A. MAUREEN ROUHI, C&EN Washington

Given the breakneck pace of research on fullerene nanotubules, it's inevitable that you immediately think of fullerenes when you hear the word nanotubules. If you need a respite from fullerene nanotubules, however, relax. What follows is about nanotubules that are not fullerenes, that are easy to prepare, and that are more versatile in form and function than fullerene nanotubules have proven to be so far.

In the din of fullerene nanotubule mania, these other nanotubules offer a breath of fresh air. They are prepared from almost anything by templating onto porous membranes that are commercially available for use as filters. In the lab of Charles R. Martin, a chemistry professor at the University of Florida, the pores in these membranes become nanotubules and nanowires, which are among the hardware required in a nanotechnological world.

	ALL IN A ROW Template-based synthesis produces parallel arrays of monodisperse nanotubules.
	COURTESY OF CHARLES MARTIN

The idea of transforming filtration membranes to nanomaterials came to Martin during a conference about 15 years ago, when he saw an enlarged image of the micropores of a porous alumina membrane projected on a screen. He remembers thinking, "If I deposited metals in those pores, I could make very small electrodes."

Depending on the pore's diameter, the electrodes could have very small diameters, down to tens of nanometers or even smaller. Martin, an electrochemist, could use them to make nanoscopic electrodes. But Martin also noticed that when depositing gold into pores to form gold nanowires, what forms first is a tubule, as the metal initially deposits onto the pore wall. He remembers telling his students: "We can make these nanoscopic gold electrodes, and they have proved to be interesting. But these nanotubules that we're making basically by accident could be interesting, too. I wonder what they are good for."

Now, these nanotubules synthesized within membrane pores--or nanotubule membranes--are proving to be interesting and useful. They are enabling new ways of separating and detecting analytes. Currently, Martin's group is working feverishly to interface the nanotubule membrane architecture with biological recognition agents for applications in chiral separations and single-molecule sensing.

"It's a classic example of going along a route and taking the blinders off," Martin notes. "When you're so focused on a particular goal, you often can have blinders on and not see the other interesting things that might be evolving."

 A MAJOR ADVANTAGE of nanotubule membranes is that they are easy to make. Each pore in the membrane essentially is a nanobeaker in which chemistry can occur. "Almost anything can be synthesized in these pores, as long as a suitable chemistry is available," Martin says.

Metals can be deposited electrochemically or through so-called electroless plating by using a reducing agent to plate the metal from the solution onto the surface. Conductive polymers, such as polypyrrole, can be synthesized electrochemically, as well as chemically, by immersing the membrane in a solution of monomer and a polymerization reagent. Nanotubules of inorganic materials such as silica or titania can be prepared through sol-gel chemistry, and carbon nanotubules can be made through chemical vapor deposition.

VERY SMALL The inside diameter of this carbon nanotubule prepared by template-based synthesis, about 60 nm, is three orders of magnitude smaller than the diameter of a human hair.

COURTESY OF CHARLES MARTIN

The nanotubules come out embedded in a membrane, aligned, and monodisperse--that is, of uniform diameter and of a uniform length equal to the thickness of the membrane. Various porous membranes can be used. Martin's group works mostly with monodisperse polycarbonate and alumina membranes. These can be bought in different pore sizes and densities (pores per unit surface area on the membrane). Sometimes, membranes of unusual pore characteristics can be custom ordered from commercial suppliers. But for the most part, Martin says he and his coworkers prepare their own microporous aluminas with various pore sizes for their templating applications.

Depending on reaction times, thin- or thick-wall tubules are formed. The tubules can be capped on both ends to create a cluster of individual confined spaces. Or the membrane can be removed to release individual nanotubules.

The generality of the template-based synthetic methodology broadens the scope of nanomaterials that can be prepared. Over the years, with funding from the National Science Foundation and the Office of Naval Research, Martin has found many clever uses for these nanotubule membranes. Separation is one big application area. 

ONE OF THE EARLIEST separation applications to be explored was selective-ion transport. A simple experiment with a gold nanotubule membrane in a U-tube cell demonstrates this phenomenon. On one side of the membrane is a feed solution containing KCl, which is colorless, and the cationic dye methylene blue. On the other side is a receiver solution of KCl.

After a while, the colorless receiver solution turns blue, indicating transport of the cationic dye across the membrane. But when the feed contains permanganate anion, which is red, the receiver remains colorless. The anion, although smaller than the cationic dye, can't cross the membrane. Only cations pass because the nanotubule walls have excess negative charge due to adsorbed chloride ions, Martin explains.

In another experiment, the excess charge is controlled by applying a potential to the membrane. If a positive potential is applied, the nanotubule wall will have excess positive charge and the nanotubule membrane will reject cations and transport only anions. At negative applied potentials, the membrane will transport cations and reject anions. At the specific potential where the wall is electrically neutral, the membrane will be nonselective. Thus, the gold nanotubules make up a switchable ion-exchange membrane.

