1. Introduction
Surface waves at the interface between a semi-infinite periodic layered dielectric medium and a homogeneous medium has long been studied as electromagnetic Bloch modes [
1
P. Yeh, “Guided waves in layered media,” in Optical waves in layered media , (John Wiley & Sons, New York, 1988).
]. In general, at any terminations of a two-dimensional (2-D) or three-dimensional photonic crystal (PC), surface modes exist as localized electromagnetic field distributions [
2–5
R. D. Meade, K. D. Brommer, A. M. Rappe, and J. D. Joannopoulos, “Electromagnetic Bloch waves at the surface of a photonic crystal,” Phys. Rev. B
44, 10961–10964 (1991). [CrossRef]
]. Various surface wave related phenomena have also been explored including enhanced transmission through slits in PC slabs and strong beaming of light emerging from a PC waveguide [
6
E. Moreno, F. J. Garcìa-Vidal, and L. Martìn-Moreno, “Enhanced transmission and beaming of light via photonic crystal surface modes,” Phys. Rev. B
69, 121402 (2004). [CrossRef]
], and leaky surface modes assisted transmissions in the forbidden band [
7
M. Laroche, R. Carminati, and J.-J. Greffet, “Resonant optical transmission through a photonic crystal in the forbidden gap,” Phys. Rev. B
71, 155113 (2005). [CrossRef]
]. Truncated 2-D PC slabs also support surface waves, as calculated and experimentally observed in a double-trench PC defect waveguide [
8–10
W. T. Lau and S. Fan, “Creating large bandwidth line defects by embedding dielectric waveguides into photonic crystal slabs,” Appl. Phys. Lett.
81, 3915–3917 (2002). [CrossRef]
]. These surface waves are localized at the interface between the trench and the PC region, rather than guided in the defect waveguide region. It has also been recognized that at frequencies near the PC Brillouin zone edge, surface waves exhibit a relatively large group refractive index (implying a slow group velocity), and thus a surface-mode microcavity is expected to be high-Q [
11
J.-K. Yang, S.-H. Kim, G.-H. Kim, H.-G. Park, Y.-H. Lee, and S.-B. Kim, “Slab-edge modes in two-dimensional photonic crystals,” Appl. Phys. Lett.
84, 3016–3018 (2004). [CrossRef]
,
12
S. Xiao and M. Qiu, “Surface-mode microcavity,” Appl. Phys. Lett.
87, 111102 (2005). [CrossRef]
].
Fig. 1 Schematic of a two-dimensional photonic crystal slab surface waveguide in the z-direction (GK symmetry direction) with a dielectric margin of width m. The waveguide has a finite length of L. Shaded frame indicates the PML region in the FDTD simulation domain. (b) Unit cell adopted (dashed-line box) in the plane wave expansion calculation using super-cell approach. (c) Schematic of a 4-port device comprising a finite-length surface waveguide side coupled to a strip waveguide. Port 1: input, Port 2: throughput, Port 3: forward couple, and Port 4: backward couple.
Here we report numerical simulations of surface waves guided in a relatively wide and flat dielectric margin in a high-index-contrast 2-D PC slab [
Fig. 1(a)]. Lightwave in such surface waveguide is partially confined by total internal reflections at the flat sidewall and by coherent scatterings at the PC holes lattice. The PC slab gives rise to much of the interesting dispersion characteristics similar to those of line defect waveguides in PC slabs. Compared to truncated PC slabs that also support surface states, a wide margin in structure under study allows fabrication tolerance, and also introduces multiple partially transmission modes in the band gap that can be relevant for multi-channel wavelength multiplexing applications (e.g. in passive optical networks (PON)). The flat sidewall enables convenient lateral evanescent coupling to the surface modes, e.g. through coupling the finite-size PC slab to a conventional dielectric waveguide [
Fig. 1(c)]. The coupling forms an asymmetric coupler with regular waveguide modes coupled to highly dispersive surface modes. Another example is our previously proposed surface guided photonic crystal embedded microcavities (PCEMs) [
13
K. K. Tsia and A. W. Poon, “Dispersion-guided resonances in two-dimensional photonic-crystal-embedded microcavities,” Opt. Express
12, 5711–5722 (2004). [CrossRef] [PubMed]
,
14
K. K. Tsia and A. W. Poon, “Dispersion-guided and bandgap-guided resonances in semiconductor waveguide-coupled hexagonal photonic-crystal-embedded microcavities,” presented at 2005 Conference on Lasers and Electro Optics/ Quantum Electronics and Laser Science Conference (CLEO/QELS), Baltimore, USA, 22–27 May 2005.
