* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Bandgap-assisted surface-plasmon sensing Arnaud J. Benahmed* and Chih-Ming Ho
Optical amplifier wikipedia , lookup
Birefringence wikipedia , lookup
Vibrational analysis with scanning probe microscopy wikipedia , lookup
Optical aberration wikipedia , lookup
Magnetic circular dichroism wikipedia , lookup
Diffraction topography wikipedia , lookup
Ellipsometry wikipedia , lookup
Ultraviolet–visible spectroscopy wikipedia , lookup
Silicon photonics wikipedia , lookup
3D optical data storage wikipedia , lookup
Gaseous detection device wikipedia , lookup
Dispersion staining wikipedia , lookup
Nonimaging optics wikipedia , lookup
Phase-contrast X-ray imaging wikipedia , lookup
Laser beam profiler wikipedia , lookup
Optical flat wikipedia , lookup
Optical coherence tomography wikipedia , lookup
Harold Hopkins (physicist) wikipedia , lookup
Rutherford backscattering spectrometry wikipedia , lookup
Diffraction grating wikipedia , lookup
Photon scanning microscopy wikipedia , lookup
Nonlinear optics wikipedia , lookup
Optical tweezers wikipedia , lookup
Retroreflector wikipedia , lookup
Bandgap-assisted surface-plasmon sensing Arnaud J. Benahmed* and Chih-Ming Ho Department of Mechanical Engineering, University of California, Los Angeles, 420 Westwood Plaza, Los Angeles, California 90095 *Corresponding author: [email protected] Received 1 September 2006; revised 30 January 2007; accepted 30 January 2007; posted 30 January 2007 (Doc. ID 74665); published 15 May 2007 Surface-plasmon resonance (SPR) is a sensing technique widely used for its label-free feature. However, its sensitivity is contingent on the divergence angle of the excitation beam. The problem becomes pronounced for compact systems when a low-cost LED is used as the light source. When the Kretschmann configuration with a periodically modulated surface is used, a bandgap appears in the surface plasmon dispersion relation. We recognize that the high density of modes on the edge of the surface-plasmon bandgap permits the coupling of a wider range of incidence angles of excitation photons to surfaceplasmon polaritons than what is possible in the traditional Kretschmann configuration. Here, the numerical simulation illustrates that the sensitivity, detection limit, and reflectivity minimum of an amplitude-based SPR bandgap-assisted surface-plasmon sensor are almost independent of the divergence angle. Two different bandgap structures are compared with the Kretschmann configuration using the rigorous coupled-wave analysis technique. The results indicate that the bandgap-assisted sensing outperforms traditional SPR sensing when the angular standard deviation of the excitation beam is above 1°. © 2007 Optical Society of America OCIS codes: 000.3110, 120.4640, 120.5700, 130.6010, 240.6680, 260.6970. 1. Introduction A. Surface-Plasmon Sensing Surface-plasmon waves (SPWs) are electromagnetic waves propagating on the surface of a plasma and created by fluctuations of the surface charge density [1]. At optical frequencies, SPWs can be observed on the surface of metals such as gold and silver. SPWs are extremely sensitive to any changes of optical index on the interface and are therefore used in a wide range of surface-sensing applications [2,3], the most common being called surface-plasmon resonance (SPR). On a flat surface, the SPW dispersion relation has the following analytical form, with kSP being the complex SPW wavenumber, k0 being the wavenumber of light propagating at the same frequency in vacuum, and ⑀2 and ⑀3 being the complex relative permittivities of the metal and the dielectric, respectively [1]: 0003-6935/07/163369-07$15.00/0 © 2007 Optical Society of America 冑 kSP ⫽ k0 ⑀2⑀3 . ⑀2 ⫹ ⑀3 (1) SP sensing is the monitoring of ⑀3 by measuring kSP. In the SPR technique, the coupling between TM polarized light and SPW is used to measure kSP using the momentum-matching condition at the resonance. A convenient technique is to use the Kretschmann configuration [4], which consists of a high-index prism with a thin layer of gold or silver deposited on one face. TM-polarized light is sent inside the prism where it internally reflects at the metal– dielectric interface. When the transverse momentum of the incident beam kx is equal to kSP, the energy of the incident