"This experiment may be the coolest idea we've ever had," Martin tells C&EN. It is well known that the transport properties of polymer membranes can be changed by chemically attaching positive or negative groups, he explains. That's the basis of ion-exchange resins, for example. "But our gold nanotubules are the first example of a material whose transport properties can be changed by electrostatic charging. There was no precedent for this phenomenon prior to our publication in Science [268, 700 (1995)]."

Another early application was filtration based on size. Using the same simple U-tube setup, Martin demonstrated size selectivity in permeation experiments using a pair of chemically similar compounds that differ in size. For example, when a mixture of pyridine (molecular weight 79) and quinine (molecular weight 324) is introduced to one side of the U-tube, only pyridine is detected on the other side of the nanotubule membrane. 

USING MEMBRANES to filter molecules on the basis of size is not new. Dialysis machines, for example, separate macromolecules from small molecules. However, prior to Martin's nanotubular membranes, synthetic membranes with uniform, molecule-sized pores that can filter small molecules from other small molecules did not exist. The work was also published in Science [278, 655 (1997)].

Having shown these membranes to have charge- and size-based selectivities, Martin next attempted to engender chemical selectivity by modifying the chemistry of the nanotubule walls. Taking advantage of how easily gold reacts with thiols, Martin and coworkers John C. Hulteen and Kshama B. Jirage have prepared nanotubule membranes in which alkyl and other groups are attached to the nanotubule walls through gold-sulfur bonds.

	Deck The Walls (click here for full-size image)

They showed that the attached groups affect transport through the nanotubule membrane. For example, walls decked with long hydrocarbon chains (such as C₁₆H₃₃) preferentially transport hydrophobic molecules, whereas those decorated with short polar groups (like C₂H₄OH) behave in just the opposite way.

"My postdoc Sang Bok Lee then asked, What if we used cysteine to modify the nanotubule walls?" Martin recalls.

Cysteine is a thiol as well as an amino acid. So it has both amino and carboxyl groups. At low pH, it will be positively charged; at high pH, it will be negatively charged. Lee has shown that the cysteine-decorated membrane rejects cations at low pH and anions at high pH [Anal. Chem., 73, 768 (2001)].

"This membrane thus has pH-switchable transport properties," Martin says. "There's a lot of interest in gating the transport of molecules, in being able to change transport properties just by throwing a switch. For lab-on-a-chip assemblies, for example, you need to be able to switch between different transporting states."

With cysteine anchored to the nanotubule walls, Martin begins to bridge the gap between nonbiological and biological agents that modify transport across the nanotubule membrane. His goal now is to combine nanotubule membranes with the exquisite recognition agents already designed by nature. This melding of biology with nanotechnology, or bio/nano interfacing, "is just exploding right now," he says. "It is one of the hottest things on the scientific horizon."

In the first attempt at bio/nano interfacing, Martin trapped proteins within nanotubule membranes by capping both faces of the membrane with a thin layer of porous polymer. In effect, he formed arrays of protein-filled microcapsules. Small molecules can pass through the porous polymer plugs, but not the proteins, which float freely in the confined space. Loaded with enzymes, the membrane works like a bioreactor. But if apoenzymes, which lack the cofactors that the enzymatic reaction requires, are used, the membrane acts like a chemically selective filter. 

THE APOENZYME selects its substrate from a mixture. But unable to carry out the enzymatic reaction, it releases the substrate unchanged. With Brinda B. Lakshmi, Martin used such a "sandwich" assembly to separate D-phenylalanine from L-phenylalanine [Nature, 388, 758 (1997)].

"Chiral separation is hugely important," Martin says--especially in the pharmaceutical industry, as it has become clear that the enantiomers of drugs produced as racemic mixtures could have very different properties. "We'd like to develop simple membrane-based processes for separating enantiomers."

However, the sandwich setup isn't ideal. The thin polymer films impede the flow of material, and assembling the sandwich is quite involved. "Wouldn't it be much better to have open nanotubes, where the biochemical molecular recognition agent is just chemically attached to the walls?" Martin asks.

	GATOR FANS Nanotubule researchers (from left) Lee, Naichao Li, Martin, Scott Miller, David Mitchell, and Lacramioara Trofin pose beside a replica of the University of Florida's mascot, the alligator.
	PHOTO BY ELIZABETH MEDEIROS

Taking advantage of well-established silane chemistry, the Martin group has begun anchoring biological recognition elements to the walls of silica nanotubules synthesized in microporous alumina membranes. The hydroxyl groups on silica react readily with silanes, and silanes that end with aldehyde groups react with the free amino groups in proteins. With that, Martin and his coworkers have created bio/nano assemblies of proteins immobilized in silica nanotubule membranes.

