A porous silicon membrane that is a few nanometers thick can quickly filter liquids and separate molecules that are very close in size, researchers at the University of Rochester report in this week’s Nature. The new membrane could lead to efficient protein purification for use in research and drug discovery. It could also act roughly 10 times faster than current membranes used for blood dialysis, the artificial purification of blood. In addition, the membrane could be employed as a filter to separate molecules in microfluidics devices used to study DNA and proteins and as a substrate for growing neurological stem cells.

The polymer-based membranes currently used for filtering proteins are typically many micrometers thick and have an elaborate pore structure much like a sponge. “[Filtration] takes longer because there’s much longer distance to go through, and the pores are convoluted,” says Philippe Fauchet, the electrical- and computer-engineering professor at the University of Rochester who led the research. “And a fair fraction of what needs to go through remains stuck forever in the membrane.” Researchers end up losing the smaller protein molecules that are lodged inside, says James McGrath, a biomedical-engineering professor at Rochester and coauthor of the Nature paper.

The new membrane is 15 nanometers thick, so it filters faster without trapping the molecules that pass through it, which is important if researchers want to retain both the larger and smaller proteins. “Once a molecule gets to the membrane, it takes one step, and it’s on the back side,” McGrath says.

To make the membranes, the researchers employ tools that are used to create integrated circuit chips. This should make the filters easy to integrate into silicon-based microfluidic devices that are used for protein research, where they would be useful if scientists wanted to separate a particular protein of interest from a biological fluid sample. The researchers made the membranes by first depositing a stack of three thin layers–an amorphous silicon layer sandwiched between two silicon-dioxide layers–on a silicon wafer. Exposing the wafer to temperatures higher than 700 ºC crystallizes the amorphous silicon, and it forms pores. Then the researchers etch the wafer and silicon-dioxide layers to expose small squares of the nanoporous membrane that are 200 micrometers on each side. The temperature controls the pore diameter, allowing the researchers to fine-tune the membranes: at 715 ºC the membrane has an average pore size of 7 nanometers, while at 729 ºC the average is about 14 nanometers.

McGrath says that the membrane would make a good substrate to culture neurological stem cells. Certain “helper” cells nurture stem cells and coax them into turning into neurons. To get a pure culture of the neurons, researchers are looking for ways to physically separate the helper cells from the stem cells while allowing them to exchange chemicals. “[With the new membrane,] the distance they’ll be separated by will roughly be the same size as their own plasma membrane,” McGrath says. “The pores will allow a signaling molecule to diffuse very quickly.”

The researchers believe that because of a narrower range of pore diameters, the silicon membranes could separate proteins that are much closer in size than is possible with current sponge-like filters. There are thousands of different proteins serving crucial functions in the human body, and separating an individual protein is key to understanding its structure and function. Fauchet says that by engineering a narrower range of pore diameters, the researchers could get 100 percent separation of proteins–even those that are close in size.

In laboratory tests, one-nanometer-wide dye molecules in a solution pass through the nanoporous membrane 10 times faster than through a commercial blood-dialysis membrane. The researchers plan to make the membrane stronger–it can sustain pressures of 15 pounds per square inch–so that they can push more molecules through, potentially improving dialysis speed by a factor of 100 over commercial membranes.

Some experts, however, feel that it is too early to say whether the membrane will be useful for large-scale applications such as protein purification and blood dialysis. The drawback of the ultrathin membrane is that it is difficult to make large-area membranes using the technique, says Andrew Zydney, a chemical-engineering professor at Penn State University. Current protein-purification systems in the biotechnology industry effectively use 100 square meters of membrane, he says. Even if the new membrane filters 10 times faster, which means it can filter the same amount of fluid with a 10-times-smaller area, “you’re still talking about 10 square meters of silicon membranes,” Zydney says. “I’m not convinced that that can be done in a cost-effective way.”