Follow Slashdot stories on Twitter

 



Forgot your password?
typodupeerror
×
Science

See the Highest-Resolution Atomic Image Ever Captured (scientificamerican.com) 39

An anonymous reader quotes a report from Scientific American: Behold the highest-resolution image of atoms ever seen. Cornell University researchers captured a sample from a crystal in three dimensions and magnified it 100 million times, doubling the resolution that earned the same scientists a Guinness World Record in 2018. Their work could help develop materials for designing more powerful and efficient phones, computers and other electronics, as well as longer-lasting batteries. The researchers obtained the image using a technique called electron ptychography. It involves shooting a beam of electrons, about a billion of them per second, at a target material. The beam moves infinitesimally as the electrons are fired, so they hit the sample from slightly different angles each time -- sometimes they pass through cleanly, and other times they hit atoms and bounce around inside the sample on their way out. Cornell physicist David Muller, whose team conducted the recent study, likens the technique to playing dodgeball against opponents who are standing in the dark. The dodgeballs are electrons, and the targets are individual atoms. Though Muller cannot see the targets, he can see where the "dodgeballs" end up, thanks to advanced detectors. Based on the speckle pattern generated by billions of electrons, machine-learning algorithms can calculate where the atoms were in the sample and what their shapes might be.

Previously, electron ptychography had only been used to image extremely flat samples: those merely one to a few atoms thick. The new study, published in Science, now allows it to capture multiple layers tens to hundreds of atoms thick. That makes the technique much more relevant to materials scientists, who typically study the properties of samples with a thickness of about 30 to 50 nanometers. (That range is smaller than the length your fingernails grow in a minute but many times thicker than what electron ptychography could image in the past.) "They can actually look at stacks of atoms now, so it's amazing," says Andrew Maiden, an engineer at the University of Sheffield in England, who helped develop ptychography but was not involved with the new study. "The resolution is just staggering."

This discussion has been archived. No new comments can be posted.

See the Highest-Resolution Atomic Image Ever Captured

Comments Filter:
  • by RoccamOccam ( 953524 ) on Tuesday June 29, 2021 @08:28AM (#61533270)
    I've just beaten that record by zooming in on the image!
    • I've just beaten that record by zooming in on the image!

      You were probably too excited to remember to "enhance"

      • In TV series, the police can do the same. Someone just says "enhance", and a license plate reflected in sunglasses captured on the cheapest camera can be easily read. I think that if the say "enhance further" a few times, they can see the atoms of both the sunglasses and the license plate!
        • Mathematically you probably could enhance an poor quality image, if you have multiple shots, perhaps a video. Not to the atomic level, but to a point where you may be able to extrapolate some details. Also if you know what you are looking for say a License Plate Number. You could use the Library of License Plate Fonts to figure out which blob best meets the actual letter. an 8 vs a 1 may have more color. If you are really good you may be able to find the Difference between a B and 8 if you notice the blo

          • Sure, if you apply enough "Google Machine Learning AI" you can enhance any pixelated footage to a high-res image of this one black guy.
          • by Tx ( 96709 )

            Mathematically you probably could enhance an poor quality image, if you have multiple shots, perhaps a video. Not to the atomic level, but to a point where you may be able to extrapolate some details.

            You can, it's called super-resolution [wikipedia.org]. If you have several images of the object, perhaps from several frames of video, you can generate a higher resolution image. Unfortunately it doesn't work well with frames from video compressed with modern video codecs, they tend to destroy the information the techniques de

        • You forgot to perform the uncrop command prior to enhancing it.

        • I love satellite images "enhanced" to show license plates not visible from the vertical.

          Bending and acquiring photons after the fact.

        • by SirSlud ( 67381 )

          Yes, that is the joke. You just explained it to anybody who has never seen TV or watched movies.

        • by hey! ( 33014 )

          You kids. Back in the days of print magazines, if you looked at the color photos through a magnifying glass you could *clearly* see the atoms. They looked just like the atoms in this picture.

      • NCIS!

    • Digital zoom doesn't count
  • by BigDukeSix ( 832501 ) on Tuesday June 29, 2021 @08:53AM (#61533366)
    The crystal is PrScO3; you can easily distinguish all three elements in the lattice structure, as well as their positions (note the alternating positions of the oxygen atoms). Semiconductor designers are going to be able to actually see where dopants end up in the crystal structure. That is very cool.
    • Can you help label them?

      The crystaline structure is clear. There is a dim element, then two types of bright elements, one with what appears to be a single atom, and one with two, each with a cloud (electron?).

      How do those map to PrScO3? It's not obvious to those of us without a background in crystals.