Of particular interest is one example involving immobilized antibodies. The specific antibody that the Martin group has been using is one that recognizes a promising antitumor agent. Martin believes that the bio/nano assembly could be used to separate the bioactive enantiomer from four stereoisomers that are formed when the compound is made in the lab. So far, they have achieved a twofold higher transport of the desired enantiomer over the other stereoisomers. "That's not good enough," Martin says. "But it's a great first start."

Nanotubule membranes could also lead to more sensitive analytical methods. "In analytical chemistry, we are always striving to push detection limits to lower and lower values," Martin says. "Recent work from other labs has pushed them to the absolute limit--single molecules. We have demonstrated detection limits as low as 10^-11 M, which is very, very low. And we're working very hard to get to single molecules."

Martin and Yoshio Kobayashi achieved the highly sensitive detection at 10^{- 11} M with unadorned gold nanotubule membranes through which an ionic current passes. And they demonstrated it with nonelectroactive molecules [Anal. Chem., 71, 3665 (1999)].

In the presence of an electric field, cations in an electrolyte solution migrate in one direction through the nanotubules, and anions travel the other way. Undisturbed, the two-way traffic registers a steady current.

"Now, suppose we throw in a molecule of comparable dimension to the inside diameter of the nanotubules," Martin explains. When these molecules diffuse into the membrane, they seem like big boulders to the small ions making their way through the tunnels. Martin reasoned that the ions would have to wind their way around the big boulders and that the zig-zagging should create a drop in the current.

"And that's exactly what happens," Martin says. "Even though the molecule is nonelectroactive, its ability to disrupt the current passing through the nanotubules allows it to be detected by electrochemical means."

The experiment brings to mind the work of Hagan Bayley, professor of medical biochemistry and genetics at Texas A&M University Health Science Center. Bayley uses pores formed in lipid bilayers by the bacterial protein a-hemolysin to detect individual molecules via changes in the current through a single pore. With genetic engineering and chemical modification of a-hemolysin, the Bayley group has been using this approach to sense a wide range of analytes, from small ions to oligonucleotides.

Single-molecule sensing is possible because the a-hemolysin pores have molecule-size dimensions in both diameter and length. Also, "we introduce binding sites for our analytes in the lumen of a-hemolysin," Bayley says. "This increases the dwell time of a single analyte and introduces the element of specificity."

In contrast, the nanotubules prepared by Martin, although of molecular size in diameter, are long tunnels. At any time, many molecules could be in a tunnel, and it's not possible to see only one going in and coming out.

To detect individual molecular events, the tunnel needs a constriction of molecular dimensions. Martin believes this can be achieved with conical tubules. On the end where the tubule is wide, molecules wouldn't block the pathway. Only when a molecule reaches the narrow end would it look like a boulder. As it gets lodged in a hole with a diameter and length comparable to its own dimensions, the molecule stops the current.

Another way to detect single molecules would be to ring the tubule exit with molecules that will obstruct the opening. And a third way, involving bio/nano interfacing, would be to take Bayley's pore-forming a-hemolysin and install it on one end of the tunnel. Martin is collaborating with Bayley to explore this approach

Interfacing the pore-forming protein with the nanotubular architecture could yield sensors that can be used in field experiments outside the laboratory. "One problem we have is making a robust device," Bayley says. "Planar lipid bilayers are not sufficiently stable for use in commercial sensors. If we can place our pores within the nanotubules, they should be stabilized because we will have dispensed with lipid bilayers altogether."

Both Bayley and Martin are excited about this collaboration. "Integration of the protein pores and the nanotubules will require some nifty chemistry, and this is what we are focusing on initially," Bayley says.

 ACHIEVING LOW pore density on the membrane is another problem to be solved. Commercial membranes have as many as 10⁹ pores per cm²--much too high for single-molecule detection--where 10⁹ pores means a billion molecules. "There's not much interest for the filtration industry to have membranes with low porosities," Martin notes. But Martin has been able to obtain custom-made porous membranes with as few as 40 pores per cm².

With such low-pore-density membranes, "we have the possibility of isolating a piece of membrane with only one pore," Martin says. To do that, they prepare gold nanotubules in the membrane and deck them with fluorescent thiols. "You hit the tubule with black light and it will glow. It would be so cool if we could install a single a-hemolysin channel on an individual nanotubule."

Martin is not alone in exploring the possibilities of template-synthesized nanotubule membranes. For example, the group of Pieter Stroeve, a professor of chemical engineering and materials science at the University of California, Davis, uses gold nanotubule membranes with self-assembled monolayers for separations of proteins of similar molecular weights as well as stereoisomers. And at the University of Illinois, Urbana-Champaign, chemistry professor Paul W. Bohn and collaborators Jonathan V. Sweedler and Mark A. Shannon are using nanotubule membranes to develop a fluidic processor for detecting biological agents at extremely low levels.