"We chemists have borrowed a couple of pages from the natural world to demonstrate the potential for a new approach for purifying drugs," said team member Charles Martin, a professor at the University of Florida.

As the basis for their nanofilters, the team used alumina films roughly 40 µm thick containing cylindrical pores with diameters of 20 and 35 nm. The researchers deposited silica nanotubes in the pores by a sol-gel template synthesis technique. The nanotube walls were less than 3 nm thick.

Then the scientists attached specially selected antibodies to the internal walls of the silica nanotubes. Roughly 5 nm in size, the antibodies were chosen to bind to one specific enantiomer of the chemical under test, an experimental antitumour drug.

Membranes with a pore size of 35 nm transported one enantiomer of the drug twice as fast as the other enantiomer, while membranes with 20 nm diameter pores transported the selected enantiomer 4.5 times as fast. According to Martin, the antibodies latch onto the desired drug molecules and pass them along the tube like a "bucket brigade".

The team also found that they could tune the binding affinity of the antibody by adding dimethyl suphoxide to the drug solutions.

Today, chromatography is often used to purify drugs, but the technique requires expensive solvents and is difficult to employ in large-scale production. Ultimately, nanotube membranes could be used instead.

"The idea is that you don't need to put this stuff in big columns, you don't need to pour a bunch of solvents over it, you just have this smart membrane in contact with the mixture that knows what molecule it wants and grabs it," explained Martin.

That said, for such nanofilters to be a useful industry tool, the rate and volume of flow of the desired molecule through the membrane would need to be far higher. The team suggest that techniques for achieving this might include making an ultrathin film composite and augmenting diffusive flux with pressure-driven or electro-osmotic transport.

The scientists reckon the process could reach the drug-making industry within five to 10 years. What's more, they say that in principle it is possible to obtain antibodies that selectively bind to any desired molecule or enantiomer.

The researchers reported their work in the 21 June issue of Science.