We have demonstrated a very large enhancement of the transmission of light through sub-wavelength apertures in metallic films when the metal surface surrounding the aperture is corrugated. Through coupling to surface plasmons in the metal surface, resonant enhancement of the electromagnetic fields near the aperture results in transmission far higher than that obtained for a bare aperture. Transmission as large as T/f~3 has been observed, where f is the cross sectional area of the aperture with diameter d; i.e. three times more light is transmitted than is directly incident on the aperture, despite the fact that for d << l the transmission through the aperture is evanescent. The wavelength of the resonance may be tuned by the design of the corrugation geometry, as well as the index of refraction of the adjacent medium. For this reason, plasmon-enhanced devices are highly promising for use as NSOM probes and for high-density optical data storage.
In near-field scanning optical microscopy (NSOM), the probe resolution is determined largely by the diameter of an aperture, typically approximately a few tens of nanometers. However, in the tapered optical fibers often used for NSOM applications, the single mode fiber ceases to function as a single mode waveguide when the tip diameter decreases below the nominal core diameter. This results in multimode performance, and much of the light is lost in the taper before reaching the probe aperture.
The throughput of a subwavelength aperture in even a flat metal film is extremely low: Classical diffraction theory (Bethe, 1944; Bouwkamp, 1954) predicts that the transmission of a electromagnetic radiation through an aperture, of radius d, in a perfectly conducting plane obstacle should be given by,
When the radius is smaller than the wavelength of the light (d << l), the transmission of energy through the aperture falls as (d/l)4. This analysis assumes the transmission is measured far from the aperture (far field diffraction). Far-field diffraction theory neglects those terms in the radiation field that fall off faster than quadratic in the inverse distance from the aperture.
For instance, part of the transmitted energy emerges in an evanescent wave that travels along the reflecting interface between the media. It is possible to couple to this wave, despite the fact that its intensity diminishes exponentially with distance from the interface, and this allows one to create surface plasmons in thin metal films (Raether, 1988).
However, a comprehensive theory for the near field radiation distribution does not exist (especially for thick partially-absorbing barriers.) A numerical analysis (Leviatan, 1986) has demonstrated that the transmitted wavefronts may be viewed as approximately collimated within a distance of about one diameter of the aperture (z~d). Thus, it is possible to employ the radiation field near the aperture (near-field) to study optical emission from surfaces with high resolution, although with very low optical power throughput. This may be seen in Figure 1, which compares the measured transmission of several types of NSOM probes. The transmission of such probes is often less than that expected for purely diffraction-limited optics, presumably because real metals have finite conductivity and thus require finite thickness in order to be opaque.
Figure 1. Power throughput as a function of aperture diameter d normalized to wavelength l, comparing surface plasmon enhanced (solid squares) and "bare" (open squares) apertures in a free-standing metal membrane, microwave emission at l=118µm (solid circles) and tapered optic fiber tips (open circles). Also shown are the Bethe theory (solid line) and the geometrical filling fraction f (dashed line), assuming incident spot size has a diameter d=5l.
We have demonstrated a very large enhancement of the transmission of light through sub-wavelength apertures in metallic films when the metal surface surrounding the aperture is corrugated (Grupp et. al., 1999; Ebbesen et. al., 1998; Ghaemi et. al., 1998). Transmission as large as T/pd2~3 have been observed, i.e. three times more light is transmitted than is directly incident on the aperture, despite the fact that for d<<l the transmission through the aperture is evanescent. The wavelength of the resonance may be tuned by the design of the corrugation geometry, as well as the index of refraction of the adjacent medium. For this reason, plasmon-enhanced devices are highly promising for use as NSOM probes. The transmission of radiation through such holes is enhanced by the coupling of the incident light to surface plasmons--the creation of Surface Plasmon-Polaritons. If the coupling is resonant (i.e. the wavelength of the light matches the wavelength of the surface plasmons), the electric fields are enhanced, resulting in enhanced transmission. (Martín-Moreno, et. al. 2000; Schröter and Heitmann 1998).
Plasmons (plasma oscillations, or charge density waves) in a metal are vibrational modes of the electron gas density oscillating about the metallic ion cores. Surface plasmons describe the special case in which the charges are confined to the surface of the metal. In this case, the electric field is strongest in the plane of the metallic surface (Figure 2). Plasmons confined to a plane do not radiate light (Raether, 1988); however, when the local planar symmetry is disturbed, plasmons can radiate (Smolyaninov et. al. 1997; Sanchez-Gil and A. A. Maradudin, 1999.) Surface plasmon-mediated emission from defects in metal surfaces has been observed (for example, by Hecht et. al., 1992.) Such features have been described as "plasmon flashlights", and represent the localization of photon emission in a region generally experiencing non-radiative, collective oscillation of the surface electron density. It is possible to design structures that take advantage of this emission of surface plasmon radiation. Surface plasmons may be used to enhance radiative processes, providing bright light sources with a size less than the wavelength of the emitted light. The enhancement may also be used to provide highly efficient collectors of light. These so-called nanoscopic emitters (or collectors) may prove useful when integrated into a nanoscale assembly.