]. The surface PC waveguide can be configured to form a closed loop in a hexagonal-shaped microresonator. A conventional strip waveguide can then be side-coupled to one surface waveguide facet of such hexagonal microcavity that has embedded PC holes array.
2. Plane wave expansion calculations
Figure 1 shows the schematic of a 2-D PC slab surface waveguide. The PC slab comprises a triangular lattice of air holes, with lattice constant a and hole radius
r. The surface waveguide axis is in the
z-direction (ΓK symmetry direction). The waveguide transverse dimension is in the
x-direction (ΓM symmetry direction). The dielectric margin of width
m spanning between the flat sidewall and the edge of the first row of holes constitutes an asymmetric waveguide. We focus on the E-modes (electric field parallel to plane of the periodicity) in such PC slab surface waveguides.
Fig. 2. (a) - (c) PWE-calculated projected dispersion diagrams for PC surface waveguides of (a) m = 0.1a, (b) m = 0.7a, and (c) m = 1.2a. Surface modes A0 – A4 are highlighted. (d) – (g) PWE-calculated field patterns of modes (d) A0 in m = 0.1a waveguide, (e) A1, (f) A2 in m = 0.8a waveguide, and (g) A3 in m = 1.2a waveguide. (h) PWE-calculated surface waveguide modes (A0 – A5) zone edge frequencies with margin width spanning from 0.1a to 2.0a. 1st PC band lines are zone-edge frequency boundaries. Lines are for visual aid.
We first apply the plane wave expansion (PWE) method with super-cell approach in order to numerically calculate the projected dispersion diagrams of the PC slab surface waveguides of an infinite length. It is important to note that for
r = 0.3
a and refractive index
n = 3.5 (e.g. silicon refractive index), the PC slab displays a band gap between ~0.205
c/
a and ~0.275
c/
a, where
c is the speed of light in vacuum. We consider surface waveguides for a range of margin widths from
m = 0.1
a to
m = 2.0
a. We define a super-cell as one unit cell in the
z-direction (ΓK) and 11 unit cells in the
x-direction (ΓM), as shown in
Fig. 1(b). We discretize each unit cell into 32 computation steps. From the calculated dispersion diagrams, we extract several pertinent parameters about the surface waveguide modes including the group refractive index
n
g, the guided mode bandwidth, and the effective span in projected wavevector
k that displays a relatively large
n
g
.
Figures 2(a)–2(c) show the representative projected dispersion diagrams (projected to the
z- direction) for surface waveguides of
m = 0.1
a, 0.7
a, and 1.2
a. The band gap can be clearly discerned. The thin dark lines are the PC slab modes in the 1
st and 2
nd PC bands. The dashed line is the light line. We highlight the surface modes and label them as A0, A1, etc. For
m = 0.1
a [
Fig. 2(a)], we observe only a single surface mode A0. For
m = 0.7
a [
Fig. 2(b)], two surface modes A1 and A2 exist within the band gap, whereas mode A0 redshifts to below the lower PC zone edge frequency. For
m = 1.2
a [
(Fig. 2(c)], three surface modes A2, A3, and A4 exist within the band gap, whereas modes A0 and A1 redshift to below the lower PC zone edge frequency. It is noteworthy that mode A3 displays a flat band (~0.004
c/
a bandwidth) over a relatively wide
k span (~0.3 (2
π/
a) to ~0.5 (2
π/
a)).
Figures 2(d)–2(g) show the PWE-calculated field patterns at the zone edge frequencies for modes A0 – A3 with various
m’s. Modes A0 and A1 exhibit familiar field patterns as previously shown in corrugated surface waveguides [
11
J.-K. Yang, S.-H. Kim, G.-H. Kim, H.-G. Park, Y.-H. Lee, and S.-B. Kim, “Slab-edge modes in two-dimensional photonic crystals,” Appl. Phys. Lett.
84, 3016–3018 (2004). [CrossRef]
]. Whereas, modes A2 and A3 display two field extrema in the
x-direction.
Figure 2(h) shows the PWE-calculated surface modes zone edge frequencies for
m spanning from 0.1
a to 2.0
a. We identify a total of six different guided modes. As
m increases, the surface modes dispersion curves and thereby the zone edge frequencies redshift, and the number of surface modes within the band gap increases. The surface waveguide exhibits single guided mode (fundamental mode) in the PC band gap only for
m below ~0.3
a.
