Fabrication and design of an integrable ... - Semantic Scholar
APPLIED PHYSICS LETTERS 88, 031102 共2006兲
Fabrication and design of an integrable subwavelength ultrabroadband dielectric mirror Lu Chen Department of Materials Science and Engineering, Cornell University, Ithaca, New York 14853
Michael C. Y. Huang, Carlos F. R. Mateus, and Connie J. Chang-Hasnain Department of Electrical Engineering and Computer Sciences, University of California, Berkeley, California 94720
Y. Suzukia兲 Department of Material Science and Engineering, University of California, Berkeley, California 94720
共Received 22 February 2005; accepted 21 November 2005; published online 17 January 2006兲 We have designed and fabricated a subwavelength grating 共SWG兲 broadband mirror whose performance depends on key factors, including SWG period, duty cycle, and angle of incident light. The fabricated SWGs exhibit high reflectivity 共艌96% 兲, when the grating periods are varied from 650 to 750 nm and duty cycles are varied from 55% to 65%. The bandwidth and reflectivity of these mirrors are remarkably robust to variations in design and fabrication. The SWGs can be designed as broadband mirrors from microwave to visible wavelengths. © 2006 American Institute of Physics. 关DOI: 10.1063/1.2164920兴 In optical integrated circuits, electro-optic modulators play an important role in switching and signal encoding. Ideally, electro-optic modulators should have low insertion loss and wide bandwidth. Mirrors are key components in many modulators so that a low insertion loss, broad bandwidth mirror would greatly improve the performance of these modulators. Among the candidates for mirrors are metal mirrors1,2 and dielectric mirrors.3,4 Metal mirrors have very wide reflection bands, but the absorption of light introduces a large insertion loss that limits their use in devices based on transmission. Dielectric mirrors are composed of multilayer dielectric materials with different dielectric indices, such as the distributed Bragg reflectors. These mirrors have low absorption loss, but their modulation depth, bandwidth and band location depend on the refractive index contrast of the constituent materials and material growth control of the layer thickness. In order to minimize the interface disorder and strain in the multilayer structures, typical combinations of materials often have small refractive index differences, thus resulting in rather small bandwidths 共⌬ / ⬇ 3 % – 9 % 兲. Consequently, the narrow bandwidth limits the tuning range of the electro-optic modulators such as the etalon-type devices.5–7 Recent theoretical work has reported that subwavelength gratings 共SWGs兲 exhibit a very high reflectivity 共R ⬎ 99% 兲 and broad reflection bandwidth 共⌬ / ⬎ 30% 兲.8 The structure is comprised of a grating made of materials with high index contrast sandwiched between two low index materials. The large index contrast ensures the broad bandwidth and large modulation depth 共over 10 dB兲 of the mirror. The current mature 90 nm photolithography technology in industry also provides a cheap and precise fabrication method, and it enables potential integration with other silicon-based optoelectronics components. In our previous work, we have demonstrated the first prototypical SWG broadband mirrors that were in good agreement with theoretical simulations.9 In this letter, we study the fabrication tolerance of the highly refleca兲
Electronic mail: [email protected]
0003-6951/2006/88共3兲/031102/3/$23.00
tive SWG broadband mirror—in particular the effects of geometrical variations on its reflectivity. Figure 1共a兲 shows the schematic design of the SWG structure. The structure is composed of periodic grating lines consisting of high-low refractive index material sandwiched in between two low refractive index materials on top and bottom. The index difference between the high and low index materials determines the bandwidth and modulation depth. The larger difference in refractive indices gives rise to wider reflection bands. The low index layer under the grating is critical for the broadband mirror effect. The reflection is sensitive to design parameters including the grating period, grating thickness, duty cycle, refractive index of materials and thickness of the low index layer underneath the grating.8 Duty cycle is defined as the ratio of the width of the high index material with respect to the grating period. In order to obtain high reflection across a wide bandwidth, we chose a thermally grown 580 nm silicon dioxide as the low index layer below the grating and a 400 nm polysilicon as the high index component of the grating. The low index layer above
FIG. 1. 共a兲 Schematic design of the subwavelength broadband grating reflector. 共b兲 SEM image of a fabricated SWG. 88, 031102-1
© 2006 American Institute of Physics
031102-2
Chen et al.