      • by Jaegs ( 645749 )

        I'm not a chemist/physicist, but I suspect the dimmer ones are the oxygen atoms. I cannot differentiate the brighter ones that appear as singles and pairs.

        • by Anonymous Coward on Tuesday June 29, 2021 @11:12AM (#61533820)

          I'm not a chemist/physicist, but I suspect the dimmer ones are the oxygen atoms. I cannot differentiate the brighter ones that appear as singles and pairs.

          I have journal access. You are right.
          Pr - the bright pairs
          Sc - the bright points with two arms
          O - the faint points
          The figure caption refers to a square as a rectangle which is technically true, but a bit confusing.

          • > Pr - the bright pairs
            > Sc - the bright points with two arms
            > O - the faint points

            Thanks. Confirming I knew nothing about how crystals are structured!

            Neat.

      • I read (what I could understand) of the actual article. The doublets are Pr, the singlets are Sc, and the little red circles are oxygen. The authors say that the imaging uncertainty is related to thermal vibrations of the atoms; I think this is where the 'clouds' come from in this image.
    • by The Raven ( 30575 ) on Tuesday June 29, 2021 @11:38AM (#61533898) Homepage

      Be aware that inclusions raise the computational difficulty; part of the reason they can get this level of detail is because they're assuming the material is a repetitive crystal lattice, and having the computer solve for 'what repetitive lattice will produce this scatter pattern'.

      But when you have noise in the lattice... random inclusions of unknown location... then it becomes several orders of magnitude harder to calculate. Like the Checkers -> Go progression, just because you've solved one game doesn't mean you'll be good at another where the rules are different, even if both involve just two colors of token. It was decades after solving Checkers that we had computers that were good at Go.

      Similarly, using the scatter scattering to deduce the pattern of a repeating lattice is a completely different class of computational problem than determining the arbitrary location of thousands of individual atoms, only some of which are within a crystal structure.

      I'm not saying electron ptychography won't be useful... I'm just saying that your example use-case is still science fiction, despite the progress made here.

      • Hmm, if true it sounds more like measuring the parameters of a known pattern, moreso than individual elements of that pattern. But the features in the image are not that regular so I don't understand.
  • That is not at all how I expected such a molecule to look. The Pr and Sc are almost touching. I assumed they would be separated by some of the oxygen atoms, but instead they are shoved together by them. Neat!
    • by znrt ( 2424692 )

      I assumed they would be separated by some of the oxygen atoms

      according to this diagram that's actually what happens: it's pairs of pr atoms almost touching, the pr pairs and individual sc atoms neatly separated by oxygen.
      https://science.sciencemag.org... [sciencemag.org]

      this is amazing indeed.

      • by keithdowsett ( 260998 ) on Tuesday June 29, 2021 @10:06AM (#61533592) Homepage

        I think it's a matter or perspective rather than closeness of the praesodymium and scandium atoms.

        If you look at https://materialsproject.org/m... [materialsproject.org] you can see the structure calculated from X-ray diffraction. This clearly shows that there are oxygen atoms between the Pr and Sc atoms in every direction.

         

        • Look at your picture, select the grey Sc atom on the left side, look at the oxygens attached to it, and the Pr molecule on the right of it. I see the same structure there, with more 3d than the article picture. The two Pr are stacked front to back in the center. It looks like the Pr might be oriented in a different direction in the article, but it looks like the same structure, just turned some and 2d instead of 3d.

        • Ah! That makes so much more sense. Thanks.
  • All they cared about was if they could do it! No one cared to ask if they should do it.
  • New unit of measure: nail growth per minute.... how many swimming pools of congress is that?
  • by ZipprHead ( 106133 ) on Tuesday June 29, 2021 @11:02AM (#61533790) Homepage

    The article states that they used machine learning to generate the image based off of data from their 'advanced detectors'.

    Does the generated algorithm produce an image that's accurate or an image that is pleasing to the scientist?

    How do we know the difference?

    • It's not generating an image, what it's doing is determining a lattice that computationally answers the question 'what atomic layout would produce this scattering pattern'. There are error bars (uncertainty), but it's closer to the concept of solving an array of linear equations in algebra. But instead of 4 variables in 4 equations, you have 500,000 variables in 10,000,000 equations... just straight up solving it isn't computationally feasible, so they use AI to approach the result heuristically.

      What this m

    • Although, approximately counting photons landing on an array of receptors over the course of some time period (i.e. all vision and imaging) is also a statistical process. But I too am curious about the odds of spurious results here.
  • Let me go get my glasses (cant' see a darned thing without them)

Time is the most valuable thing a man can spend. -- Theophrastus

Working...