In November 2004, a woman in North Carolina with potentially suffocating growths in her larynx and trachea had them removed by a high-power laser and went home the same day. This condition had never before been treated without anesthesia and operating-room surgery. Six years earlier, physicists at MIT used supercomputers to learn something no one knew about mirrors.
The two events are linked. A new laser technology, developed from a startling insight into the physics of light, may have saved the woman’s life and, at the least, promises huge savings in the treatment of her disease — recurrent respiratory papillomatosis — one that affects tens of thousands of people in the United States alone.
Yoel Fink (left) and John Joannopoulos hold a length of OmniGuide flexible optical fiber with a lens with an image of an open human throat on the computer screen.
It’s a success, furthermore, that exemplifies how supercomputing is no longer merely a supporting character, but with increasing frequency plays a lead role in scientific discovery. In 1998, John Joannopoulos and his team of researchers at MIT discovered what has come to be called a “perfect mirror.” Their eureka moment came not in the laboratory or with pencil and paper working out of mathematical theory. It happened because a computational model produced results no one expected.
For the past decade, Joannopoulos and his team have pushed forward new understanding of “photonic crystals” — fascinating materials, crafted from layers of silicon, that have unprecedented ability to trap, guide and control light. While he works closely with a laboratory team, headed by MIT professor Yoel Fink to fabricate these challenging materials, a key to this work, driving it forward, has been computational simulations that predict — successfully and precisely — how photonic crystals will work in advance of actually making them. “Computation,” says Joannopoulos, “has played a dominant role in the study of photonic crystals.”
These two cross-sectional images represent a schematic (top) of a model OmniGuide hollow core fiber and the first visualization (above) from a computational simulation by John Joannopoulos of the same fiber. The “perfect mirror” photonic reflector consists of alternating concentric layers (green and blue) of dielectric with differing indices of refraction. The visualization shows boundaries between the dielectric shells (blue circles) and power density (increasing from red to yellow) of a light beam contained within the hollow core.
It may be the most significant advance in mirror technology, said the New York Times, since Narcissus fell in love with his own image in a pool of water. The perfect mirror is so called because it reflects light at any angle with virtually no loss of energy. As a result it makes possible a number of applications in optical technology, the most significant to date being flexible optical fiber that can transmit the high-powered CO2 lasers used in endoscopic surgery.
Until the Joannopoulos team’s 1998 finding, reported with a paper in Science, mirrors were understood to come in two basic flavors, both with inherent limitations. Everyone who looks in the bathroom mirror for signs of life in the morning knows about metallic mirrors. They work all too well for seeing your own face, but they don’t work to make optical fiber because a large portion of the light leaks away, absorbed by the metal, rather than reflected.
For optical fiber and other applications where energy loss matters, the choice has been mirrors made from dielectrics — materials that don’t conduct electricity well. Dielectrics generally don’t reflect light well either, but scientists have found ways to alternate thin dielectric layers of different reflective properties to achieve reflection without energy loss. The drawback has been that these dielectric mirrors reflect light only from certain angles, and their application depends on being able to use light at a limited range of angles and frequencies.
This limitation was thought to be a law of nature, like gravity - no way to get around it — until 1998, when Joannopoulos’ team noticed anomalous results from a computational model of a photonic crystal mirror they were running at the San Diego Supercomputer Center. The light seemed to reflect at a much larger angle than was thought possible. “We saw some interesting results in the computation,” says Joannopoulos. “Then came the theory to explain the computation, and then came a real experiment making something like this and testing it.”
The result: a multi-layered dielectric mirror that reflects light from all angles without energy loss. Within a few years, the perfect mirror proved to be the solution for delivering a high-powered laser via flexible optical fiber.
Fiber optics to transmit visible light, based on conventional dielectric mirror technology, has been around for years. These silica-based fibers have a light-carrying core with an index-of-refraction higher than the surrounding material. This layered approach traps light within the inner core -called “total internal reflection.” It works well for visible light, but high-power lasers — such as CO2 lasers used in endoscopic surgery — will melt conventional optical fiber.
Joannopoulos and Fink realized that the perfect mirror offered a potential solution for high-power transmission. With further computations and pioneering laboratory work, the team developed a hollow-core fiber — essentially a dielectric perfect mirror rolled up into a tube — designed in such a way, based on photonics, to transmit high-power lasers.
These scanning electron microscope images show three cross-sectional views (at increasing magnification) of a hollow cylindrical fiber, fabricated by Yoel Fink and colleagues, that can transmit a CO2 laser beam with minimal energy loss. Layers of a chalcogenide glass (white), arsenic triselenide (As2Se3), alternating with a thermoplastic polymer (gray) surround the hollow core (black). The same polymer is used to coat the fiber. The scale bars (white) indicate microns (100, 10 and one). A normal human hair is about 50 microns.
To take this idea beyond the laboratory into useful applications, Joannopoulos and Fink in 2000 helped to found a company, OmniGuide Communications, to develop and market the new hollow-core fiber. Further computations over the next few years — at San Diego, Illinois and Pittsburgh — explored other fundamental issues and phenomena of this new class of cylindrical photonic-crystal fiber.
In endoscopic surgery, the lack of a fiber for high-power transmission has meant that the laser had to be delivered to a patient via an apparatus with an articulated arm and large handpiece — which has precluded using these precise lasers for many minimally invasive procedures. For this reason, the surgery to treat RRP required dislocating the patient’s jaw and general anesthesia, so that the laser could be brought close enough to the affected area.
A test case for the OmniGuide hollow-core fiber presented itself last year. In serious cases of RRP, the surgery often must be repeated to keep the breathing passage open. Dr. Jamie Koufman, director of the Center for Voice and Swallowing Disorders of Wake Forest University Baptist Medical Center, had a woman RRP patient who had undergone several previous RRP surgeries, but once again had developed near-total obstruction of the larynx and trachea.
Koufman obtained FDA approval to use the prototype fiber. She used a numbing topical spray in the throat and trachea, requiring no anesthesia, and with a CO2 laser delivered via an OmniGuide fiber cleared the RRP growths. The patient, who went home that day, is doing fine.
“Unsedated, laryngeal laser surgery with the OmniGuide fiber is a dream come true for me as an endoscopic surgeon,” said Koufman, “and the patient loved it because it was easy for her.” Typical cost of RRP operating-room surgery with general anesthesia is $25,000. With expected FDA approval, the new procedure promises very large cost savings nationally.
“These novel optical fibers, based on photonic crystals,” says Joannopoulos, “offer a new approach for medical lasers, making it possible to guide a CO2 laser beam, which can cut tissue with high precision, into a patient’s body through a very small incision. It will likely prove itself useful for many procedures.”
“Computational science has come a long way over the past 20 years. Even well known equations can have remarkable unexpected consequences that we would never learn about without these powerful computational engines, such as LeMieux (PSC’s terascale system). This is just one advance that highlights how these machines are invaluable tools of discovery.”