light is transferred to SPW and the specular reflectivity becomes zero, if the metal thickness is well chosen. This phenomenon can be used for sensing in several ways, using the measure of the resonance angle [3,5], the resonance wavelength [6], or the phase [7]. The easiest technique to implement in a compact format is the amplitude technique [8], which can be described as follows: When 冑⑀3 ⫽ n3 changes while 1 June 2007 兾 Vol. 46, No. 16 兾 APPLIED OPTICS 3369 Fig. 1. Reflectivity on each structure for a Gaussian beam of divergence 0° and 2° at 633 nm and with a central angle of incidence of 51°. the incident angle, i, and the wavelength, , are kept fixed at some well-chosen values, the reflectivity, R共n3兲, presents a dip whose minimum corresponds to the momentum-matching condition kSP共n3兲 ⫽ kx. In the range of n3 where the slope of R is maximal, it is possible to measure n3 by monitoring the reflectivity. The two important characteristics of the reflectivity dip are its slope [9], which defines the sensitivity ⭸R兾⭸n, and the value of the reflectivity minimum, which has an effect on the detection limit. Villatoro and Garcia-Valenzuela [10] pointed out that these two parameters are strongly dependent on the quality of the excitation beam. Particularly, for a Gaussian beam with a small waist, 0, the slope becomes lower and the minimum of reflectivity is higher. For example, when the beam waist is 32 m at 633 nm, which corresponds to a 0.2° divergence, the reflectivity minimum is 30 times larger and the sensitivity is 65% less than a perfect beam [10]. The reflectivity plots for zero divergence 共 ⫽ 0兲 SPR in Fig. 1 illustrate this effect. Such a situation occurs when the light source has a small aperture or the optical system cannot achieve perfect collimation. Also, the beam can be decollimated for focused measurements in a very precise location such as a spot in DNA or protein microarray. To remedy this problem, we propose to take advantage of the increase of the SP density of modes on the edge of the surface-plasmon bandgaps (SPBG) to decrease the influence of the beam divergence on the signal. B. Surface-Plasmon Bandgap When a SPW propagates through a periodic structure, its dispersion relation is modified [11,12]. Notably, when the periodicity of the structure, g (See Fig. 2), is equal to half of the periodicity of the SPW, the SPW is Bragg reflected, and the propagation of the wave through the structure is blocked [13,14]. This phenomenon, which is similar to a photon in a photonic crystal [15], creates energy gaps in the dispersion relation of the surface plasmon. This type of structure is of particular use in the field of plasmonics, 3370 APPLIED OPTICS 兾 Vol. 46, No. 16 兾 1 June 2007 Fig. 2. Schematic of the different structures studied. where they are used as SPW mirrors and waveguides for integrated plasmonic circuits [16,17]. This phenomenon has been analytically explained and described by Barnes et al. [14] and by Kitson et al. [18] in the case of a corrugated surface on a conical mount. There are different methods for defining the surface-plasmon density of modes in one dimension. A simple definition for the density of modes 共兲 can be written as 共兲 ⫽ dkSP . d (2) The SPW dispersion is flat on the edge of the bandgap. Therefore, according to Eq. (2), the density of modes will be increased on the band edge. This is a recurrent phenomenon in most bandgap situations. In this paper, we propose to take advantage of this phenomenon to reduce the SPR signal decay when the beam divergence increases. The principle is as follows: the system is first prepared so that an excitation beam excites SPWs on the edge of the bandgap (see Fig. 3). Since the density of modes is large at this point, the coupling between a larger range of incident Fig. 3. Illustration of the concept of the sensor. On the band edge, a wider range of excitation angles can be coupled to SPWs because the density of modes is higher. transverse momenta and SPW modes is possible. Even if the divergence of the incoming beam is large, the reflectivity will be low at this point. When the optical index of the medium changes, so does the gap position causing the excitation beam to no longer be located on the band edge. The corresponding reflectivity increase and change can be used for sensing. In the next sections, we describe how we simulated such a system and derive the performances of this SPBG sensor. 