When Thomas Theis goes to work these days in a New York City suburb, he keeps at least one eye focused an ocean away on a key strategic goal: creating the world's first nanotechnology memory device. As director of physical sciences at IBM's T.J. Watson Research Center in Yorktown Heights, N.Y., Theis oversees the work of Forschungslaboratorium Zurich, Big Blue's research center in Switzerland.
In the international community of microelectromechanical systems (MEMS) builders, the Zurich lab shines as a nanotechnological beacon of competency. Its scanning tunneling microscope won the Nobel prize and has become the standard submicron "tweezers" used worldwide to sculpt nanotechnologies atom by atom. Now IBM plans to extend that nanotechnology expertise to superdense memories and end-user applications like specialized drugs delivered directly to specific cells.
"There are broad application areas for nanotechnology, and we have been there from the beginning," said Theis.
IBM's main strategic goal the nanotechnology memory device was spearheaded by Zurich's Millipede project. Millipede harnesses MEMS technology to mark submicron ones and zeroes into a soft polymer with a comblike cantilever that resembles its namesake the multilegged millipede insect.
"Starting with the Nobel prize-winning work at our Zurich facility on atomic-force microscopes, our nanotechnology research has led to microelectromechanical systems including our Millipede project. With our millipede-like cantilever technology, we hope to achieve information densities that will be measured in the hundreds of gigabits per square inch."
Besides making denser memories, IBM experiments have used the MEMS cantilever to detect specific molecules. The MEMS "comb" had each tooth pretreated to deflect in the presence of a single type of molecule. That enabled it to detect molecules with only a single missing bond on a long protein chain, an impossible feat in real-time at any price with conventional equipment today. IBM experimenters used a laser measuring system to read out which comb teeth were being deflected, and thus which molecules were present in a test sample.
The Zurich researchers already had expertise in quality control and measurement, such as detecting harmful vapor in process control systems. The MEMS cantilever, however, works for liquids where it is submerged, making its molecular recognition capabilities applicable in areas like DNA testing.
The Zurich research center cites possible drug-delivery applications for their basic research, too. Drug companies could tailor cantilevers to puncture tiny medicine capsules in the bloodstream, but only when they are located within a cancer tumor. That would eliminate the side effects of cancer drugs which, taken orally, saturate all the tissues.
Using standard silicon fabrication techniques, the researchers created a tiny cantilever with teeth 500 microns long and 100 microns wide, but less than a micron thick. One side of the cantilever was then coated with gold, to which about 10 billion DNA half-helixes were bonded.
"These cantilevers are purely mechanical devices, etched out of pure crystalline silicon using standard chip-fabrication equipment," said Theis. "Then, one side is coated with gold and the DNA half-helix was attached to the ends of the comb's teeth."
When submerged in solution, the DNA half-helixes bonded only with a single type of molecule their other half. When they do, the additional stress caused by the tightly packed vertically oriented half-helixes makes them bend the cantilever's tooth, which can be detected with a laser.
DNA molecules are composed of thousands of base-4 bits called "base pairs" in long strings. Identification of a molecule occurs when its exact base-pair bits match with those of a string present in the test sample. The match must be perfect, right down to a single missing "bit" to have an exact match something nearly impossible to verify with 100 percent accuracy using current methods. But IBM's prototype cantilever did perform 100 percent accurate identification.
"We were able to show that not only can you use this technology to see individual DNA molecules in real-time, but also that its identification accuracy can be within a missing single base pair," said Theis.
IBM's technology demonstration depended on a phenomenon often used for molecular recognition called hybridization, where half the double helix of a DNA molecule is ripped out. Since the double helix can only re-form in the presence of exactly the right other half, the single helix can serve as a highly accurate molecular-recognition mechanism.
Various mammal antibodies tested
No matter which type of biomolecule is coated to one side of the cantilever, it is so thin that even small amounts of the detected substance are present. The cantilever's receptors become so stressed with the extra molecules that the cantilever bends about 10 to 20 nanometers, which can be measured with laser beam deflection. Both single-stranded DNA and proteins known to recognize antibodies of various mammals have been tested by IBM so far.
"It's very easy to see expanding the experiment by taking a thousand different fragments of DNA and testing whether any one of those fragments is present," said Theis. "Certainly there are so-called gene chips that can do this, but ours is a real-time readout of the composition of a sample."
