Sunday, February 14, 2021

Majoranas - a brief follow-up

As you can always tell by the frequency of my posting, work-related activities have been dominating my schedule of late.  In addition to the usual stuff (papers, proposals, the normal academic activities), this is the time of year when as department chair there are deadlines and activities associated with faculty and staff evaluations, departmental budgets, graduate admissions, teaching assignments for next year, etc.  Still, in the wake of this article from Wired and some breathless reactions in the news and social media, it's worth following up my prior post on the topic of solid state implementations of Majorana fermions and what the pending retraction of this paper means. 

There are two main issues.  First, it has become clear that it can be very challenging to achieve the experimental conditions needed to have clear, unambiguous evidence of Majorana quasiparticles in the superconductor/semiconductor nanowire architecture.  This is explained in detail here, for example.   The interface quality of the semiconductor and of the semiconductor/superconductor boundary is extremely important, as disorder can lead to various confounding effects.  Interfaces are notoriously challenging.  ("God created the bulk; surfaces were invented by the devil." - Pauli)  There is no reason to think that it is impossible to reach the cleanliness level needed to see Majoranas in this type of structure, but like many material-related problems, this seems like it will require even more effort. 

Second is the particular issue of data presentation in this paper and whether it was misleading.  I have not personally looked at this in depth, but others have (twitter thread).  Snipping out segments of gate voltage without making that clear, and only plotting a limited range of gate voltage (leaving out where the conductance exceeds what is supposed to be the limiting value), is problematic.  

It's important to separate these two issues.  The issues with this particular paper are not a reason to stop this experimental approach or give up trying to confirm Majoranas this way.  It's just hard, the community isn't there just yet, and this is a cautionary tale about triumphal press releases.

8 comments:

Peter Armitage said...

This whole thing makes me terribly sad, but its hard to see how some of this doesn't qualify as scientific misconduct e.g. what is detailed here...
https://twitter.com/spinespresso/status/1360355302491774980

Douglas Natelson said...

Peter, yes. That and the fact that some bits of gate voltage range in the middle of the figure were clipped out (acc to a different thread of Sergey Frolov's presentation on twitter).

Anonymous said...

As I remember, Hendrik Schon's explanation for his data falsification was that he knew what the experiment was supposed to show, so he made his data fit his own expectation. Any data that did not fit the "expected result" had to be removed. I think this attitude, to a lesser degree, is more common than most would think.

Anonymous said...

We should not put this in the "Schon" level of falsification, after all the group did come out and say it. Putting possiblemalintent aside, I think we should be encouraging authors to come out and fix these errors, rather than leaving them alone like the tons of Nature/Science papers that are wrong and will never be reproduced till the end of the universe.

Peter Armitage said...

I agree that this is nothing remotely like the Schon situation.

Anonymous said...

I'm sorry, but you are missing my point, and probably have not read the details of the Schon case, or the "excuses" he gave for his behavior. The root of the problem here is that you have scientists who have a certain idea or model in mind, who become very attached to an interpretation of their data, and will throw away or modify the data in a way that makes it fit their idea; while the rest of us obviously do not have access to the data that was thrown away. There is certainly a difference in degree between Schon's case and this one, but both cross the same line and corrupt the scientific process in the same way.

To the Majorana authors' credit, they apparently provided their raw data for review, which helped uncover the error and suggests that they were open to some level of criticism; by contrast, Mr. Schon did not make his lab books accessible.

What also irks me in this new story is that the authors doctored their data for so-called "aesthetic" purposes, according to Sergey Frolov's tweet. This is unfortunately an attitude that I have observed among several other scientists, some working in prestigious universities. Experimental science is not supposed to produce "aesthetic" results; on the contrary. And I wish this could be taught more systematically to graduate students.

Douglas Natelson said...

Anon, Schon wholesale made up results and never did the vast majority of experiments he claimed to have done. I was a postdoc at Bell that overlapped with him, and I know the story in detail. This is very different than that.

Markus said...

Thanks for sharing very informative blog post, I like to read only quality stuffs. You may also browse sites to entertain yourself.
Cheshire escort Jobs
Incalls Escort Manchester
Escorts in Manchester
Manchester Airport Escorts

Escorts in Chester
Bolton Escorts