2. Numerical Simulations We numerically investigated the SPBG sensor by simulating three different types of structure. The simulation technique consisted of two steps. First, a map of the reflectivity of the structures for plane waves was calculated. Then, the reflectivity for a Gaussian beam was computed by integrating the plane-wave reflectivity over a range of angles of incidence. A. Simulated Structures In two dimensions, periodic structures modify the SP dispersion relation by creating new branches that correspond to ⫾kSP共兲 shifted m times by the reciprocal periodicity of the grating, kg, where m is an integer. Each branch is indexed by its corresponding m and the sign of the direction of propagation in the superscript. For example, the branch that corresponds to the SPW propagation on a flat surface in the forward direction is indexed 0⫹. Bandgaps appear at branch crossings. In the past, SP bandgaps have been observed at normal incidence at the crossing of the ⫹1⫺ and ⫺1⫹ branches. In this situation, the first order of the grating is used for momentum matching between the photon and the SPW, while the second order is used for creation of the surface plasmon bandgap at kx ⫽ 0 [14]. The main advantage of this technique is that no prism is necessary since momentum matching is ensured by the grating. Since it is difficult to precisely control the second order of corrugation of the structure at optical scale, we instead use in this paper the gap appearing at the 0⫹ and 1⫺ branch intersections, which is controlled by the the first order of the grating. However, in order to probe this gap by the SPR technique, a prism needs to be used to increase the excitation light transverse momentum. As illustrated in Fig. 2, we simulated three different types of structure. The first, further referred to as (a), corresponds to the classic Kretschmann configuration. It is a thin 共39 nm兲 layer of silver on top of a glass prism 共n1 ⫽ 冑⑀1 ⫽ 1.5兲. The values for the permittivity 共⑀2兲 of silver have been linearly extrapolated at each wavelength from tables [19]. We chose the silver thickness in order to minimize the reflectivity minimum at 600 nm. The second structure (b) is a direct modification of the Kretschmann configuration, where a square-shaped dielectric layer is patterned on the metallic layer. We chose nd ⫽ 冑⑀d ⫽ 1.5 as the optical index of the dielectric layer and a layer thickness of 200 nm, so that the full extent of the surface-plasmon field is perturbed by the grating. We also chose a filling ratio of 0.5 to maximize the size of the bandgap. The last structure (c) that we simulated was a uniform thin layer of silver 共39 nm兲 on top of a sinusoidally corrugated prism. The corrugation height was 30 nm, so that the bandgap was large enough to be observed. It is also a convenient structure to fabricate in an optical lithography setup. In the last two devices, the pitch of the grating was chosen so that the bandgap was observable at ⬃600 nm. The position of the bandgap can be approximated by the relationship: kg ⫽ ᑬ共k̃SP兲. 2 (3) In this equation, k̃SP is the wavenumber of the SPW that would propagate through the grating without the Bragg reflection, and ᑬ is the real part operator. In the assumption that k̃SP ⫽ kSP at 600 nm and for ⑀3 ⫽ 1, we obtain 冉冑 冊 kg 2 ᑬ ⫽ ᑬ共kSP兲 ⫽ 2 0 2 ⑀2⑀3 ⬇ 1.0378 . ⑀2 ⫹ ⑀3 0 (4) This leads to a pitch, g, of 280 nm. The assumption is that k̃SP ⫽ kSP is reasonable in structure (c) but excessive in structure (b). In this case, an effective optical index should be used instead of ⑀3. We observed that a simple weighted average between ⑀3 and ⑀d leads to reasonable results. For these three structures, the reflectivity maps, R共, i兲, were calculated in the range 关400–800 nm兴 and 关40°–70°兴 and by varying the optical index n3 in the range 关1–1.2兴. B. Reflectivity Calculations In structure (a), which corresponds to the classic Kretschmann configuration, we calculated the reflec1 June 2007 兾 Vol. 46, No. 16 兾 APPLIED OPTICS 3371 tivity of the interface using the Fresnel coefficients [10]. To calculate the reflectivity of structures (b) and (c), we implemented a code based on rigorous coupled wave analysis (RCWA). This technique is