To directly transduce from a sample to a physically detectable motion gives IBM's MEMS technology a distinct advantage over gene chips that depend on batch-operations. Vapor-sniffing versions of IBM's cantilever technology, in theory, could read out the chemical composition of a sample by merely pointing the device in the sample's direction, like the Tri-Corder on Star Trek.
"Real-time detection of such a highly specific biomolecular reaction has never been done before. I believe that our Zurich lab was the world's first," said Theis.
Other methods, including gene chips, label the molecules under test with a radioactive isotope, then expose them to the sample. On a gene chip, thousands of pits on the chip contain the DNA half-strands. Then the chip is submerged in the test sample, but an extra step of chemistry has to be done after exposure to the sample to detect the radioactive isotope markers.
IBM's cantilever, however, can be directly observed with a laser to deflect within milliseconds of the time it is exposed to the sample. Since an unlimited number of different types of molecules can be bonded to the cantilever's teeth, an unlimited number of real-time recognizers can be designed.
"I must caution that we have, so far, only demonstrated the principle of transducing the [constituents of matter] in real-time, but the principle so demonstrated is very general. Any molecule has some specificity for other molecules and in fact two members of this team are working on an artificial nose that uses a set of molecules which range over the various kinds of scents permitting them to recognize one kind of wine over another," said Theis.
Molecular recognition, however, is not a strategic direction for IBM. "We are in the storage business, not molecular recognition, so we will license out any medical applications we uncover. Our major emphasis has got to be on the Millipede project to commercialize gigabit-per-square-inch memory technologies," said Theis.
Millipede historyAt the Millipede project, tiny indentations are poked into a polymer layer by the submicron cantilever tips representing stored bits. Since all tips on a cantilever can be read out at the same time, remarkably high data rates can be achieved. Storage densities could theoretically approach hundreds of billion of bits per square inch more than five times the theoretical highest density for magnetic storage media.
Using the scanning tunneling microscope like a pair of molecular-sized tweezers has enabled IBM's Zurich lab to demonstrate the feasibility of making modifications on a nanometer scale, including the precise positioning of individual atoms on molecules. In comparison to electrical energy consumed on chips, the movement of nanoscale mechanical components consumes tiny amounts of energy and operates faster, plus wear is less of a problem than with macroscopic mechanical systems.
In cooperation with the IBM Almaden Research group (San Jose, Calif.), IBM has demonstrated a spinning disk on a turntable similar to a spinning magnetic disk. The cantilever was positioned so that its free end could mark and detect 100-nm-sized 0s and 1s resulting in 10-Gbit/cm2 densities 10 times the density of a DVD disk.
With a single-bit cantilever, the nanotechnology disk streamed 10 Mbits/second, but by using a cantilever with thousands of teeth, future disks could stream at tens of gigabits per second. IBM has already demonstrated the feasibility of multibit disks, with a 25-tooth cantilever, arranged as a 5-by-5 array measuring just 25 mm squared. A 1,024-tooth array, already in fabrication at IBM, features a 32-by-32 form factor squeezed into 3 mm squared. The highest measured density so far 30 nm marks puts the 1,024-tooth cantilever at 80 Gbits/cm-squared density.
No battery required
Another advantage to IBM's cantilever approach may be its utter non-dependence on power sources. For instance, to deflect a cantilever's tooth in the molecular-recognition application, no external power need be applied.
"No external power source of any kind is needed to get this transduction of the signal. Nanoscale mechanics doesn't require any kind of external battery or power supply in order to achieve a mechanical motion," said Theis.
In fact, the deflection of the cantilever is measured by a laser beam, which, of course, must be powered externally. However, once you can bend cantilevers without power, you can, in principle, use the motion to pass kinetic energy along to other MEMS devices.
"It's not hard to imagine a mechanical amplification system where the tiny motion of the cantilever is used to control a mechanism that punctures a medicine capsule when it recognizes that it's inside a tumor," said Theis.
In July, German company MicroTec announced an MEMS smart pill and a microsized submarine, both designed to do the same thing harness MEMS devices to deliver drugs directly to specific needy cells.
MicroTec builds its microstructures in submicron layers, permitting different types of materials to be integrated into working mechanisms. Its submarine, for instance, measures just 4 mm in length and has 600 micron propellers driving a medicine capsule measuring just 650 microns in diameter.
MicroTech reports that it used a focused UV laser polymerization technique to structurally harden layers of materials before etching them into three-dimensional devices, including integrated magnets in the propeller and its 10-micron-diameter drive shaft.