Hacker News new | past | comments | ask | show | jobs | submit login
Cornell researchers see atoms at record resolution (cornell.edu)
231 points by filoeleven on May 22, 2021 | hide | past | favorite | 37 comments



I am awestruck that the direction of the blur/motion further reveals the crystalline structure:

https://i.imgur.com/g7A24w6.png

This is the coolest image I've seen in a long while.


Not quite, a crystal is a periodic structure which is defined by lattice vectors and the content of the corresponding unit cell. The pattern drawn in the linked image does not match this definition.

Imagine it like tiling a floor. In the simplest case, all tiles are the same and the unit cell is a "primitive cell". The way they are layed out gives the lattice vectors.

Now if you are using two differently styled tiles (e.g. a checkerboard pattern of black and white tiles), then the unit cell is now a "conventional cell" and consists of these two tiles. The lattice vectors now point accordingly from one tile of a type to the next one of the same type (e.g. white->white). You need two vectors that do not point in exactly the same direction to be able to cover the floor in two dimensions. Remember, you have to be able to cover the plane by moving the unit cell by integer multiples of the lattice vectors if you did it correctly.

Edit: this might be it:

https://i.imgur.com/75ltMjb.png

If you shift this pattern by any integer multiple of these vectors, the pattern that it comes to lay upon will look the same as the one that was shifted.

Here are some additional links.

https://en.wikipedia.org/wiki/Bravais_lattice

https://en.wikipedia.org/wiki/Unit_cell


I’m curious to know where you found this image? Because I looked through the paper (https://arxiv.org/abs/2101.00465) and couldn’t find it…


I edited the image in the article. It appears to be cropped from Fig 2A on page 11.


Ah, thank you! I had assumed the white lines were present in the original image.


Great spot!


It will always amaze me that 100 years ago when microscopes were quite primitive that physical chemists could predict the shapes of the electron shells. The knew the inner ones were spherical, a little further out were dumbbell shaped, etc. Now pictures like this help add more context to these more complex structures, but the hybrid dumbbell shape of the shell is easily recognizable.


At first my reaction was: "No way, it has to be a lot more recent than a 100 years." But then I realized that 1920s was a 100 years ago.

The "History" section of this page was interesting: https://en.wikipedia.org/wiki/Molecular_orbital_theory It apparently took most of 30s and 40s for those theories to mature, and the ideas of the then new quantum mechanics to be applied to molecules.


"Student describes the ‘90s as 'the late 1900s' and now we all feel old" — https://mashable.com/article/late-1900s-student-professor-bu...

:-\


I love how the meaning of "atom" is evolving with our understanding of the world. The origin is from that which is uncuttable/indivisible while our modern understanding of physical atoms are things made up of sub-atomic particles. Even in CS we have "atomic operations" which can be made up of multiple instructions.

I wonder how we will define "atom" in another 100 years.


An atomic operation is a sequence of steps that are indivisible in their effects. That's a return to the original definition of "atom." Atoms as constituents of molecules are the exception from the trend and a misnomer.


And on your computer, your cpu is doing a lot of stuff during that atomic operation even cutting it in smaller parts...


There are two major uses of the term "atomic" in CS. The sense is either that of database transactions, in which either everything happens together or nothing happens at all, or that of concurrency, in which... really, it's the same thing, but with a focus on whether an intermediate state can be visible from another thread/process, as opposed to whether an intermediate state can be written to persistent media. An atomic operation is one in which, whatever method you use to observe the affected state, you will only see the state from before the operation begins or the state from after it completes.

It doesn't matter whether the CPU does the operation by magic or by some physical process involving the passage of time; that's not part of the concept. The operation is not divisible; it isn't possible for part of it to occur.


Mostly, I agree. However, in a multitasking environment anything that is multiple instructions can be significantly split up in time. While not the primary effect of the op, the op itself is still divided which can have other non-negligible secondary effects on a large system.


Atomic operations are made up of multiple instructions, but the atomic in the CS sense means that they need to be done together. This is a different concept from being made up of subatomic parts.


Always thought we stopped to soon. At some point things may cease to be divisible. Plank scale is still pretty deep down the well.


Or perhaps they'll be some form of Mobius strip-like string-particle item that you can divide in two but still have a single item afterwards. Like self-healing strings that you divide in one [set of] dimension[s] and which simultaneously close up in some other dimension.

Whether such an item is mathematically possible is left as an exercise for the reader ;o) (just an idea I had).


I like it.

There is a super fun ted(?) talk, that I can’t seem to find, with a guy constructing and cutting Möbius strips, and variations on that theme. Pretty unintuitive what a little twist can do!

...and I seemingly still cannot find it. A google will reveal many such videos, I think, but no time to search them out.


Matt Parker has demonstrated this at least a couple of times on YouTube, I bet he’s who you’re thinking of. It is surprising!

https://youtu.be/-u-O_2Hz82I


I think I was thinking of a Matt Parker video. I hadn’t seen this one though, thanks!


perhaps you are thinking of this video from Numberphile with Tadashi Tokieda https://m.youtube.com/watch?v=wKV0GYvR2X8


I forgot about this one. It’s great, thank you! His video with the counterintuitive bead flows was great as well, if I recall. Lots of good stuff from this guy.

Well, both numberphile and Tadashi Tokieda have lots of good stuff, I should say!


Atomicity in CS is about isolation not about the smallest number of operations.


"Ptychography works by scanning overlapping scattering patterns from a material sample and looking for changes in the overlapping region"

Interesting, we use multi-patterning to produce silicon already to get the light down to scale for process nodes - this is kinda the reverse and do wonder if some of this research may have applications in silicon production.


Very cool. The image looks to have a pair of atoms but the material is PrScO3, I would expect to see three. Maybe someone more knowledgeable in chemistry could explain this?



Alternative rock album cover in 3... 2... 1... /s(?)


Funny, I immediately thought prog rock album cover.


what's the empty space between the atoms made of?


Nothing (or a probabilistic electron cloud? I don’t think it counts as “being made of” something in a meaningful way)


when you’re at that scale you need to have a good idea of what you mean by “made of” before you can get an answer. You end up trying to figure out what it means for a space to be occupied or unoccupied and coming to the conclusion that space is either 100% occupied or 0% occupied but not really anything in between which isn’t just an arbitrary threshold


Beautiful image, truly amazing work.


And for a limited time only, ptycho.com is still available!


Praseodymium orthoscandate (PrScO3) is beautiful.


I tried to find info about this compound but only got links to various articles reporting this story...


I skimmed a couple of the top hits on GScholar, seems like it’s mostly used as an epitaxial substrate for growing thin film functional materials like BaSnO3. Crystals that would be interesting to research as potential ferroelectric or multiferroic devices, that sort of thing

Edit: I found the context in this paper really helpful / interesting: https://dx.doi.org/10.1063/1.2396920





Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: