Ultrathin Silicon Filter Sorts Single Molecules

A recently developed porous silicon membrane that is thousands of times thinner than similar filters used today sorts objects as small as proteins and has potential applications ranging from fuel cells to stem cells to microchip manufacturing.

The 50-atom-thick filter -- more than 4000 times thinner than a human hair -- was developed at the University of Rochester. It can withstand surprisingly high pressures and may be a key to better separation of blood proteins for dialysis patients, speeding ion exchange in fuel cells, creating a new environment for growing neurological stem cells, and purifying air and water in hospitals and cleanrooms at the nanoscopic level, its creators said.

The 50-atom-thick filter developed at the University of Rochester is as thin as some of the molecules it sorts. The nanofilter array pictured is a 4-in. wafer with 160 membranes. (Images courtesy of the University of Rochester)

"It's amazing, we have a material as thin as some of the molecules it's sorting, and even riddled with holes, but can withstand enough pressure to make real-world nanofiltering a practical reality," said research associate Christopher Striemer, co-creator of the membrane. "That ultrathinness means much higher efficiency and lower sample loss, so we can do things that can't normally be done with current materials."

The membrane is a 15-nm-thick slice of the same silicon that's used in computer chip manufacturing. In the lab of Philippe Fauchet, professor of electrical and computer engineering, Striemer discovered the membrane as he was looking for a way to better understand how silicon crystallizes when heated.

He used such a thin piece of silicon because it would allow him to use an electron microscope to see the crystal structure in his samples, formed with different heat treatments. Striemer found that as parts of the silicon contracted into crystals, holes opened up in their wakes.

The nanomembrane sorts molecules by size.

In talks with Striemer and Fauchet, James L. McGrath, assistant professor of biomedical engineering, and his graduate student, Tom Gaborski, realized that since the membrane's holes were only nanometers in size, it might be possible to separate objects as small as proteins much more effectively than is being done now.

Current molecular-level filters use a polymer-based design that is a jumble of varying holes and tunnels. The sizes of holes in the polymer model vary greatly, and since its "holes" are really convoluted tunnels through the material, they require much more time for proteins to pass through and easily clog.

Recently researchers tried to design an ideal filter by drilling holes into a thin slice of another silicon-based material with an ion beam. While the effort did result in a filter with regular holes, the process was too laborious to be cost effective, and its membrane was so brittle that it required an elaborate support structure to prevent it from shattering.

While McGrath knew he might have the exact filter researchers have been searching for, he needed to test if the predictions held up. "When you build something at this scale, you're closing in on the quantum world and you never know what the properties are going to be," he said.

When Striemer tested his design, he found that the same 50-atom thickness could hold back an astonishing 15 pounds per square inch of pressure. To test the membrane, Gaborski placed a solution of two blood proteins, albumin and IgG, behind the membrane and forced it gently through the nanoscopic holes. In just over six minutes, the albumin had passed through, but the larger IgG protein was stopped.

The team has also found a way for the nanofilter to carry a fixed charge, effectively making the hole "smaller" for molecules of a certain charge than for others. In a single filter it's now possible to quickly and easily separate molecules by their size and their charge -- a serious boon for fuel cell researchers, who wish to move only certain ions from one part of a fuel cell to another.

Separating molecules by size and charge efficiently is also the goal of kidney dialysis researchers. Johnson & Johnson recently gave the team a $100,000 grant to pursue developing the membrane's use in separating blood proteins with the hope of creating a more efficient method of dialysis.

"Its potential applications to neuroscience, cell biology and medical research may be profound. Kidneys do a much better job than dialysis machines of filtering blood proteins and keeping the ones you need, like albumin, and getting rid of toxins, which in some cases are smaller proteins," said McGrath.

"They use a type of cellulose or plastic membrane with relatively poor discrimination. We think we can engineer these membranes to provide superior discrimination of proteins, which may make the process of dialysis faster and more effective than it is today," he said.

The group sees many more applications for the membrane in the future. One of the most intriguing ideas is that it may play a role in growing neurons from stem cells.

In this electron microscope image, the pores in the ultrathin membrane are white.

Steve Goldman, Glenn-Zutes Chair in Biology of the Aging Brain and Rochester professor of neurology, discussed the technology with McGrath and colleagues and said he was impressed.

"It's a spectacularly interesting technology that opens a realm of new possibilities in fields as diverse as organ reconstitution, proteomics and microfluidics," said Goldman. "Its potential applications to neuroscience, cell biology and medical research may be profound."

Recent evidence suggests that neurological stem cells may grow better when in the immediate vicinity of certain "helper" cells. A problem arises after the new neurons are grown, when scientists need to separate the neurons from these helper cells. McGrath suggests that the neurological stem cells can be adhered to one side of the membrane, and the helper cells on the other.

The silicon membrane is about the thickness of the cell's own membranes, meaning the two groups of cells can actually touch each other through the membrane's pores without passing through themselves. The chemical communication between the helper and stem cells can continue as if the two sets of cells were in direct contact, but after the neurons are fully formed, they can easily be separated from the helper cells.

The team is working to realize the potential of the membrane by refining its fabrication. Striemer found he could "tune" the size of the filter holes depending on the temperature to which the silicon is heated, but the process is not yet accurate enough for engineers to simply select any pore size and fabricate it.

The researchers have founded a company, SiMPore, to commercialize the numerous applications of the nanomembrane. Many of the university's laboratories will be involved in testing and developing the membrane, and the founders have already been approached by semiconductor companies such as Intel to see if the filter could remove nanoparticles from solutions used in microchip manufacturing.

The team is currently testing the membranes to see how they stand up to regular wear and tear, and how resistant they are to clogging, which is a chief problem with conventional filters.

Details on the membrane appear this week in the journal Nature. For more information, visit: www.rochester.edu