Using AI to improve water filtration

Researchers in the journal ACS Central Science have reported that artificial intelligence (AI) could speed up the development of promising materials that can be used to improve water filter systems. In a proof-of-concept study, the researchers simulated different patterns of water-attracting and water-repelling groups lining a filter’s porous membrane and found arrangements that could let water through and slow down contaminants.

Filter systems, ranging from faucets to industrial systems, clean up water for drinking and other uses. Current filtration membranes struggle with water that is extremely dirty or has small, neutral molecules such as boric acid. Synthetic porous materials are generally limited to sorting compounds by either size or charge whereas biological membranes have pores made of proteins, such as aquaporin, that can separate water from other molecules by both size and charge. This is because of the different types of functional groups, or collections of atoms, lining the channels.

In the study, M. Scott Shell and colleagues wanted to use computers to design the inside of a carbon nanotube pore that can filter water containing boric acid.

The researchers simulated a carbon nanotube channel with hydroxyl (water-attracting) and/or methyl (water-repelling) groups tethered to the atoms on the inner wall. They designed and tested thousands of functional group patterns with algorithms and machine learning, a type of AI, to assess the speed of water and boric acid moving through the pore. They found that:

• The optimal patterns had one or two rows of hydroxyl groups sandwiched between methyl groups, forming rings around the midsection of the pore.
• In these simulations, water went through the pore nearly twice as fast as boric acid.
• Another series of simulations showed that the optimised carbon nanotube designs could also separate other neutral solutes, including phenol, benzene and isopropanol from water.
   

According to the researchers, the study demonstrates AI’s usefulness toward developing water purification membranes with novel properties and could form the basis of a new type of filter system. The approach could also be adapted to design surfaces that have unique interactions with water or other molecules, such as coatings that resist fouling.

The study was supported with funding from the US Department of Energy (via the Centre for Materials for Water and Energy Systems (M-WET), an Energy Frontier Research Centre) as well as additional support from the US National Science Foundation, the California NanoSystems Institute, the Materials Research Science and Engineering Centre (MRSEC) and a National Science Foundation Graduate Research Fellowship.

Image caption: iStock.com/amixstudio