3 ms·

I have a question for someone on here. When I began research in grad school in 2012, graphene had already been discovered eight years previously. I understand

I have a question for someone on here. When I began research in grad school in 2012, graphene had already been discovered eight years previously. I understand that before this happened, a "2D" atomic layer of a material had never been isolated. What I want to know is: why was this considered impossible before then? I do molecular dynamics simulations (and am in fact performing simulations on graphene currently), and an individual layer of graphene easily stays intact in both a vacuum and other environments. The capability to simulate one layer of graphene was also available way before 2004. So what was it that made single-layer isolation so astonishing at the time? It seems like it would have been a perfectly reasonable assumption to me (but like I said, I suppose I'm biased by the benefit of hindsight).

jxm262 12y ago
I'm not really qualified to answer this, but I recently came across a similar question that might shed some light (old post though). https://www.physicsforums.com/threads/why-is-graphene-only-a-recent-discovery.661080/#post-4209868 https://www.physicsforums.com/threads/why-is-graphene-only-a...
tfgg 12y ago
Peierls and Landau predicted that 2D crystals can't be stable: http://physics.stackexchange.com/questions/110079/why-a-mono-atomic-crystal-layer-2d-cant-be-stable http://physics.stackexchange.com/questions/110079/why-a-mono... Either they made assumptions that don't hold, or graphene actually isn't "stable" as such but is stable enough in reality. I don't know much of the history of quantum chemistry simulations of graphene, but I'm fairly sure they existed beforehand and the experiments were a nice confirmation. Quite often it's possible to simulate things which are thermodynamically impractical, with things like an incomplete thermodynamic sampling in MD, or small periodic cell sizes artificially stabilising systems.
Osmium 12y ago
> I do molecular dynamics simulations (and am in fact performing simulations on graphene currently), and an individual layer of graphene easily stays intact in both a vacuum and other environments. You can make a lot of 'stable' things in molecular dynamics simulations that don't exist in reality ;) You could construct any number of 2D materials that exist quite happily because they're close to some local minima or other, or because of some quirk of the potential forcefields you use. So I don't think "I can make it exist in a computer" is necessarily a good reason for expecting it to exist in reality, even if in the case of graphene it's relatively straight forward. Though useful, classical molecular dynamics is also quite dumb: it's really good for simulating known materials, but not so much for discovering new ones. As for "2D" materials generally...well, we can make monolayers of a lot of things, sandwiched between other materials. And we can make other things that act like they're 2D even if they're not. There's nothing spectacular about it per se. The difficulty, as with all these materials, is that predicting that something has useful properties and actually being able to manufacture it at scale and relatively defect free is huge. You're fighting entropy, basically. So I think the surprise was not that graphene could exist in principle, but that it could be made reliably.
- Xcelerate 12y ago
  > You could construct any number of 2D materials that exist quite happily because they're close to some local minima or other, or because of some quirk of the potential forcefields you use ... though useful, classical molecular dynamics is also quite dumb I believe you're thinking of the state of simulation in the 1980s. There's been quite a bit of progress since then ;) For my work, I'm using a new reactive forcefield that's based off of DFT calculations and has a very close energetic match with experiment. Another forcefield I have used is optimized for structure and almost exactly reproduced neutron spectra for a variety of hierarchical materials, including long-range order (and that was from model systems alone). Also, QM-MM methods have recently become more popular with additional computational power, so the parts that have significant non-classical effects can be computed correctly without wasting computational power on the parts that aren't going to make a difference. As with any tool, molecular dynamics works very well if you know the theory behind it and know what you can and can't use it for. Structural carbon simulations is one of the areas where a lot of good forcefields exist (AIREBO for instance).
tim333 12y ago
Andre Geim talks about it 21 mins in to his Nobel Lecture - http://www.nobelprize.org/mediaplayer/index.php?id=1418 http://www.nobelprize.org/mediaplayer/index.php?id=1418

jxm262 12y ago

I'm not really qualified to answer this, but I recently came across a similar question that might shed some light (old post though). https://www.physicsforums.com/threads/why-is-graphene-only-a-recent-discovery.661080/#post-4209868 https://www.physicsforums.com/threads/why-is-graphene-only-a...

tfgg 12y ago

Peierls and Landau predicted that 2D crystals can't be stable: http://physics.stackexchange.com/questions/110079/why-a-mono-atomic-crystal-layer-2d-cant-be-stable http://physics.stackexchange.com/questions/110079/why-a-mono... Either they made assumptions that don't hold, or graphene actually isn't "stable" as such but is stable enough in reality. I don't know much of the history of quantum chemistry simulations of graphene, but I'm fairly sure they existed beforehand and the experiments were a nice confirmation. Quite often it's possible to simulate things which are thermodynamically impractical, with things like an incomplete thermodynamic sampling in MD, or small periodic cell sizes artificially stabilising systems.

Osmium 12y ago

> I do molecular dynamics simulations (and am in fact performing simulations on graphene currently), and an individual layer of graphene easily stays intact in both a vacuum and other environments. You can make a lot of 'stable' things in molecular dynamics simulations that don't exist in reality ;) You could construct any number of 2D materials that exist quite happily because they're close to some local minima or other, or because of some quirk of the potential forcefields you use. So I don't think "I can make it exist in a computer" is necessarily a good reason for expecting it to exist in reality, even if in the case of graphene it's relatively straight forward. Though useful, classical molecular dynamics is also quite dumb: it's really good for simulating known materials, but not so much for discovering new ones. As for "2D" materials generally...well, we can make monolayers of a lot of things, sandwiched between other materials. And we can make other things that act like they're 2D even if they're not. There's nothing spectacular about it per se. The difficulty, as with all these materials, is that predicting that something has useful properties and actually being able to manufacture it at scale and relatively defect free is huge. You're fighting entropy, basically. So I think the surprise was not that graphene could exist in principle, but that it could be made reliably.

Xcelerate 12y ago
> You could construct any number of 2D materials that exist quite happily because they're close to some local minima or other, or because of some quirk of the potential forcefields you use ... though useful, classical molecular dynamics is also quite dumb I believe you're thinking of the state of simulation in the 1980s. There's been quite a bit of progress since then ;) For my work, I'm using a new reactive forcefield that's based off of DFT calculations and has a very close energetic match with experiment. Another forcefield I have used is optimized for structure and almost exactly reproduced neutron spectra for a variety of hierarchical materials, including long-range order (and that was from model systems alone). Also, QM-MM methods have recently become more popular with additional computational power, so the parts that have significant non-classical effects can be computed correctly without wasting computational power on the parts that aren't going to make a difference. As with any tool, molecular dynamics works very well if you know the theory behind it and know what you can and can't use it for. Structural carbon simulations is one of the areas where a lot of good forcefields exist (AIREBO for instance).

tim333 12y ago

Andre Geim talks about it 21 mins in to his Nobel Lecture - http://www.nobelprize.org/mediaplayer/index.php?id=1418 http://www.nobelprize.org/mediaplayer/index.php?id=1418