Appl. Phys. Lett. 88, 031102 共2006兲
FIG. 2. Experimental 共a兲 and simulation 共b兲 results of the reflectivity of SWGs with a grating period of 700 nm, poly-Si thickness of 0.4 m, PECVD oxide thickness of 100 nm and thermal oxide thickness of 580 nm with the incident polarization perpendicular to the grating lines.
FIG. 3. Experimental 共a兲 and simulation 共b兲 contour maps for the reflectivity of SWGs with a duty cycle around 65%, polysilicon thickness of 0.4 m, PECVD oxide thickness of 100 nm, and thermal oxide thickness of 580 nm with the incident polarization perpendicular to the grating lines.
the grating is simply air. Since the SWG design only contains one-dimensional symmetry from the periodic grating lines, its reflectivity is polarization sensitive 共i.e., only TM polarized light would be highly reflective兲. With these parameters as a starting point, we studied the effects of grating period, duty cycle, and incident light angle on the performance of the mirror. The SWG structures were fabricated using silicon-based nanofabrication and microfabrication techniques. In order to fabricate the structures just described, a 580 nm silicon dioxide layer and a 400 nm polysilicon layer were sequentially grown on bare silicon surfaces by thermal methods. These layers were covered with a thin layer of plasma-enhanced chemical vapor deposition 共PECVD兲 oxide that would later serve as the dry etch mask. The SWG patterns were then defined by electron beam lithography using a bilayer poly共methylmethacrylate兲 resist, followed by a lift-off process. Small windows were opened around the SWG pattern area to eliminate loading effects during the dry etch process. The defined patterns were then transferred to the PECVD oxide layer by reactive ion etching. Lastly, a high density Cl plasma was used to etch through the polysilicon layer to define the rectangular grating profile of the SWG structure using a Plasma Therm etcher. In this work, multiple SWGs were fabricated with grating periods ranging from 600 to 900 nm and duty cycles from 30% to 90%. After the fabrication process, a very thin layer 共20– 100 nm兲 of residual PECVD oxide layer was left on the top of the structure. Figure 1共b兲 shows a scanning electron microscope 共SEM兲 image of a fabricated SWG device.
The optical measurement setup includes a tungsten halogen light source, bifurcated fiber bundle, Glan-Thompson polarizer, focusing lens 关numerical aperture 共NA兲 = 0.1兴 and an optical spectrum analyzer 共OSA兲. The output of the light source is coupled into the bifurcated fiber bundle that has the common end aligned with the polarizer, focusing lens, and grating, respectively. Reflected light from the grating is collected by the bundle and coupled into the OSA. The trace is then normalized by the reflection of a commercial goldcoated mirror 共rated 艌96% reflectivity from 1.1 to 20 m兲 in order to eliminate the influence of the spectral response from both the optics and light source. In order to understand the effects of periodicity and duty cycle on the performance of the SWG structures, we performed both simulations and experiments with SWG structures over a range of periodicities and duty cycles. For a given grating period of 700 nm with duty cycles varying from 30% to 90%, we summarize the reflectivity data from experiment 共a兲 and simulation 共b兲 in a contour map, as shown in Fig. 2. Our simulation results illustrate a wide window of high reflectivity for duty cycles from 50% to 70%. Increasing or decreasing the duty cycle beyond this range reduces the reflection bandwidth. This relatively large range of duty cycles provides a large tolerance for fabrication variations, thus making these structures more appealing for applications. Despite the noise in the measurement, the basic spectral characteristics are consistent with the simulation. Figure 3共a兲 shows the measured reflectivity of SWG structures with a fixed duty cycle around 65% and a variety of periods ranging from 600 to 900 nm. The experimental results illustrate that SWG structures with periods between 650 and 750 nm exhibit very high reflectivity over a wide
031102-3
Appl. Phys. Lett. 88, 031102 共2006兲
Chen et al.
FIG. 4. 共a兲 Simulated results of SWG reflection integrated over incident angles. Curve 1 is for a collimated beam, curve 2 is for NA= 0.1, and curve 3 is for NA= 0.45. 共b兲 Optical measurements of SWG with NA= 0.1 共curve 1兲 and 0.45 共curve 2兲.
wavelength ranging from 1.1 to 1.8 m. These results suggest the robustness of the spectral characteristics to variations in structural dimensions. Simulations of these SWG structures are shown in Fig. 3共b兲 and are in excellent agreement with the experiment results in Fig. 3共a兲. Despite the qualitative agreement between our simulation and experiment results, there are a few marked differences. Deviations of experiment from simulations can be explained by the sensitivity of the experiment to the angle of incidence at longer wavelengths as well as decreased resolution of the OSA for ⬎ 1.6 m. Deviations of the incident angle from normal in the experiment accounts for the energy drop in the reflection spectra at longer wavelengths. In addition, fabrication imperfection, especially the trapezoidal grating profile resulting from the isotropic dry etching, may affect the performance of SWGs and shrink the reflection band.10 The optical characteristics of the SWG structures were remarkably robust to dimensional variations due to possible fabrication imperfections. However, the reflectivity of these mirrors also exhibits a dependence on the angle of the incident light. In our experimental setup, the focusing lens 共NA= 0.1兲 at the end of the optical path converts the collimated beam into a convergent beam whose divergence depends on the numerical aperture of the lens. Simulation results indicate that the influence of incident angle was relatively small when the incident angle was smaller than 5°. However, as the incident angle increases, the reflection decreases. Figure 4共a兲 shows the simulated reflection spectra for an SWG structure for different values of numerical aper-
ture, with the grating period of 700 nm, an underlying oxide layer of about 580 nm and a duty cycle around 65%. Within the reflection band, the reflectivity has sudden drops at = 1.13 m and = 1.44 m, and the amplitude of the drop increases as the incident angle increases. Furthermore, with the increase of the incident angle, the reflectivity over the reflection band is suppressed. When the incident angle reaches 25° 共NA= 0.45兲, the reflectivity drops about 2 dB over wavelengths longer than 1.13 m. This phenomenon is due to the change of the polarization of light when the beam passes through the focusing lens. In the SWG design, since the grating only has dielectric modulation in one dimension, the broadband reflection is only present for light with its electric field perpendicular to the grating line. After the polarized collimated beam passes through the focus lenses, the polarization of the beam is no longer perpendicular to the grating, thus giving rise to reflectivity suppression and the sudden drop in reflectivity at = 1.13 m and = 1.44 m. The measurement results of the same SWG structure are shown in Fig. 4共b兲. Experimental results showed the same trends in sudden reflectivity drops at = 1.13 m and = 1.44 m as simulation. In this letter, we have presented the influences of several key factors, including SWG period, duty cycle, and angle of incident light, on the performance of a broadband mirror based on a SWG structure. The reflectivity of the SWGs does not change significantly when the SWG period is varied from 650 to 750 nm and the duty cycle is varied from 55% to 65%. These experimental results indicate that these SWG structures have high flexibility in design and fabrication that make them ideally suited for broadband mirror applications. This work was performed in part at the Cornell NanoScale Science & Technology Facility 共a member of the National Nanofabrication Users Network兲 which is supported by the National Science Foundation under Grant ECS9731293, its users, Cornell University, and Industrial Affiliates. This work is supported primarily by the Packard Foundation, the Nanoscale Science and Engineering Initiative of the National Science Foundation under NSF Award # ECS0117770 and the New York State Office of Science, Technology & Academic Research under NYSTAR Contract No. C020071. 1
L. V. Poperenko, V. S. Voitsenya, M. V. Vinnichenko, V. G. Konovalov, O. Motojima, K. Sato, A. Sagara, and K. Tsuzuki, Surf. Coat. Technol. 114, 200 共1999兲. 2 J. N. Huiberts, R. Griessen, J. H. Rector, R. J. Wijngaarden, J. P. Dekker, D. G. de Groot, and N. J. Koeman, Nature 共London兲 380, 231 共1996兲. 3 A. Scherer, M. Walther, L. M. Schiavone, B. P. Van der Gaag, and E. D. Beebe, J. Vac. Sci. Technol. A 10, 3305 共1992兲. 4 J. M. Zanardi Ocampo, P. O. Vaccaro, T. Fleischmann, T.-S. Wang, T. Ohnishi, A. Sugimura, R. Izumoto, M. Hosoda, and S. Nashima, Appl. Phys. Lett. 83, 3647 共2003兲. 5 C. J. Chang-Hasnain, IEEE J. Sel. Top. Quantum Electron. 6, 978 共2000兲. 6 G. S. Li, W. Yuen, and C. J. Chang-Hasnain, Electron. Lett. 33, 1122 共1997兲. 7 D. Sadot and E. Boimovich, IEEE Commun. Mag. 36, 50 共1998兲. 8 C. F. R. Mateus, M. C. Y. Huang, Y. Deng, A. R. Neureuther, and C. J. Chang-Hasnain, IEEE Photonics Technol. Lett. 16, 518 共2004兲. 9 C. F. R. Mateus, M. C. Y. Huang, L. Chen, C. J. Chang-Hasnain, and Y. Suzuki, IEEE Photonics Technol. Lett. 16, 1676 共2004兲. 10 M. C. Y. Huang, K. K. Y. Lee, D. Parekh, and C. J. Chang-Hasnain 共unpublished兲.
Fabrication and design of an integrable subwavelength ultrabroadband dielectric mirror Lu Chen Department of Materials Science and Engineering, Cornell University, Ithaca, New York 14853
Michael C. Y. Huang, Carlos F. R. Mateus, and Connie J. Chang-Hasnain Department of Electrical Engineering and Computer Sciences, University of California, Berkeley, California 94720
Y. Suzukia兲 Department of Material Science and Engineering, University of California, Berkeley, California 94720
共Received 22 February 2005; accepted 21 November 2005; published online 17 January 2006兲 We have designed and fabricated a subwavelength grating 共SWG兲 broadband mirror whose performance depends on key factors, including SWG period, duty cycle, and angle of incident light. The fabricated SWGs exhibit high reflectivity 共艌96% 兲, when the grating periods are varied from 650 to 750 nm and duty cycles are varied from 55% to 65%. The bandwidth and reflectivity of these mirrors are remarkably robust to variations in design and fabrication. The SWGs can be designed as broadband mirrors from microwave to visible wavelengths. © 2006 American Institute of Physics. 关DOI: 10.1063/1.2164920兴 In optical integrated circuits, electro-optic modulators play an important role in switching and signal encoding. Ideally, electro-optic modulators should have low insertion loss and wide bandwidth. Mirrors are key components in many modulators so that a low insertion loss, broad bandwidth mirror would greatly improve the performance of these modulators. Among the candidates for mirrors are metal mirrors1,2 and dielectric mirrors.3,4 Metal mirrors have very wide reflection bands, but the absorption of light introduces a large insertion loss that limits their use in devices based on transmission. Dielectric mirrors are composed of multilayer dielectric materials with different dielectric indices, such as the distributed Bragg reflectors. These mirrors have low absorption loss, but their modulation depth, bandwidth and band location depend on the refractive index contrast of the constituent materials and material growth control of the layer thickness. In order to minimize the interface disorder and strain in the multilayer structures, typical combinations of materials often have small refractive index differences, thus resulting in rather small bandwidths 共⌬ / ⬇ 3 % – 9 % 兲. Consequently, the narrow bandwidth limits the tuning range of the electro-optic modulators such as the etalon-type devices.5–7 Recent theoretical work has reported that subwavelength gratings 共SWGs兲 exhibit a very high reflectivity 共R ⬎ 99% 兲 and broad reflection bandwidth 共⌬ / ⬎ 30% 兲.8 The structure is comprised of a grating made of materials with high index contrast sandwiched between two low index materials. The large index contrast ensures the broad bandwidth and large modulation depth 共over 10 dB兲 of the mirror. The current mature 90 nm photolithography technology in industry also provides a cheap and precise fabrication method, and it enables potential integration with other silicon-based optoelectronics components. In our previous work, we have demonstrated the first prototypical SWG broadband mirrors that were in good agreement with theoretical simulations.9 In this letter, we study the fabrication tolerance of the highly refleca兲
Electronic mail: [email protected]
0003-6951/2006/88共3兲/031102/3/$23.00
tive SWG broadband mirror—in particular the effects of geometrical variations on its reflectivity. Figure 1共a兲 shows the schematic design of the SWG structure. The structure is composed of periodic grating lines consisting of high-low refractive index material sandwiched in between two low refractive index materials on top and bottom. The index difference between the high and low index materials determines the bandwidth and modulation depth. The larger difference in refractive indices gives rise to wider reflection bands. The low index layer under the grating is critical for the broadband mirror effect. The reflection is sensitive to design parameters including the grating period, grating thickness, duty cycle, refractive index of materials and thickness of the low index layer underneath the grating.8 Duty cycle is defined as the ratio of the width of the high index material with respect to the grating period. In order to obtain high reflection across a wide bandwidth, we chose a thermally grown 580 nm silicon dioxide as the low index layer below the grating and a 400 nm polysilicon as the high index component of the grating. The low index layer above
FIG. 1. 共a兲 Schematic design of the subwavelength broadband grating reflector. 共b兲 SEM image of a fabricated SWG. 88, 031102-1
© 2006 American Institute of Physics
031102-2
Chen et al.
Appl. Phys. Lett. 88, 031102 共2006兲
FIG. 2. Experimental 共a兲 and simulation 共b兲 results of the reflectivity of SWGs with a grating period of 700 nm, poly-Si thickness of 0.4 m, PECVD oxide thickness of 100 nm and thermal oxide thickness of 580 nm with the incident polarization perpendicular to the grating lines.
FIG. 3. Experimental 共a兲 and simulation 共b兲 contour maps for the reflectivity of SWGs with a duty cycle around 65%, polysilicon thickness of 0.4 m, PECVD oxide thickness of 100 nm, and thermal oxide thickness of 580 nm with the incident polarization perpendicular to the grating lines.
the grating is simply air. Since the SWG design only contains one-dimensional symmetry from the periodic grating lines, its reflectivity is polarization sensitive 共i.e., only TM polarized light would be highly reflective兲. With these parameters as a starting point, we studied the effects of grating period, duty cycle, and incident light angle on the performance of the mirror. The SWG structures were fabricated using silicon-based nanofabrication and microfabrication techniques. In order to fabricate the structures just described, a 580 nm silicon dioxide layer and a 400 nm polysilicon layer were sequentially grown on bare silicon surfaces by thermal methods. These layers were covered with a thin layer of plasma-enhanced chemical vapor deposition 共PECVD兲 oxide that would later serve as the dry etch mask. The SWG patterns were then defined by electron beam lithography using a bilayer poly共methylmethacrylate兲 resist, followed by a lift-off process. Small windows were opened around the SWG pattern area to eliminate loading effects during the dry etch process. The defined patterns were then transferred to the PECVD oxide layer by reactive ion etching. Lastly, a high density Cl plasma was used to etch through the polysilicon layer to define the rectangular grating profile of the SWG structure using a Plasma Therm etcher. In this work, multiple SWGs were fabricated with grating periods ranging from 600 to 900 nm and duty cycles from 30% to 90%. After the fabrication process, a very thin layer 共20– 100 nm兲 of residual PECVD oxide layer was left on the top of the structure. Figure 1共b兲 shows a scanning electron microscope 共SEM兲 image of a fabricated SWG device.
The optical measurement setup includes a tungsten halogen light source, bifurcated fiber bundle, Glan-Thompson polarizer, focusing lens 关numerical aperture 共NA兲 = 0.1兴 and an optical spectrum analyzer 共OSA兲. The output of the light source is coupled into the bifurcated fiber bundle that has the common end aligned with the polarizer, focusing lens, and grating, respectively. Reflected light from the grating is collected by the bundle and coupled into the OSA. The trace is then normalized by the reflection of a commercial goldcoated mirror 共rated 艌96% reflectivity from 1.1 to 20 m兲 in order to eliminate the influence of the spectral response from both the optics and light source. In order to understand the effects of periodicity and duty cycle on the performance of the SWG structures, we performed both simulations and experiments with SWG structures over a range of periodicities and duty cycles. For a given grating period of 700 nm with duty cycles varying from 30% to 90%, we summarize the reflectivity data from experiment 共a兲 and simulation 共b兲 in a contour map, as shown in Fig. 2. Our simulation results illustrate a wide window of high reflectivity for duty cycles from 50% to 70%. Increasing or decreasing the duty cycle beyond this range reduces the reflection bandwidth. This relatively large range of duty cycles provides a large tolerance for fabrication variations, thus making these structures more appealing for applications. Despite the noise in the measurement, the basic spectral characteristics are consistent with the simulation. Figure 3共a兲 shows the measured reflectivity of SWG structures with a fixed duty cycle around 65% and a variety of periods ranging from 600 to 900 nm. The experimental results illustrate that SWG structures with periods between 650 and 750 nm exhibit very high reflectivity over a wide
031102-3
Appl. Phys. Lett. 88, 031102 共2006兲
Chen et al.
FIG. 4. 共a兲 Simulated results of SWG reflection integrated over incident angles. Curve 1 is for a collimated beam, curve 2 is for NA= 0.1, and curve 3 is for NA= 0.45. 共b兲 Optical measurements of SWG with NA= 0.1 共curve 1兲 and 0.45 共curve 2兲.
wavelength ranging from 1.1 to 1.8 m. These results suggest the robustness of the spectral characteristics to variations in structural dimensions. Simulations of these SWG structures are shown in Fig. 3共b兲 and are in excellent agreement with the experiment results in Fig. 3共a兲. Despite the qualitative agreement between our simulation and experiment results, there are a few marked differences. Deviations of experiment from simulations can be explained by the sensitivity of the experiment to the angle of incidence at longer wavelengths as well as decreased resolution of the OSA for ⬎ 1.6 m. Deviations of the incident angle from normal in the experiment accounts for the energy drop in the reflection spectra at longer wavelengths. In addition, fabrication imperfection, especially the trapezoidal grating profile resulting from the isotropic dry etching, may affect the performance of SWGs and shrink the reflection band.10 The optical characteristics of the SWG structures were remarkably robust to dimensional variations due to possible fabrication imperfections. However, the reflectivity of these mirrors also exhibits a dependence on the angle of the incident light. In our experimental setup, the focusing lens 共NA= 0.1兲 at the end of the optical path converts the collimated beam into a convergent beam whose divergence depends on the numerical aperture of the lens. Simulation results indicate that the influence of incident angle was relatively small when the incident angle was smaller than 5°. However, as the incident angle increases, the reflection decreases. Figure 4共a兲 shows the simulated reflection spectra for an SWG structure for different values of numerical aper-
ture, with the grating period of 700 nm, an underlying oxide layer of about 580 nm and a duty cycle around 65%. Within the reflection band, the reflectivity has sudden drops at = 1.13 m and = 1.44 m, and the amplitude of the drop increases as the incident angle increases. Furthermore, with the increase of the incident angle, the reflectivity over the reflection band is suppressed. When the incident angle reaches 25° 共NA= 0.45兲, the reflectivity drops about 2 dB over wavelengths longer than 1.13 m. This phenomenon is due to the change of the polarization of light when the beam passes through the focusing lens. In the SWG design, since the grating only has dielectric modulation in one dimension, the broadband reflection is only present for light with its electric field perpendicular to the grating line. After the polarized collimated beam passes through the focus lenses, the polarization of the beam is no longer perpendicular to the grating, thus giving rise to reflectivity suppression and the sudden drop in reflectivity at = 1.13 m and = 1.44 m. The measurement results of the same SWG structure are shown in Fig. 4共b兲. Experimental results showed the same trends in sudden reflectivity drops at = 1.13 m and = 1.44 m as simulation. In this letter, we have presented the influences of several key factors, including SWG period, duty cycle, and angle of incident light, on the performance of a broadband mirror based on a SWG structure. The reflectivity of the SWGs does not change significantly when the SWG period is varied from 650 to 750 nm and the duty cycle is varied from 55% to 65%. These experimental results indicate that these SWG structures have high flexibility in design and fabrication that make them ideally suited for broadband mirror applications. This work was performed in part at the Cornell NanoScale Science & Technology Facility 共a member of the National Nanofabrication Users Network兲 which is supported by the National Science Foundation under Grant ECS9731293, its users, Cornell University, and Industrial Affiliates. This work is supported primarily by the Packard Foundation, the Nanoscale Science and Engineering Initiative of the National Science Foundation under NSF Award # ECS0117770 and the New York State Office of Science, Technology & Academic Research under NYSTAR Contract No. C020071. 1
L. V. Poperenko, V. S. Voitsenya, M. V. Vinnichenko, V. G. Konovalov, O. Motojima, K. Sato, A. Sagara, and K. Tsuzuki, Surf. Coat. Technol. 114, 200 共1999兲. 2 J. N. Huiberts, R. Griessen, J. H. Rector, R. J. Wijngaarden, J. P. Dekker, D. G. de Groot, and N. J. Koeman, Nature 共London兲 380, 231 共1996兲. 3 A. Scherer, M. Walther, L. M. Schiavone, B. P. Van der Gaag, and E. D. Beebe, J. Vac. Sci. Technol. A 10, 3305 共1992兲. 4 J. M. Zanardi Ocampo, P. O. Vaccaro, T. Fleischmann, T.-S. Wang, T. Ohnishi, A. Sugimura, R. Izumoto, M. Hosoda, and S. Nashima, Appl. Phys. Lett. 83, 3647 共2003兲. 5 C. J. Chang-Hasnain, IEEE J. Sel. Top. Quantum Electron. 6, 978 共2000兲. 6 G. S. Li, W. Yuen, and C. J. Chang-Hasnain, Electron. Lett. 33, 1122 共1997兲. 7 D. Sadot and E. Boimovich, IEEE Commun. Mag. 36, 50 共1998兲. 8 C. F. R. Mateus, M. C. Y. Huang, Y. Deng, A. R. Neureuther, and C. J. Chang-Hasnain, IEEE Photonics Technol. Lett. 16, 518 共2004兲. 9 C. F. R. Mateus, M. C. Y. Huang, L. Chen, C. J. Chang-Hasnain, and Y. Suzuki, IEEE Photonics Technol. Lett. 16, 1676 共2004兲. 10 M. C. Y. Huang, K. K. Y. Lee, D. Parekh, and C. J. Chang-Hasnain 共unpublished兲.