Figure 2. Surface plasmons on a smooth metal surface do not radiate, as the dispersion curve for the plasmons lies below the "light line" (Figure 2a). The oscillations in the surface charge density are illustrated in Figure 3B, along with the electric field lines; the electric field intensity decays exponentially with distance from the surface.
A periodic pattern of surface topography provides grating coupling of the incident radiation to the surface plasmon modes. When energy and momentum (including grating momentum) conservation are obeyed, a resonant enhancement of the electric fields around the hole leads to a transmitted intensity far in excess of what would be found in the absence of the enhanced fields. (Martin-Moreno et. al, 2000) An example of such a structure is given in Figure 3, which depicts a rectangular array of dimples surrounding an aperture in a sputtered silver film.
Figure 3. Focused ion beam (FIB) image of a single aperture in a free-standing metal membrane, surrounded by a square array of dimples.
For normal incidence onto a surface structure with square symmetry the resonant wavelength is given by
where the effective refractive index, neff, (see Eqn. 3) is determined by em anded, the dielectric constants of the metal, and dielectric, respectively (Ghaemi et. al., 1998), and P is the periodicity of the surface corrugation, and (i,j) are integers. Thus, this effect is tunable by altering the period P, or the material characteristics (em, ed).
For arrays such as that depicted in Figure 3, several resonances are observed for the various values of (i,j). However, it is the longest-wavelength resonance (i,j)=(1,0) which yields the largest transmission enhancement, because at that wavelength (l>P) no diffraction occurs off the periodically corrugated surface. For circularly symmetric structures which are periodic in the radial direction, there is only one relevant integer: The ring corrugation structure restricts the resonance condition for plasmon enhanced radiation to harmonics of a single spatial wavelength (j=0 in Eq. 2). An actual sample, fabricated by focused ion beam (FIB) milling, is shown in Figure 4.
Figure 4. SEM image of an aperture surrounded by grooves with circular symmetry. The aperture diameter, d=400nm. The grooves have spacing P=750nm, and width W=375nm. The structure was fabricated by focused ion beam (FIB) milling in a free-standing metal film, with silver (Ag) overcoating on both top and bottom surfaces. The grain structure in the silver film is clearly visible in this SEM image.
Two distinct surface plasmon modes are apparent in the transmission spectrum of such a device (Figure 5), due to level splitting of the surface plasmon modes. (Thio et. al., 2001) The difference between the modes is depicted in Figure 6, which shows that the modulation of the surface charge density is concentrated in the in the corrugation ridges and valleys (Figure 6b), or the top surface of the metal near the edges of the corrugation (Figure 6a). The peak electric fields at the input of the aperture result in a larger transmission through the aperture than would occur in the absence of the surface plasmon resonance.
Figure 5. Transmission spectrum of a ring structure with sinusoidal corrugation. Two distinct SP modes are observed at 760nm (Mode 1) and 800nm (Mode 2). The peak transmission is T/pd2~3. The dashed line indicates the transmission of a bare aperture which does not show the SP enhancement.
Figure 6. Two SP modes around gap in dispersion relation, corresponding to double peak structure in Figure 5. The two SP modes (Thio et. al., 2001) differ in the distribution of surface charge displacement and therefore in the pattern of local electric field enhancement. "Hot spots" indicate regions of peak electric field intensity.
The optimal surface corrugation depth is about 90nm, or a few times the skin depth of the metal at this wavelength. The width of the grooves is also optimal at W=P/2. In a series of single holes with varying aperture size and ring spacing, it is found that T/pd2 ~ 3 is independent of aperture diameter (see Figure 7).
Figure 7. Transmission, normalized to aperture area at the peak wavelength l=800nm, for a ring structure with P=750nm, W=375nm and h=150nm (solid squares) is almost independent of aperture diameter, in contrast to "bare" apertures of which the normalized transmission drops rapidly with decreasing diameter (open squares).
Figure 8. The transmission enhancement factor, the ratio of the enhanced transmission (at the peak wavelength) to that of a similar bare aperture (from data of Figure 7), is a measure of the relative enhancement due to the surface plasmon coupling. While the throughput (not normalized to aperture area) decreases with d, FSP increases dramatically, reaching values greater than 100 at the length scales commonly used for NSOM applications, <100nm.