Fig. 3 (a) Surface modes k span (Δk) for n
g
exceeding 50 as a function of the zone-edge frequency. Lines are for visual aid. (b)–(d) Dispersion diagrams highlighting mode A3 dispersion curves in waveguides with m = 1.1a, m = 1.7a, and m = 1.9a.
Figure 3(a) shows the surface modes
k span (Δ
k) within which the modes exhibit a relatively large n
g
(arbitrarily set at exceeding
n
g
= 50) as a function of the zone-edge frequency. We find that Δ
k overall narrows as the mode redshifts with the margin width. For example, mode A2 displays a Δ
k that is narrowed from ~0.06 (2
π/a) to ~0.01 (2
π/a) with
m widens from 0.6
a to 1.4
a. Mode A2 exhibits ~3 to ~5 times wider Δ
k than mode A1 at the same zone-edge frequencies. Whereas mode A4 only displays marginally wider Δ
k than mode A1. It is noteworthy that mode A3 Δ
k exceeds ~0.1 (2
π/
a) for a relatively wide span of
m between 1.1
a and 1.7
a [see
Fig. 2(h)]. We remark that the lowest order mode A0 also exhibits wide Δ
k up to 0.1 (2
π/
a) in
m = 0.1
a waveguide, which is however prohibitively narrow and not favorable for device fabrication. Increasing the margin width to 0.2
a renders A0 mode Δ
k narrows to 0.01 (2
π/
a).
Figures 3(b)–
3(d) show the PWE-calculated dispersion diagrams for
m = 1.1
a,
m = 1.7
a, and
m = 1.9
a, with mode A3 Δ
k of ~0.128 (2
π/
a), ~0.112 (2
π/
a), and 0.054 (2
π/
a), respectively.
3. Finite-difference time-domain calculations
Here we examine surface waveguides of finite lengths
L = 16
a and
L = 32
a by means of 2-D finite-difference time-domain (FDTD) simulations. We employ 36 computation points per period in our FDTD calculations. We launch at longitudinal position z = 0 a single-pulse source (pulse width
c
Δt ~ 4.3
a) with center frequency ~0.2322
c/
a, and detect the transmission field at the waveguide forward end and the reflection field at the waveguide backward end. We apply perfectly matched layers (PMLs) of reflectivity 10
-8 and thickness of 0.5 μm to all four boundaries of the computation domain [
Fig. 1(a)]. The finite simulation domain effectively imposes mode-mismatch reflections at the two waveguide ends.
Fig. 4. (a) PWE-calculated projected dispersion diagram for a surface waveguide with a margin width of 0.8
a. Symbols are k-vectors retrieved from FDTD-simulated field patterns for modes A1 (

) and A2 (

). Dashed line shows the light line. (b) FDTD-simulated transmission spectra and (c) reflection spectra for
L = 16a (blue) and
L = 32
a (red). (d), (e) Zoom-in view of transmission bands for modes A1 and A2.
Figure 4(b) shows the simulated transmission spectra for
m = 0.8
a with
L = 16
a and
L = 32
a. The two waveguides show largely comparable spectra with sharper spectral features in the case of
L = 32
a. In both cases we observe two transmission bands. We contrast the transmission spectra with the PWE-calculated dispersion diagram, as shown in
Fig. 4(a). We find that the two transmission bands lie in the band gap. The minor one corresponds to mode A1, with peak transmissions below ~0.3 and spans from ~0.2
c/
a to ~0.209
c/
a. The transmission bandwidth is bound by mode A1 zone-edge frequency and the maximum frequency of the mode A1 dispersion curve. The major one corresponds to mode A2, with peak transmissions of ~0.75 and spans from ~0.233
c/
a to ~0.26
c/
a. The transmission bandwidth is bound by mode A2 zone-edge frequency and the intersection between the light line and the mode A2 dispersion curve. In between the two transmission bands is a forbidden band with a bandwidth of ~0.024
c/
a.
Figure 4(c) shows the corresponding simulated reflection spectra. We see strong reflections within the forbidden band. Reflection corresponding to mode A1 is relatively high (below ~0.5), whereas reflection corresponding to mode A2 is relatively low (below ~0.25). We attribute the reflection to the mode-mismatch reflection at the forward end of the surface waveguide. However, we note that for each waveguide the transmission and reflection spectra are not entirely complimentary, thus suggesting that light energy is partially lost to radiation modes in the PC slab and in the air.
