IMEC’s “noble” efforts to integrate nanophotonics with hi-speed CMOS
07/01/2009
Plasmonic technology, today still in an experimental stage, has the potential to be used in future applications such as nanoscale optical interconnects for high-performance computer chips, extremely sensitive (bio)molecular sensors, and highly efficient thin-film solar cells. IMEC recently reported a method to integrate high-speed CMOS electronics and nanophotonic circuitry based on plasmonic effects.
Interest in the photonic properties of nanostructured (noble) metals has accelerated in the last 10 years because they “show great promise for use in nanophotonic applications,” according to Pol Van Dorpe, senior researcher in plasmonics at IMEC. In an optics context, metals are typically seen as either reflective or absorbing materials; in an optics context that’s true for most metals and in many circumstances, he explained, but metals with optical properties that can be described by the Drude model (i.e., whose electronic properties resemble a free electron gas, only weakly bound to the metal ions) support charge density waves (surface plasmons) that can couple to visible electromagnetic radiation.
There are a number of interesting features concerning surface plasmons, he noted. First of all, their wavelength can be reduced significantly below the free space wavelength (which has the same frequency). “The reduction of the wavelength can result in a strongly improved degree of confinement, which allows deep sub-wavelength waveguides whose sizes can match state-of-the-art transistors,” explained Van Dorpe. “A number of designs allow such strong confinement.” One of the more advanced designs, which is explained in IMEC’s paper in the May issue of Nature Photonics, consists of a metal/dielectric/metal layered structure (a MIM waveguide).
Second, surface plasmons can be focused in small holes or slits, giving rise to extraordinary transmission through deep sub-wavelength holes. “This property is also particularly interesting to reduce the noise and the capacitance of photodetectors, as light can be captured on a metal film (i.e., converted to surface plasmons) by the appropriate gratings and focused in a deep sub-wavelength aperture of slit that connects to a semiconductor,” noted Van Dorpe. Ultrafast photodetectors could thus be constructed, with a small semiconductor active area resulting in limited capacitance and low noise, “without sacrificing the total signal to noise.”
Moreover, when patterned into deep sub-wavelength nanostructures (or nanoparticles), noble metals respond in-phase to the exciting electromagnetic field. Depending on the shape, the specific metal used (Au, Ag, Cu, Al, etc.) and the dielectric surrounding the metal, the polarizability of the metal nanostructures shows a resonant behavior in the visible. “This goes hand in hand with strongly enhanced local electric fields, absorption and/or scattering,” said Van Dorpe. Local surface plasmon resonance (LSPR) properties of metal nanoparticles have applications in several areas, he pointed out, including biosensing (shifts in the resonant wavelengths upon molecular binding events); surface enhanced Raman scattering (utilizing the enhanced local electric fields); and cancer treatment (local heating of cancer cells by labeling them with metal nanoparticles, and irradiating with near-infrared light).
The technology also has application in solar cell enhancements. Reducing the thickness of solar cells not only promises to lower material costs and therefore the intrinsic cost of solar cells, but it also results in efficiency reductions. “There are a number of ways that plasmonic effects can be used to boost the efficiencies of thin-film solar cells,” notes Van Dorpe. “Most importantly, the strong scattering of metal nanoparticles can result in a significant enhancement of the optical path inside the photo-absorbing material, allowing a strong absorption enhancement for near-bandgap photons that don’t get completely absorbed otherwise.” This situation is typically realized by surface texturing for thicker solar cells, he noted, but as the texture features are in the range of several micrometers, this strategy obviously breaks down for thin-film solar cells whose thickness is in the same range. “Secondly, the enhanced local fields of plasmon supporting metal nanoparticles can result in an enhanced optical absorption of light in organic semiconductors,” he added.
Looking to the future, nanoscale plasmonic circuits could allow massive parallel routing of optical information on ICs—but eventually that high-bandwidth optical information has to be converted to electrical signals, notes Van Dorpe. ICs combining high-speed CMOS electronics and plasmonic circuitry will require efficient and fast interfacing components that couple the signals from plasmon waveguides to electrical devices.
As an important stepping-stone to such components, IMEC has now demonstrated integrated electrical detection of highly confined short-wavelength surface plasmon polaritons in metal-dielectric-metal plasmon waveguides. The detection was done by embedding a photodetector in a metal plasmon waveguide. Because the waveguide and the photodetector have the same nanoscale dimensions, there is an efficient coupling of the surface plasmons into the photodetector and an ultrafast response.
Numerous IMEC experiments have “unambiguously demonstrate[d] this electrical detection,” explained Van Dorpe. “The strong measured polarization dependence, the experimentally obtained influence of the waveguide length, and the measured spectral response are all in line with theoretical expectations, obtained from finite element and finite-difference-time-domain calculations.” These results pave the way for the integration of nanoscale plasmonic circuitry and high-speed electronics, he added.
The highly confined plasmonic waveguide described in IMEC’s Nature Photonics paper is only the final part of the optical chain, consisting from less confined, but long-range waveguides (either dielectric waveguides or long-range surface plasmon waveguides) that eventually couple to nanosized waveguides that connect to individual devices. “More work is necessary to build efficient optical couplers between the different types of waveguides,” noted Van Dorpe.
The device IMEC fabricated and described is a proof-of-principle device, making use of GaAs as the photodetector material. “Similar devices can be constructed from, as an example, (Si)Ge, which is better suited for telecom wavelengths,” said Van Dorpe, “and we are currently working on measuring the ultimate speed limits of such photodetectors.” Because plasmonic circuitry also relies on optical sources, IMEC is also working on building fast, nanosized optical sources that efficiently couple to surface plasmon waveguides. “The fast-interfacing components on the detector side is something that can be realized rather quickly, in the first couple of years, while the latter part [integrated sources] probably will take a longer time, i.e., 5-10 years,” he said. —D.V.