The NSOM is based on the fact that the diffraction limit to resolution in optical microscopy is not a fundamental restriction. By scanning a source or detector of light very close to a sample, it is possible to generate an image whose resolution is dependent only on the probe’s size and the probe-to-sample separation, each of which can be made much smaller than the wavelength of light.
Two AT&T physicists, Eric Betzig and Jay Trautman, were looking for ways to improve the inspection of lithographic masks used with semiconductor wafers when they devised a probe smaller than 200 nm in diameter that would let them “break” the diffraction limit.
They did so by drawing out an optical fiber to a thousandth the diameter of a human hair and wrapping it in an aluminum film. They then guided laser light down the tapered region to the aperture and collected either transmitted or reflected light on a point-by-point basis.
This produced an image 10,000 times brighter than previous nearfield sources, opening the way to new uses.
Optical storage. The most potentially lucrative use of the NSOM is in magneto-optic (MO) data storage, part of a $40 billion industry.
Last August, Trautman and Betzig, working with three other Bell researchers, demonstrated a new technique that offers data densities of 45 billion bits/|in..sup.2~ That’s 100 times more storage than current MO methods and 300 times better than today’s magnetic storage methods.
At that density, you could store the entire text of War and Peace in both English and Russian in an area about the size of a pinhead. A palm-sized computer disk could hold up to 17 hrs of HDTV-quality video.
To write bits, the light intensity in the probe is increased until the medium is hot enough to form a small region of differing magnetization in a platinum-cobalt multilayer film. The bits are then read using the probe at a lower power.
Betzig believes they may be able to reach a density of 500 Gbits/|in..sup.2~ before reaching the physical limits of the technique.
So far, however, the Murray Hill researchers have only been able to read the data at the relatively slow rate of 10,000 bits/sec; conventional magnetic disks can read and write at a speed of millions of bits/sec. This is one reason commercialization is some years in the future.
Biology and medicine. Because NSOM builds on 300 years of experience with optical contrast, it could prove particularly useful in the biosciences, where the optical microscope is an everyday tool.
For example, Betzig is working with Benedict Yen of the Univ. of California, San Francisco, on finding staining procedures suitable for the use of NSOM in clinical pathology.
NSOM may also find use as a gene mapping tool, by coupling it with fluorescence in situ hybridization techniques. “We have not done this yet, but I suspect that it will happen soon,” says Betzig.
The AT&T researchers have been able to image specimens from a monkey brain on a scale equal to that of a transmission electron microscope. This suggests uses in optical pathology.
The most exciting biological use, however, could be in the study of whole living cells.
At Pittsburgh’s Carnegie Mellon Univ., cell biologist D. Lansing Taylor is working with Betzig to couple the NSOM technique with conventional microscopy to observe the fine structure of actin, the specific protein of the cytoskeleton within cells.
“One of our goals is to map where different constituents–proteins, organelles, etc.–are and how they change during locomotion, cell division, phagocytosis, and so on,” says Taylor, director of the National Science Foundation’s Light Microscope Imaging and Biotechnology Center.
Studying cells with traditional microscopy has its limits, he says. “You have to do optical imaging, then do electron microscopy, but the problem with EM is that you don’t have the ability to look at as many components that are specifically labeled,” says Taylor. “You get better resolution, but you lose constituents.”
With NSOM, however, Taylor is able to use an optical microscope at, say, |is less than~100 x magnification, then use near-field to get a higher resolution image (|is greater than~50,000 x) of a much smaller domain–all with the same sample preparation. He can label each sample with as many as six different fluorescent probes.
“Near-field allows us to maintain the power of fluorescent-based agents, but at higher resolution,” he says.
Taylor says that his group soon will publish a paper with Betzig on the fine structure of actin. The ultimate technical challenge will be to make this kind of study from living cells–as Taylor puts it, “taking dry bumpy stuff and putting it under water.”
Semiconductor research. The NSOM holds great promise for its original purpose, the inspection (and possible repair) of lithographic masks.
“Something will replace optical microscopes for inspecting masks in the near future,” says Trautman. With its ability to image narrower line widths, the NSOM could be that device.
It could likewise have a role in super-resolution optical lithography on a research scale, comparable to that available with electron beam lithography, says Trautman. “You could presumably do this in a closed-loop system, to expose the mask and inspect it at the same time.”
Timothy Harris, supervisor of solids characterization research at Bell Labs, and physicist Robert Grober have adapted the NSOM for near-field spectroscopy of quantum devices.
In the last year, Grober constructed an NSOM designed to work in liquid helium because the fluorescence of semiconductors is far more efficient at low temperatures. This reduces the background signal while getting a brighter fluorescence spectrum from the sample.
“It’s up and running, and it can give an image, or we can do a fluorescence emission spectrum or a fluorescence excitation spectrum,” says Harris.
“So far, we’re still testing,” he adds. “The original premise was correct: NSOM was capable of creating signal without background.” And while he hasn’t yet proven single-molecule detection, says Harris, he has begun experiments that could lead to such a capability.
Another challenge, says Harris, is to probe single objects in the 100 to 200 |Angstrong~ range: quantum wires and dots. Such objects could be used to make detectors, transistors, lasers, and optical or electronic devices.
“There’s some theory that says such devices might be useful, but it’s difficult to get an ensemble of objects that are testable,” he says, adding that the NSOM could permit testing of one such object.