The classical diffraction limit, d ~ l/2, was once thought to limit the design of optical devices. Using the surface plasmon enhancement, it is possible to obtain high throughput for apertures on the sub-wavelength, or nanometer scale. Any application for high brightness, high resolution optical processing may benefit from the techniques of "surface plasmon enhancement".
High Density Optical Data Storage
Traditional approaches to increasing the density of optical data storage devices involve working at the diffraction limit of light while using far-field radiation patterns. For instance, by moving to shorter (bluer) wavelengths, and fabricating microoptical devices it is possible to force the data storage density higher in the classical optics regime (see for example, Fasol, 1996; Kawase, et. al., 2000.) The use of shorter wavelengths introduces a new class of materials-related difficulties such as photobleaching effects, and thermal degradation of the Kerr effect (Sakamoto, et. al., 2000). However, surface plasmon enhancement makes it possible to operate in the near-field with plasmon-enhanced emitters, while using well-established and relatively inexpensive source technology (visible-near IR), and nanometer-scale lithography to design bright, high-resolution emitters and collectors.
High-Throughput NSOM Probes
An obvious application for tunable plasmon enhanced transmission is in Near Field Scanning Optical Microscopy (NSOM). The extremely low transmission of traditional NSOM probes makes observation of lowefficiency processes such as fast dynamical processes, or non-linear optical effects, very difficult. Non-linear optical studies (second-order surface Raman scattering, for instance) often involve processes which require high intensity sources, and are intrinsically weak effects. Plasmon enhanced NSOM probes are capable of probing nonlinear processes at surfaces by acting as high brightness emitters. When used as collectors, even standard NSOM applications may benefit from the higher throughput probes, which at the very least will improve data acquisition times.
By structurally varying the surface of a metal-dielectric interface, it is possible to use a resonance of the incident light with surface plasmon polaritons to enhance the optical transmission through apertures that are much smaller than the wavelength. Enhancement factors larger than 100 have been observed; the enhancement factor grows with decreasing aperture diameter and increasing wavelength. Surface plasmon enhanced devices thus hold an immense potential for use in applications where both high throughput and high resolution are required, without the need to use wavelengths shorter than those which can be provided by conventional light sources, such as visible and near-IR diode lasers. Two possible applications are high-density optical data storage, and near-field optical microcopy.
Bethe, H. A. 1944 Phys. Rev.66, 163
Bouwkamp, C. J. 1954 Rep. Progress Phys.17, 35
Ebbesen, T. W., Lezec, H. J., Ghaemi, H. F., Thio, T. and Wolff, P. A. 1998 Nature391, 667
Fasol, G. 1996 Science272, 1751
Ghaemi, H. F., Thio, T., Grupp, D. E., Ebbesen, T. W. and Lezec, H. J. 1998 Phys. Rev. B58, 6779
Grupp, D. E., Lezec, H. J., Thio, T. and Ebbesen, T. W. 1999 Adv. Mater. 11, 860
Hecht, B. Sick, B., Wild, U. P., Deckert, V., Zenobi, R., Martin, O. J. F. and Pohl., D. W. 2000 J. Chem. Phys.112, 7767
Kawase, K., Muto, Y., Yamaguchi, K., Ando, N., Maeda, Y., Yamada, M., Kanno, M. 2000 Optical Data Storage, 2000 Conference Digest, 173
Leviatan, Y. 1986 J. Appl. Phys.60, 1577
Martín-Moreno, L., García-Vidal, F. J., Lezec, H. J., Pellerin, K. M., Thio, T., Pendry, J. B. and Ebbesen, T. W. 2000 Phys. Rev. Lett.86, 1114
Matsumoto, T., Ohtsu, M., Matsuda, K., Saiki, T., Saito, H., and Nishi, K. 1999 Appl. Phys. Lett.75, 3246
Michaelis, J., Hettich, C., Mlynek, J., and Sandoghar, V. 2000 Nature405, 325
H. Raether, H. 1988. Surface Plasmons on Smooth and Rough Surfaces and on Gratings, Springer Tracts in Modern Physics Vol.111, Springer-Verlag, Berlin, 1988
Sakamoto, T., Fujita, G., Takemoto, Y., Akiyama, Y., Imanishi, S., Aratani, K., Kanno, M., and Nakaoki, A. 2000 Optical Data Storage, 2000 Conference Digest, 89
Sanchez-Gil, J. A. and Maradudin, A. A. 1999 Phys. Rev. B60, 8359
Smolyaninov, I., Mazzoni, D. L., and Davis, C. C. 1997 Phys. Rev. Lett. 78, 2823
Schröter, U., and Heitmann, D. 1998 Phys. Rev. B 58, 15419
Thio, T., Pellerin, K. M., Linke, R. A., Lezec, H. J. and Ebbesen, T. W. 2001 Optics Lett. 26, 1972