AI Helps Crack Salt Water's Curious Electrical Properties (science.org) 11
sciencehabit shares a report from Science: Water is a near-universal solvent, able to dissolve substances ranging from limestone to the sugar in your coffee. That chemical superpower originates, oddly enough, in water's electrical properties. It can oppose and almost entirely cancel electric fields -- including attractions among dissolved ions that might otherwise pull them together. Curiously, dissolving salt in water weakens that electrical response. Now, a team of physicists has figured out exactly why this happens, using state-of-the-art computer simulations bolstered by artificial intelligence (AI).
'This is a fundamental property of water and one can finally do a calculation in which this can be entirely predicted from first principles,' says Roberto Car, a physicist at Princeton University who was not involved in the work. The AI-aided approach should allow physicists to probe in other settings, he says, such as batteries and fuel cells. [...] The results show that most of the salinity effect comes from the disruption of the clustering and correlations produced by hydrogen bonding, the team reports in a paper in press at Physical Review Letters. The researchers can pull out even more detail, explaining exactly how disruptions propagating through the network of water molecules make the dielectric constant vary with the salt concentration in a complex, nonlinear way.
"They can distinguish all the different contributions and identify which effect is dominant over the other," Car says. Yuki Nagata, a physicist at the Max Planck Institute for Polymer Research, says, "This is more or less conclusive." More important than this specific result may be the AI-based method, Nagata adds. It could be used for more practical problems, such as analyzing the interactions of water with membranes or surfaces. Zhang says she's doing just that, analyzing the splitting of water into hydrogen and oxygen along the surface of a titanium dioxide catalyst, one potential way to generate hydrogen for fuel.
'This is a fundamental property of water and one can finally do a calculation in which this can be entirely predicted from first principles,' says Roberto Car, a physicist at Princeton University who was not involved in the work. The AI-aided approach should allow physicists to probe in other settings, he says, such as batteries and fuel cells. [...] The results show that most of the salinity effect comes from the disruption of the clustering and correlations produced by hydrogen bonding, the team reports in a paper in press at Physical Review Letters. The researchers can pull out even more detail, explaining exactly how disruptions propagating through the network of water molecules make the dielectric constant vary with the salt concentration in a complex, nonlinear way.
"They can distinguish all the different contributions and identify which effect is dominant over the other," Car says. Yuki Nagata, a physicist at the Max Planck Institute for Polymer Research, says, "This is more or less conclusive." More important than this specific result may be the AI-based method, Nagata adds. It could be used for more practical problems, such as analyzing the interactions of water with membranes or surfaces. Zhang says she's doing just that, analyzing the splitting of water into hydrogen and oxygen along the surface of a titanium dioxide catalyst, one potential way to generate hydrogen for fuel.
Salting out (Score:5, Interesting)
I enjoyed the explanation of the dielectric constant. They explained it in a way I hadn't understood before. And it's also nice to finally understand that's relationship to salting out. Salting out is when you mix salt into water to de-solubilize something else, for example alcohol or glycerin. (Unfortunately you end up with briney alcohol.)
Re: (Score:2)
Thanks for your comment. I'd not have gone and read the very interesting article without it.
AI helped to speed up a simulation (Score:2)
Yes! (Score:4, Insightful)
Now this is a proper use of AI.
This Story Has Very Little To Do With AI (Score:2)
Re: This Story Has Very Little To Do With AI (Score:2)
Why couldn't humans make the guesses? Too smug?
No margin of error? (Score:2)
"'This is a fundamental property of water and one can finally do a calculation in which this can be entirely predicted from first principles,' says Roberto Car"
No outliers, eh?
AI's role (Score:1)