Big Tussle Over Tiny Particles

As truth seekers, scientists often challenge one another’s work and debate over the details. At the first-ever international scientific conference, for instance, leading chemists argued vociferously over how to define a molecule’s formula. A lot of very smart people at the meeting, held in Germany in 1860, insisted that water was OH, while others fought for H₂O.

That squabble might seem tame compared with a dispute that’s been raging in the nanoscience community during the past decade. Francesco Stellacci, a professor at the Swiss Federal Institute of Technology (ETH) Lausanne, says his group created gold nanoparticles coated in somewhat orderly striped domains of two different types of molecules. Because of their unique rippled topology, Stellacci claims, these particles enter cells in a way that might be useful for drug delivery and they trap toxic ions, such as methylmercury, showing promise as environmental sensors.

Stellacci’s critics—principally Raphaël Lévy of the University of Liverpool and Philip Moriarty of the University of Nottingham, both in England—say Stellacci’s data don’t prove the existence of stripes and are inconsistent and riddled with errors.

Unlike the scientific debate over a molecule’s formula, today’s differences of opinion don’t typically get hammered out at conferences. They instead get aired in papers and correspondences published in peer-reviewed journals. And increasingly, they’re being carried out on blogs and online commentary sites such as PubPeer, forums that some scientists say can be unfair.

The striped nanoparticle feud, in particular, has reached a fever pitch because of the online buzz it has generated. Lévy and Moriarty write regular posts on Lévy’s blog (raphazlab.wordpress.com), hashing out in gory detail the various aspects of the controversy. Stellacci, who refuses to engage in these discussions, says he feels as though his critics have gone beyond friendly scientific debate and are trying to discredit him.

Lévy began blogging about the disagreement in 2012, after it took three years to get a correspondence that his group wrote, “Stripy Nanoparticles Revisited,” published in the journal Small (DOI: 10.1002/smll.201001465). In it, he questions Stellacci’s original report on striped nanoparticles: a 2004 scanning tunneling microscopy (STM) paper published in Nature Materials while Stellacci was at Massachusetts Institute of Technology (DOI: 10.1038/nmat1116).

During those three years—a long period by peer review standards—the paper was seen by seven referees, and Stellacci was given time to write a full response, which included new data (Small 2012, DOI: 10.1002/smll.201202322).

“To Small’s credit, they didn’t have to consider the paper,” Lévy says. He originally sent it to Nature Materials because, as home to the original report, it seemed like the logical place to air concerns,but it was rejected. The glacial publication process and the fact that Small had a difficult time finding referees willing to evaluate the controversial topic pushed Lévy over the edge. “That’s the point at which I started blogging,” he says. “It’s important that we have more channels, more critical evaluations of research.”

The 2004 Nature Materials report mainly used STM images as evidence for stripes. Stellacci’s team coated gold nanoparticles with a mixture of short (mercaptopropionic acid) and long (octanethiol) molecules and then imaged the particles sitting on a flat surface. They saw ripples on the particles’ surfaces that they interpreted as 9- to 10-Å-wide, striped domains of each compound.

Stellacci says he didn’t dream up the idea of different molecules separating into cliques when sticking to an interface. “The community has long known that mixed molecules form patches, or islands, on flat surfaces,” he says.

The ETH Lausanne scientist was interested in whether a binary mixture of thiols would arrange themselves on curved surfaces—nanoparticles—in a similar fashion. Seeing stripes rather than islands, though, was a surprise.

When Lévy looked at the 2004 paper, he was surprised too. But it was because he thought the ripples that appeared in the STM images were artifacts caused by improper use of the microscope.

In STM, a sharp tip is scanned across a surface while a set voltage difference is maintained in between. A user doesn’t want the tip to contact the surface but to glide over it, tracing its features. One parameter a researcher sets to maintain a fixed distance between tip and surface is the gain. If the gain is set too low, the tip may not correct its position fast enough when approaching a bump on a surface and crash. If the gain is set too high, the tip corrects so rapidly that it bounces up and down on the surface in an erratic fashion.

