The promise of plasmonics

One of the most interesting technologies emerging in photonics at its heart isn’t an electromagnetic wave at all. Instead, the phenomenon of interest is an optically generated wave of free electrons that propagates along the interface between a metal and a dielectric. It is a surface plasmon, and it is the fundamental mechanism of a field known as plasmonics.

Generating a surface plasmon requires illuminating a metal/dielectric interface with an optical beam that is tuned to the resonance of the material. Surface plasmons can also be generated electrically, though the process is less understood. Surface plasmons can propagate, holding forth the prospect of chip-level interconnects, or they can be localized surface plasmons, offering multiple orders of magnitude of enhancement in techniques like Raman spectroscopy.

Caltech’s Harry Atwater, credited with coining the term plasmonics, wrote an article for SPIE’s oemagazine in 2002 on his group’s work in fabricating plasmonic waveguides that could beat the diffraction limit (read the archived article at SPIE.org/atwater).

Atwater’s group has now developed a slot waveguide, a structure that sandwiches the dielectric between layers of metal for more effective waveguiding. More recently, they’ve demonstrated an all-plasmonic modulator in which a plasmon not only modulates the propagation but also can be used to switch another plasmon operating at a different frequency. The modulator is a step along the way to Atwater’s eventual goal of developing a plasmonic transistor—dubbed a plasmonster in his group.

The modulator displays low power operation and nonlinearity, two of the four necessary characteristics of a plasmonic transistor (gain, nonlinearity, input/output isolation, and low-power operation).

As to achieving the other two? “I think there’s no fundamental obstacle,” says Atwater. “There are pragmatic obstacles. We already have a second-generation module that I think will solve the input/output isolation problem.”

That just leaves the requirement to build a device with high gain per unit length, a trickier accomplishment, Atwater acknowledges. “It’s something that’s challenging, but, of course, every laser achieves this, so it’s simply tantamount to building a plasmonic transistor with a material that has high enough gain per unit length to overcome the loss. It’s a technology issue that I think will yield with effort.”

When asked whether the eventual application would be optical computing, Atwater is circumspect. “There’s no sense replicating something that does extremely well at what it does—namely silicon CMOS,” he says. “You want to think about the applications where you want to do computing at high frequencies or speeds at which you can’t do it currently.” He cites lidar or radar image processing, or image recognition as examples.

Atwater’s group is also pursuing plasmonic interconnects, which would potentially permit chip-level and board-level interconnects with bandwidths exceeding those of electrical interconnects.

Of course, plasmonics also has practical applications in the here and now. The work of Naomi Halas at Rice University (Houston, TX) using gold nanospheres to target cancer cells is one Atwater characterizes as possibly the plasmonics killer app—literally. Equipped with biological tags, the nanospheres preferentially lodge in tumors in the body. Transdermal illumination with an IR source generates surface plasmons on the spheres to heat them, in turn destroying the cells. Nanospectra Biosciences of Texas is actively pursuing FDA approvals for clinical trials of the technology.

Elsewhere, plasmonics is changing spectroscopy. In surface-enhanced Raman spectroscopy (SERS), the plasmonic effect enhances the normally weak Raman signal to the order of 106 or even 1014. Researchers such as Katrin Kneipp of Harvard University (Boston, MA) have pushed the technique with one and two-photon Raman spectroscopy, using it to detect even single molecules.

In two-photon, or hyper Raman spectroscopy, two photons interact simultaneously with a vibrational quantum state to generate a Raman signal that depends on the excitation intensity to the second power. The technique is particularly appealing for biomedical applications because the excitation wavelengths are in the near-IR spectral region, where light propagates deeper into biological materials and simultaneously lower-energy photons minimize sample damage.

Not surprisingly, the cross section of the two-photon process is even smaller than that of conventional Raman spectroscopy—on the order of 10–64 cm4s. Enter plasmonics. “If this process takes place in the enhanced local optical field around tiny gold or silver particles, the strong local optical fields enhance this very weak effect by up to 20 orders of magnitude,” says Kneipp.

“In a localized surface plasmon, there’s no propagation. All the propagating wave vectors sort of collapse and you get a standing wave around the nanoscopic object,” says SERS pioneer Richard Van Duyne of Northeastern University (Boston, MA). “The question is what portion of [the enhancement] can be accounted for by plasmonic phenomena. Both Katrin and I can agree that if not all, it’s a very, very large portion of it.”

“Basically, one could say we do Raman scattering using the help of plasmonics,” adds Kneipp. “It is Raman spectroscopy using plasmonics as a tool.”

Van Duyne is developing a SERS-based in vivo glucose sensor. Diabetics would inject the 0.1 mm2 to 1 mm2 sensors into the body with a hypodermic needle, then read them by NIR illumination through the skin. The sensors would last for months, provide continuous monitoring, if desired, and test for multiple different analytes. Van Duyne is quick to caution that the work is not likely to come to fruition for years, but his group has already demonstrated transdermal Raman scattering.

The same electric field that surrounds metal nanospheres in SERS also provides a performance boost to LEDs. “We’ve been able to enhance emission by an order of magnitude and we think there’s potential for going even one more,” says Atwater. “You can pick your favorite emitter, whether an indirect bandgap semiconductor or a dye molecule or a quantum dot. Each of those can benefit from a resonance plasmon enhancement of the local optical field.”

If the mechanism is essentially the same, then why does SERS see substantially higher enhancements? “In SERS, [the enhancement] is to the fourth power of the electric field whereas in photoluminescence enhancement, it goes as the second power,” Atwater says. In other words, optical processes that go as the electrical field to a higher power stand to get a big boost from plasmonics.

“Enhanced Raman scattering is only the most extensively and historically researched of the optical phenomena that could possibly benefit from plasmon resonance enhancement,” he notes. “Over the next couple of years, I think we’ll probably see nonlinear processes such as second-harmonic generation, parametric oscillation, and so forth.”

And that makes the future pretty exciting.

Thanks to Mark Stockman of Georgia State University for helpful discussions.

When Richard Van Duyne experimented in the 1970s with ways to increase the signal-to-noise ratio in Raman spectroscopy for the study of molecular monolayers, he never expected to play a part in the development of a whole new sub-discipline of Raman spectroscopy.

After evaluating his options, he pursued exciting an adsorbate with a laser frequency that corresponded with an electronic transition in the adsorbate/metal system. Then Martin Fleischmann’s group at the University of Southampton, UK, reported count rates of 500 to 1000 counts per second from a resonantly excited monolayer of pyridine adsorbed on a roughened silver electrode immersed in aqueous electrolyte solution. Van Duyne and his group worked to duplicate and enhance those results. What they discovered was that the smoother the electrode surface, the stronger the surface Raman signal.

Eventually, Van Duyne’s group optimized the experiment to achieve enhancement factors of 105 to 106, substantially greater than the factors of 103 or 104 normally attributable to resonance Raman spectroscopy. In addition, they couldn’t ascribe the enhancement to a resonance effect because they were exciting the pyridine at 514.5 nm, a wavelength far outside of its absorption range. Further experiments, plus work performed by Albrecht and Creighton at the University of Kent, demonstrated that it was an unknown Raman enhancement mechanism—and surface-enhanced Raman spectroscopy (SERS) was born. –K.L.