well known [20 –22] and will not be described in further detail here. RCWA simulates the transmission and the reflectivity for a periodic stack of square-shaped dielectric layers. Structure (b) can be exactly simulated with a four-layer system. To simulate structure (c), we had to discretize the sinusoidal shape by approximating it with a large number of flat dielectric slabs [see Fig. 2(c)]. RCWA is notoriously known to have convergence problems when simulating sinusoidal cases [23,24]. In our particular simulation, we did obtain convergence for a 30-layer approximation and 61-wave decomposition. C. Calculation of the Reflectivity of a Gaussian Beam Once we calculated the reflectivity, R共, 兲, for each wave, we used the technique described by Villatoro and Garcia-Valenzuela [10] to calculate the intensity of the reflected beam. The Gaussian beam can be expressed as a continuum of plane waves traveling in different directions. The angular intensity follows a normal distribution around the average direction, i. In the approximation of low angular divergence and by assuming that the changes of reflectivity in the polar angle of incidence are neglected, the reflectivity of a Gaussian beam with a central angle of incidence i and angular divergence can be calculated by RGauss共, i兲 ⫽ 1 冑2 冕 ⫹⬁ ⫺⬁ 冋 R共, 兲exp ⫺ 共 ⫺ i兲2 22 册 d. (5) This integral was computed using a simple Riemann approximation technique in the range of 关i ⫺ 3, i ⫹ 3兴. Divergence can be related to the diameter of beam waist w0 by the relationship ⫽ 1兾w0k1. 3. Results A. Reflectivity Maps Figure 4 represents the results obtained for the calculation of the reflectivity for each structure for n3 ⫽ 1 and n3 ⫽ 1.2. In these maps, the reflectivity minima (dark areas) represent conditions for which the incident beam excites SPW and is absorbed. The center of these minima corresponds to the SP dispersion relation. As expected, the dip of reflectivity shifts to a higher angle of incidence when the optical index of the medium, n3, increases, which corresponds to the increase in k̃SP. The value of the resonant angle of incidence (corresponding to kx ⫽ k̃SP) is higher for structure (b) than for structures (a) or (c). This is to be expected since half of the SPW is in a region of high optical index, nd. However, the resonant angle for structure (c) is very close to that for normal SPR [structure (a)]. This means that the corrugation only 3372 APPLIED OPTICS 兾 Vol. 46, No. 16 兾 1 June 2007 Fig. 4. Reflectivity maps for each of the structures. The left column corresponds to n3 ⫽ 1 and the right n3 ⫽ 1.2. On all the figures, the arrow represents the position where the signal was monitored in the further sensing results. The black curves in (b) and (c) correspond to kx ⫽ kg兾2. The white dashed curve in (b1) corresponds to k̃SP (see text). The white dashed curve in (c1) corresponds to the SP excitation on the glass–metal interface. weakly perturbs the SP propagation outside the gap, and our earlier assumption k̃SP ⫽ kSP is valid in this case. For structures (b) and (c), the locus corresponding to kx ⫽ k̃SP can be visualized as the prolongation of the 0⫹ SP resonance branch through the gap [dashed curve in Fig. 4(b1)]. In Figs. 4(b1), 4(b2) and 4(c1), 4(c2), the SP bandgap appearing at ⬃600 nm for (b) and 550 nm for (c) can clearly be seen. As explained in the previous section, the expected position of the gap is at k̃SP ⫽ kg兾2. The black curve in Figs. 4(b1), 4(b2) and 4(c1), 4(c2) represents the set of 共, i兲 for which kx is equal to kg兾2. As predicted, the bandgap appears at the intersection of the black and dashed curves, and the slope of the dispersion relation is zero on each side of the gap. This confirms the expected increase in the SPW density of modes on the edge of the gap and justifies our approach for SP sensing tolerant to beam divergence. By comparing Figs. 4(b1) and 4(b2), it can be seen that for structure B, the size of the bandgap decreases substantially when the optical index in- creases. In this type of structure, the size of the gap is intimately related to the contrast between the optical index of the medium n3 and the dielectric that forms the grating nd [13]. The reflectivity maps corresponding to structure (c) present two distinctive features. First, the average reflectivity, even outside the SPR, is lower than what is obtained for structures (a) and (b). The main reason for this is due to the fact that the ⫺1 reflected diffractive mode is propagative. Indeed, around the bandgap position, the transverse momentum of the ⫺1 diffracted mode is ⫺kx and therefore necessarily propagative. Note that this corresponds to the Littrow mount configuration: the ⫺1 reflected beam propagates in the direction opposite the excitation beam. The diffraction efficiency of this order depends primarily on the shape of the metal– glass interface. In structure B, this interface is flat; the diffraction efficiency for this mode is low. Alternatively, the metal– glass interface in structure (c) is corrugated, and the diffraction efficiency is much higher while the specular reflectivity is lower. The second notable feature on the reflectivity map corresponding to structure (c) is the transverse minimum of reflectivity for longer wavelengths [white dashed curve in Fig. 4(c1)]. This feature corresponds to the excitation of SPW on the metal– glass interface in the negative direction. These modes are usually not accessible since their wavenumber kSPglass,metal is greater than n1k0. However, in this case, it is possible to excite the glass–metal SPW propagating in the opposite direction when ⫺kSPglass,metal ⫽ kx ⫺ kg. B. Bandgap-Assisted Surface Plasmon Sensing Once the reflectivity maps were computed, we calculated the reflected intensity for a Gaussian beam of angular deviation using Eq. (5). The arrows in Fig. 4 point to the value of the mean angle of incidence, i, and wavelength, , for which we chose to monitor the reflectivity, while n3 was varied from 1 to 1.2. Figure 1 is the plot of the reflectivity obtained for each structure for beam divergence of 0° and 2°. The choice of the side of the bandgap where the resonance was monitored was dictated by the angular width of the resonance on the edge of the bandgap. We observed that the resonance below (low energy) the bandgap was wider than the one above it, which is advantageous to diminish the effect of the divergence upon the sensitivity but detrimental for the overall sensitivity. However, in the case of structure (b), the size of the bandgap is dictated by the contrast of refractive indices between the ridges of the grating and the medium and diminishes when n3 increases. We took advantage of this effect by monitoring the resonance above the band for structure (b) in Fig. 1 and below the gap for structure (c). While the degradation of the traditional SPR signal [structure (a)] when the divergence increases is clearly visible, the signal for structures (b) and (c) stays almost the same because the dip of reflectivity corresponds to a position where the density of SP modes is high. This is the effect that we were seeking in order to maintain the quality of the sensing when the beam divergence increases. However, at ⫽ 0 (plane wave) the width of the reflectivity is much larger, and the slope is lower in the two SPBG cases when compared with the SPR technique. This corresponds to a sensitivity as measured by dR兾dn3 that is approximately 10 times lower. If we assume that the effect of changing n3 is only to shift the reflectivity map, it is possible to express the sensitivity as SBG ⫽ dR ⭸R ⭸BG ⭸R ⭸BG ⫽ ⫹ , dn3 ⭸ ⭸n3 ⭸ ⭸n3 (6) with BG and BG the angle and wavelength (in vacuum) corresponding to the center of the SPBG. Since we probe the signal on the band edge, we can make the approximation ⭸R兾⭸ ⬇ 0. Also, since the bandgap is situated at k̃SP ⫽ kg兾2, we have n1 2 sin BG ⫽ . BG g (7) By derivating this expression and introducing it in Eq. (6): dR ⭸R ⭸BG ⫽ dn3 ⭸ ⭸n3 ⭸R ⭸BG ⭸BG ⫽ ⭸ ⭸BG ⭸n3 SBG ⫽ ⫽ ⫺2n1g cos BG ⭸R ⭸BG , ⭸ ⭸n3 (8) making the approximation k̃SP ⫽ kSP, valid in structure (c) results in BG ⫽ SP, with SP the angle of SP resonance without corrugation. This allows us to compare SBG with the sensitivity of traditional SPR, SSPR: SSPR ⫽ ⭸R ⭸SPR , ⭸ ⭸n3 SBG 2g⭸R兾⭸ ⫽ ⫺n1 cos BG . SSPR ⭸R兾⭸ (9) (10) The relative sensitivity of the SPBG sensor compared with the traditional SPR sensor depends primarily on the comparison between the slopes of R in the wavelength direction versus the angle direction. By measuring slopes on the reflectivity map, and by using BG ⫽ 53°, we obtain SBG兾SSPR ⬇ 12%, which is in good agreement with