Lévy thought the latter was what he was seeing in Stellacci’s images and made that argument, among others, in his 2012 Small correspondence.

In his response, published simultaneously, Stellacci fired back that he and his group had confirmed the stripes over and over again in peer-reviewed papers after 2004 and that Lévy was ignoring those data. “Science is a process, and it would be fair to assume that later papers offer a better understanding of a system,” Stellacci and coworkers wrote.

One of the “later” papers to which Stellacci frequently refers was published in collaboration with Sharon C. Glotzer, a chemical engineer at the University of Michigan, Ann Arbor. Glotzer’s group ran computer simulations of two types of molecules on a nanoparticle’s surface and observed that under certain conditions—when the molecules had a particular length or bulkiness relative to one another—they arranged into stripelike domains (Phys. Rev. Lett. 2007, DOI: 10.1103/physrevlett.99.226106).

According to Glotzer, the molecules separate into stripes because of entropy. Longer or bulkier ligands have more room to wiggle their tails at the edges of domains than in the middle of them, she says. Stripes form because they have more edges.

And having now run many simulations that find stripes under conditions similar to those in Stellacci’s experiments, Glotzer says, “from an Occam’s razor point of view, it would be very surprising to me if the stripes weren’t there.”

Nottingham’s Moriarty, a scanning-probe microscopy expert, is careful to say that simulations carried out by Glotzer and others are high quality. But he suggests that a lot of assumptions get made in running them, so they might not reflect reality. Moriarty is also careful to say that stripes may very well exist; he just doesn’t think Stellacci’s evidence—principally, the microscopy data—is convincing.

Moriarty began writing guest posts on Lévy’s blog after the Small papers appeared online in 2012. He, too, was dismayed by the hard time Lévy had getting his criticism into a journal. Publishing in a journal, Moriarty contends, “should be the start of the debate, not the end of it.”

Once involved, the Nottingham microscopist e-mailed Stellacci to ask for the raw data from his 2012 response paper. Stellacci’s group sent over the requested images right away.

When Moriarty took a look at the data, though, he noticed something he says is a no-no in the world of STM. Typically, an STM user will scan a large area—say 100 by 100 nm—on a surface, he explains. When that user sees a feature of interest, such as a nanoparticle, the researcher will zoom in on it and take a new scan over a smaller area—maybe 10 by 10 nm—to get a better view with higher resolution.

Rather than taking the smaller area scan, Stellacci and his team had been zooming in on the large-area scans with a software program that interpolated between data points and filled in pixels.

Stellacci acknowledged to C&EN that he plotted the data incorrectly. “I wasn’t aware that the software of the microscope was adding points.” But he claims this was only for figures in papers and that the original scans, which have been recently reanalyzed, still show ripples separated by 9 or 10 Å on his particles.

He points to a new set of papers in which stripes on his particles were once again identified, this time by microscopy experts Christoph Renner of the University of Geneva, in Switzerland, Steven De Feyter of the University of Leuven, in Belgium, and Paolo Samorì of the University of Strasbourg, in France (ACS Nano 2013, DOI: 10.1021/nn402414b; Langmuir 2013, DOI: 10.1021/la403546c). These experts’ research groups imaged a set of mixed thiol particles from Stellacci’s lab, analyzed the data, and found stripes.

Fabio Biscarini, a chemist at the University of Modena & Reggio Emilia, in Italy, later surveyed these new images with a method called power spectral density (PSD) analysis, a computational technique meant to remove any interpretation bias. He found periodic features that match the stripes Stellacci has seen.

Lévy and Moriarty say this new set of data is much better quality than previous ones, but they point out that the images look nothing like prior images recorded in Stellacci’s lab—the obvious ripples are gone, replaced with dots that Stellacci and coworkers connect to form stripes.

