, big-picture applications, author Todd Krauss, a physical chemist at the University of Rochester, told us:

Off the top of my head, a quantum computer. Quantum cryptography is another one. People want a stable light source that obeys quantum physics, instead of classic physics.

The retraction note, published Wednesday, explains how the researchers found out the effect was coming from the glass, not quantum dots:

In this Letter, we reported the unusual non-blinking characteristics of the fluorescence from individual CdZnSe/ZnSe alloyed quantum dots. However, it has recently come to our attention that similar fluorescence behaviour was seen by Celso de Mello Donega, Daniel Vanmaekelbergh and co-workers from a single fluorophore on bare silica glass. In particular, individual fluorescence spots from single molecules were found to be non-blinking, and fluorescence spectra looked similar to what we reported in our Letter.

After learning of the data, the research group spent months repeating their own control experiments, which included glass without the quantum dots, and observed the same fluorescence they believed had come from the quantum dots:

We corroborated their findings by conducting experiments of our own on bare quartz coverslips, and on quartz coverslips coated with polymethyl methacrylate (PMMA). Although these same control experiments were performed by us before publication, this time we clearly observed non-blinking fluorescence from isolated spots on the coverslip. Furthermore, the fluorescence spectra from these spots were in all practical respects identical to what we reported in our Letter.

They were able to find out exactly what was creating the unblinking light:

Subsequent investigations by us have revealed that the surprising origins of the unusual fluorescence come from individual, molecular defects in silica glasses, brightened by the polymer coating. The details of these new findings will be the subject of future publications. After examining the data of de Mello Donega and colleagues, and determining that we were both observing the same phenomena, we concluded that we cannot attribute the fluorescence we observed to CdZnSe/ZnSe quantum dots. In view of these new results, we therefore wish to retract the paper and sincerely apologize for our error. All authors agree with the decision to retract the paper with the exception of X.R., who was unable to be contacted.

Karl Ziemelis, the chief physical sciences editor at Nature, affirmed that it was a indeed a mistake:

[I]t was an honest mistake based on (at that time) unknown properties of the substrate itself – properties that the authors have subsequently discovered and are now working to better understand.

We’ve seen a few cases where unexpected effects have felled a paper — for example, a filter that affected neuronal morphology. This kind of retraction note, which spells out exactly what went wrong, is very useful to anyone working with a similar apparatus so that they don’t make the same mistake. So this retraction gets a “doing the right thing” nod from us.

In a phone conversation, Krauss told us that only glass with certain properties will produce a strong, stable light source. To preserve quantum dots on glass, the researchers affixed a polymer to the glass. In looking for the source of the non-blinking fluorescence, they found that if that polymer is aged and treated in a specific way, it brightens the light from the defects in the glass:

It’s not just any polymer. You have to have some degradation level. You put this polymer that’s aged like a wine in certain ways. If you use polymer that’s fresh, you won’t see anything.

He said a series of specific “unfortunate events” led them to the mix up, including the polymer being prepared in critically different ways during the original experiment, so that the light appeared in the glass with quantum dots but not the control glass:

We really tried to do the best job. The control experiments were done with a polymer was not treated in a certain way. We used a fresh polymer.

Krauss pointed out that they can’t know for certain what was causing the effect in the original experiment, but new experiments looking at the defects in glass appear to show exactly what they reported in the 2009 paper:

The sample is long gone. There’s no way to recover that. Our best guess is what we were looking at before was a defect in the glass. The defect in glass has all the same properties that we saw before. It looked and smelled and quacked like a duck.

The original paper has been cited 416 times, according to Thomson Scientific’s Web of Knowledge, suggesting this mix-up could have implications for many labs. But Yixing Yang, a researcher at NanoPhotonica Inc in Florida who recently cited the now-retracted paper, told us he doesn’t believe the retraction will negatively set back the field:

To the best of my knowledge, this method is actually not widely used in the research field to suppress the blinking. Nowadays, two general approaches are usually used towards suppressing the blinking. The first method is to change the solution environment of nanocrystals by adding antiblinking agents, and the second one is to grow a thick inorganic shell on the nanocrystals. It was also reported a highly crystallized thin shell can achieve the same goal. So the retraction of this paper will obviously have impact on the related research (since it was published in high-impact Nature and well cited), however, will not affect the mainstream of current research of generating high-quality non-blinking nanocrystals.

The retraction hasn’t even stymied Krauss’s own research — after all, they found a source of non-blinking lights:

The premise that excited the editors is there — it’s just not a quantum dot.

Krauss, along with the the researchers who first spotted the non-blinking light in glass without quantum dots, have submitted a new paper for publication, “Non-blinking single-photon emitters in silica,” already available on arXiv. The abstract mentions the origin of the discovery:

The defects are single-photon emitters, do not blink, and have photoluminescence lifetimes of a few nanoseconds. Photoluminescence from such defects may previously have been misinterpreted as originating from single nanocrystal quantum dots.

Update: Friday October 30th 2015, 8:05 PM

Han Htoon, a scientist at Los Alamos National Laboratory, who cited the paper in a 2011 Nature letter, emailed us some more insight on the retraction. His paper was not affected by the retraction — it was just a “general reference.” However, he says the fact that Krauss’s paper didn’t pan out does matter in the larger scheme of things:

That said, for other reasons, the retraction is a big deal. Up to this point, as reviewed by my colleague in a Chemistry of Materials article on ‘Heterostructuring QDs toward suppression of blinking and Auger recombination’ (Hollingsworth, J. A. Chem. Mater. 2013, 25, 1318), only two basic types of QDs had been reported to exhibit non-blinking behavior — the graded interfacial alloy (Krauss et al. 2009) and the thick-shell or “giant” QD (similar results reported in the same year by Hollingsworth et al., J. Am. Chem. Soc. 2008, 130, 5026 and then Dubertret et al. Nat. Mater. 2008, 7, 659 for “quasi-type II” CdSe/CdS Qds; and later, for example, in Nano Lett. 2012 12, 5545 for type II InP/CdS QDs).

Significantly, the latter (thick-shell / giant QD) approach has been reproduced, further studied and used in applications ranging from LEDs to single-molecule tracking in biology by groups around the world. These QDs exhibit different spectral/lifetime characteristics (some aspects of which we explored in our Nature article) compared to those described in Krauss et al., Nature 459, 686-689 (2009).

Like Retraction Watch? Consider making a tax-deductible contribution to support our growth. You can also follow us on Twitter, like us on Facebook, add us to your RSS reader, and sign up on our homepage for an email every time there’s a new post. Click here to review our Comments Policy.