our calculations for ⫽ 0. When the optical index of the prism is increased, BG decreases, which increases the comparative sensitivity of the sensor. Alternatively, if n3 increases, the comparative sensitivity of the bandgap sensor will reach zero in the limit of high angles. This is a problem in the case of fluid sensing where the angle of 1 June 2007 兾 Vol. 46, No. 16 兾 APPLIED OPTICS 3373 Fig. 5. Plots of the sensitivity, the detection limit, and the reflectivity minimum versus the angular deviation of the excitation beam for each structure. incidence is high and the use of a high index ratio prism is required to keep cos BG relatively high. Note that in the limit of a high angle of incidence, the term ⭸SP兾⭸n3 diverges and this model is no longer valid. Figure 5 represents the values of the sensitivity, reflectivity minimum, and detection limit against for the three structures. The sensitivities were calculated at the point of maximum slope. The maximum theoretical detection limit, which is the minimum amount of change in n3 that will produce a change of signal equal to the shot noise is proportional to [8] ␦n3,min ⬀ 冑R ⭸R兾⭸n3 . (11) We plotted the minimum of this quantity for the three structures in Fig. 5(b). The SPBG sensor is notably unaffected by increases in the divergence of the beam in sensitivity, reflectivity minimum, and detection limit, which proves that the increase in SP density of state can be 3374 APPLIED OPTICS 兾 Vol. 46, No. 16 兾 1 June 2007 used for sensing. The SPBG sensor outperforms the traditional SPR in sensitivity for ⬎ 1.1° and in the detection limit for ⬎ 0.7°. The threshold is lower for the detection limit because the minimum reflectivity stays low in the case of SPBG sensing and this reduces the factor 冑R (shot noise). An angular beam divergence of 1° corresponds to a waist radius of approximately 6 m at 600 nm. This is much smaller than the typical diameter of an optical fiber, and it is therefore unlikely that the SPBG technique would be useful in this case. However, this number corresponds to a perfectly Gaussian beam, which is difficult to realize in a compact format. Therefore, the main application of a SPBG sensing technique would be in situations where the divergence is nonzero because the beam is focused on a small area or when the optical system cannot be practically made perfectly collimated such as when the source is not a laser (diode or arc lamp). Another technique for mitigating the effect of the divergence on the sensitivity is to widen the resonance peak by changing the thickness of the metallic layer [8] in the Kretschmann configuration. However, changing the thickness of the metallic layer increases the reflectivity minimum, which has negative effects on the shot noise. Conversely, SPBG sensors have a sensitivity independent of divergence and a low reflectivity minimum. The differences in signal between the two BG structures (b) and (c) are minimal. The inherent lower sensitivity of structure (b), due to the fact that only part of the metal is in contact with the sensing medium, is compensated for by a higher reflectivity average than that of structure (c). Again, the lower reflectivity average for structure (c) is due to the corrugation on the metal– glass interface, which yields to a high diffraction efficiency in the ⫺1 reflected order. Since structure (c) is easier to fabricate, it is a good candidate for experimental validation. An ideal structure would be a metallic film that is corrugated on the metal– dielectric interface and flat on the glass–metal interface. In such a structure, the SPW would be totally perturbed by the changes in optical index, and the diffraction efficiency in the ⫺1 reflected order would be low. 4. Conclusion We have analyzed the use of a SP bandgap for sensing applications in amplitude-based SP sensors. Specifically, we showed that the increase in the density of modes on the gap edge can be used to remedy the degradation of the sensor signal when a divergent Gaussian beam is used to excite SPWs. Using RCWA, we analyzed two types of SPBG structure and compared their performances to traditional SPR sensing. For a plane wave, the sensitivities of SPBG sensors are 1 order of magnitude below the sensitivity of SPR, which is due to the differences in slopes in the momentum and energy axis. However, SPBG sensors outperform SPR sensors in sensitivity and in the detection limit as soon as the angular deviation is over a degree. We envision that this technique will be used for the design and fabrication of compact and low-cost SP sensors. This work was made possible by the National Science Foundation Center for Embedded Networked Sensing (CENS) under contract CCR-0120778. References 1. H. Raether, Surface Plasmons on Smooth and Rough Surfaces and on Gratings (Springer-Verlag, 1988). 