Others agree that the existence of stripes still isn’t clear. Stellacci and his collaborators did a good PSD analysis, says Vladimir V. Tsukruk of Georgia Institute of Technology, a materials scientist with a background in scanning-probe microscopy. “But stripes haven’t yet been proven.”

Given all the work he’s done finding outside collaborators to verify the existence of stripes, Stellacci says he believes his critics are trying to smear his reputation. He says that on Twitter, Lévy has compared him with Jan Hendrik Schön, infamous for committing fraud in his research. He finds this comparison unfair, given that he has never been found guilty of such misconduct. “My critics are after me, not the science.”

Lévy, Moriarty, and coworkers have also written a new article, “Critical Assessment of the Evidence for Striped Nanoparticles,” which examines Stellacci’s data old and new, STM and otherwise. They uploaded it to the preprint site arXiv on Dec. 24 of last year and submitted it to PLoS One shortly thereafter. With more than 240 comments, it is currently the most discussed paper on the online commenting site PubPeer.

In a Feb. 3 blog post on PhysicsFocus, Moriarty responded to Stellacci’s claims of being bullied. “Can I understand why Francesco might feel victimised? Yes,” he wrote. “As a fellow scientist, I can entirely appreciate that … a challenge to our research can feel like a direct criticism of ourselves.” But, he adds, research that is publicly funded should be debated in the open.

The Nottingham microscopist tells C&EN that another reason he’s taken the gloves off in this fight is that a former student of Stellacci’s, Predrag (Pedja) Djuranovic, came forward after Lévy began blogging, indicating that he had brought up concerns about the striped nanoparticle data with his adviser back in 2005.

At the time, Djuranovic was a new grad student at MIT tasked with studying whether a mixture of molecules formed stripes on semiconducting nanoparticles rather than the gold ones usually used in the lab. When he began to play around with the microscope, he says, he noticed that rippled features disappeared when he turned the gain down.

Concerned, he went back to basics and imaged particles with only one type of molecule coated onto them as well as an uncoated, roughened gold surface. Each time, he saw that he could produce ripples in the images just by turning up the STM gain.

Djuranovic claims he went to Stellacci a number of times to discuss his findings and that they had some heated arguments. Eventually, the student was put on a new project. Stellacci says it is his policy not to discuss former students and would not comment on Djuranovic’s story.

But Djuranovic couldn’t let things go. “It felt unethical not to raise the issue further,” Djuranovic says. So he discussed his concerns with his department head at MIT and eventually put them into writing for the university’s Office of the Vice President for Research. A few months later, an official investigation of Stellacci launched at MIT and Djuranovic left Stellacci’s research group.

MIT closed its review three years later, in 2008, finding Stellacci not guilty of academic misconduct. The university did, however, suggest that Stellacci should do additional work to substantiate his 2004 findings.

Stellacci tells C&EN that he was already working on a fuller statistical analysis of his data and addressed the suggestion when he published it in 2009 in the Journal of Scanning Probe Microscopy (DOI: 10.1166/jspm.2009.1004). No other students have come forward expressing concern.

For now, it seems Stellacci and his critics are at a stalemate. “They say they don’t see stripes,” Stellacci says. “It’s not only that I say I see stripes. Three microscopists who have imaged my particles see stripes. The referees of papers see stripes.”

The ETH Lausanne scientist also says he once offered to host Lévy in his lab to run experiments and get to the bottom of the issue. Lévy acknowledges the invitation but says it was made under the condition that he withdraw his 2012 Small paper.

Moriarty also says he’s extended an invite to Stellacci to visit his lab to no avail.

Paul S. Weiss, a pioneer in mixed molecule assembly on flat surfaces and editor of ACS Nano, says the debate has to end eventually. “The early data were inconclusive,” he says. “As a field, we’re still going. New tools will be developed and applied to sort out what the arrangements of mixed molecules on nanoparticles look like. Hopefully, we’ll discover new phenomena along the way.”