FLUID FLOW AT THE NANO LEVEL
To tell his story, Narayana Aluru, professor in the department of mechanical science and engineering at the University of Illinois at Urbana-Champaign, turns quickly to his laptop and starts displaying images. Atomic scaffolds and schematics of tubes and particles cover the screen. "This part of my work," he says, "is understanding water and electrolytes in microchannels and nanochannels." If it was just about drawing pictures, though, this could be an easy area to explore. But it's not. It takes just one mouse click for Aluru to cover his laptop's screen with equations. Suddenly, trying to model microscopic systems of water and particles looks much more complicated.
With centuries of work behind fluid flow - going back to Da Vinci and Newton - it might sound easy enough to model something simple, say, a tube filled with water. It is pretty easy, as long as the tube is big, big compared to the water molecules flowing in it, that is. But make the tube smaller, so small that the tube's diameter and the water molecules start looking about the same size, and the entire problem gets complicated, quantum-complicated, you might say.
"Why are we trying to understand fluids at nanoscale?" Aluru asks. "Lots of applications could benefit from this, ranging from DNA sequencing to single-molecule sensing to applications in energy to water purification." For such nanosize systems, Aluru found a way to keep track of fine details without the conventional limitations.
In essence, Aluru faced three options for modeling such systems. He could use quantum equations for precise detail, but those can only track a few atoms and only over angstra. Molecular-dynamic techniques, on the other hand, track molecules and work on the nanosecond and nanometer scales. This turns out simpler equations, but don't get the idea that it's suddenly as easy as solving a polynomial. Aluru would like to model such systems with classical approaches, such as the Poisson equation, which follows the distribution of charge across millimeters for milliseconds. Those classical equations, though, miss lots of details at microscopic scales.
Aluru had an idea. He started with a simple system - a nanosize tube with charge on its surface and water and ions inside it - and modeled it with molecular dynamics and classical equations. Then, he looked for the trouble spots, essentially the places where the classical equations got things wrong. Next, he calculated a fudge factor of sorts, something that he calls the excess-chemical potential. This factor, all in one number, combines the impact of everything that gets tracked in molecular dynamics but neglected in classical equations. For example, the classical equations assume that particles are infinitely small, which lets atoms lie right against the tube. This doesn't happen in reality, and molecular dynamics shows that atoms cannot touch the tube. "That's because two molecules cannot go very close to each other," Aluru explains as he points to a graph on his laptop's screen. Aluru's excess-chemical potential, though, makes the classical equations "see" this problem. By adding this factor to the classical technique, Aluru improved its accuracy. "You need to do one molecular simulation," says Aluru, "but then you can do any scale problem."