2. J. Homola, “Present and future of surface plasmon resonance biosensors,” Anal. Bioanal. Chem. 377, 528 –539 (2003). 3. J. Melendez, R. Carr, D. Bartholomew, K. Kukanskis, J. Elkind, S. Yee, C. Furlong, and R. Woodbury, “A commercial solution for surface plasmon sensing,” Sens. Actuators B 35, 212–216 (1996). 4. E. Kretschmann, “Determination of optical constants of metals by excitation of surface plasmons,” Z. Phys. 241, 313–324 (1971). 5. G. Nenninger, M. Piliarik, and J. Homola, “Data analysis for optical sensors based on spectroscopy of surface plasmons,” Meas. Sci. Technol. 13, 2038 –2046 (2002). 6. J. Homola, “On the sensitivity of surface plasmon resonance sensors with spectral interrogation,” Sens. Actuators B 41, 207–211 (1997). 7. S. Wu, H. Ho, W. Law, C. Lin, and S. Kong, “Highly sensitive differential phase-sensitive surface plasmon resonance biosensor based on the Mach–Zehnder configuration,” Opt. Lett. 29, 2378 –2380 (2004). 8. A. Kolomenskii, P. Gershon, and H. Schuessler, “Sensitivity and detection limit of concentration and adsorption measurements by laser-induced surface-plasmon resonance,” Appl. Opt. 36, 6539 – 6547 (1997). 9. J. Homola, S. Yee, and G. Gauglitz, “Surface plasmon resonance sensors: review,” Sens. Actuators B 54, 3–15 (1999). 10. J. Villatoro and A. Garcia-Valenzuela, “Sensitivity of optical sensors based on laser-excited surface-plasmon waves,” Appl. Opt. 38, 4837– 4844 (1999). 11. F. Pincemin and J. Greffet, “Propagation and localization of a surface plasmon polariton on a finite grating,” J. Opt. Soc. Am. B 13, 1499 –1509 (1996). 12. B. Fischer, T. Fischer, and W. Knoll, “Dispersion of surfaceplasmons in rectangular, sinusoidal, and incoherent silver gratings,” J. Appl. Phys. 75, 1577–1581 (1994). 13. J. Yoon, G. Lee, S. Song, C. Oh, and P. Kim, “Surface-plasmon photonic band gaps in dielectric gratings on a flat metal surface,” J. Appl. Phys. 94, 123–129 (2003). 14. W. Barnes, T. Preist, S. Kitson, and J. Sambles, “Physical origin of photonic energy gaps in the propagation of surface plasmons on gratings,” Phys. Rev. B 54, 6227– 6244 (1996). 15. K. Ho, C. Chan, and C. Soukoulis, “Existence of a photonic gap in periodic dielectric structures,” Phys. Rev. Lett. 65, 3152– 3155 (1990). 16. M. Gonzalez, J. Weeber, A. Baudrion, A. Dereux, A. Stepanov, J. Krenn, E. Devaux, and T. Ebbesen, “Design, near-field characterization, and modeling of 45° circle surface-plasmon Bragg mirrors,” Phys. Rev. B 73, 155416 (2006). 17. J. Sanchez-Gil and A. Maradudin, “Surface-plasmon polariton scattering from a finite array of nanogrooves兾ridges: efficient mirrors,” Appl. Phys. Lett. 86, (2005). 18. S. Kitson, W. Barnes, and J. Sambles, “Full photonic band gap for surface modes in the visible,” Phys. Rev. Lett. 77, 2670 – 2673 (1996). 19. E. Palik, Handbook of Optical Constants of Solids (Academic, 1985). 20. M. Moharam, D. Pommet, E. Grann, and T. Gaylord, “Stable implementation of the rigorous coupled-wave analysis for surface-relief gratings— enhanced transmittance matrix approach,” J. Opt. Soc. Am. A 12, 1077–1086 (1995). 21. M. Moharam, E. Grann, D. Pommet, and T. Gaylord, “Formulation for stable and efficient implementation of the rigorous coupled-wave analysis of binary gratings,” J. Opt. Soc. Am. A 12, 1068 –1076 (1995). 22. P. Lalanne and G. Morris, “Highly improved convergence of the coupled-wave method for TM polarization,” J. Opt. Soc. Am. A 13, 779 –784 (1996). 23. J. Chandezon, D. Maystre, and G. Raoult, “A new theoretical method for diffraction gratings and its numerical application,” J. Opt. 11, 235–241 (1980). 24. L. Li, J. Chandezon, G. Granet, and J. Plumey, “Rigorous and efficient grating-analysis method made easy for optical engineers,” Appl. Opt. 38, 304 –313 (1999). 1 June 2007 兾 Vol. 46, No. 16 兾 APPLIED OPTICS 3375