Figures 4(d) and
4(e) show the zoom-in view of the transmission bands of modes A1 and A2. The spectra show strong oscillations, with a smaller period shown in the
L = 32
a spectrum than in the
L = 16
a spectrum. We attribute the spectral oscillations to the Fabry-Perot (FP) reflections between the waveguide end-faces (mode-mismatch reflections), similar to the experimentally observed FP resonance spectra in line defect waveguides [
15
M. Notomi, K. Yamada, A. Shinya, J. Takahashi, C. Takahashi, and I. Yokohama, “Extremely large group-velocity dispersion of line-defect waveguides in photonic crystal slabs,” Phys. Rev. Lett.
87, 253902 (2001). [CrossRef] [PubMed]
].
Fig. 5 FDTD-simulated time-averaged steady-state H-field intensity patterns of (a) mode Al and (b) mode A2 in m = 0.8a surface waveguides. The waveguide length L = 32a. (c) and (d) FDTD-simulated time-averaged steady-state H-field amplitude patterns. (e) and (f) Fourier analysis of the field patterns of modes (e) A1, and (f) A2. Arrows indicate minor peaks positions in the k-space.
Fig. 6. (a) PWE-calculated projected dispersion diagram for a surface waveguide with a margin width of 1.5
a. Symbols are k-vectors retrieved from the FDTD-simulated
H-field patterns for modes A3 (

) and A4 (

). FDTD-simulated
H-field patterns for modes (b) A3, and (d) A4. Fourier analysis of field patterns of (d) A3, and (e) A4.
Figures 5(a) and
5(b) show two representative FDTD-simulated time-averaged steady-state
H-field intensity patterns (
L = 32
a,
m = 0.8
a) at (a) mode A1 transmission peak at 0.20016
c/
a (near the zone-edge frequency) and (b) mode A2 transmission peak at 0.23314
c/
a (near the zone-edge frequency).
Figures 5(c) and
5(d) show the zoom-in view of the steady-state
H-field amplitude patterns.
We apply spatial Fourier transform to the simulated field patterns in order to extract
k-space information of the surface modes.
Figure 5(e) shows the Fourier transform of mode A1 field pattern [
Fig. 5(c)]. We identify three peaks in the
k-space, with major
k
z
component at ~0.5098 (2
π/
a), and minor
kz
components at ~0.246 (2
π/
a) and ~0.754 (2
π/
a). We fold
kz
components exceeding 0.5 (2
π/
a) back into the 1
st Brillouin zone (BZ). The three
kz
components are then reduced to two projected wavevectors (marked with *’s) at ~0.4902 (2
π/
a) and ~0.246 (2
π/
a), as shown in
Fig. 4(a).
Figure 5(f) shows the Fourier transform of mode A2 field pattern [
Fig. 5(d)]. We identify only one peak in the
k-space at
kz
~ 0.50098 (2
π/
a), corresponding to
kz
~ 0.49902 (2
π/
a) in the 1
st BZ. Other retrieved wavevector values in
Fig. 4(a) are obtained in the same fashion from the field patterns at various frequencies using
L = 32
a. The Fourier-transform retrieved
k-vectors show excellent agreement with the PWE-calculated surface mode dispersion curves.
We also examine the higher order modes A3 and A4 with
m = 1.5
a waveguide.
Figure 6(a) shows the PWE-calculated projected dispersion diagram. The Fourier-transform retrieved
k-vectors are shown as symbols.
Figures 6(b) and
6(c) show the FDTD-simulated
H-field patterns for modes A3 at 0.2185
c/
a and A4 at 0.2391
c/
a.
Figure 6(d) shows the
k-space distribution of the simulated mode A3 field pattern [
Fig. 6(c)]. We observe five peaks in the
k-space distribution, corresponding to three wavevectors (marked with *’s) on mode A3 dispersion curve in
Fig. 6(a). We see that the retrieved
k-vectors confirm the flat band characteristics of mode A3.
Figure 6(e) shows the Fourier analysis of mode A4 field pattern. Three major
k-space peaks are folded to two wavevectors on mode A4 dispersion curve in
Fig. 6(a). The retrieved
k-vectors to the left of the light line are due to the
k-vectors in the 2
nd BZ folded back to the 1
st BZ.
Fig. 7 (a) Mode A1 transmission spectra of the surface photonic crystal waveguide m = 0.6
a, and
L = 32
a with different ambient refractive indices: 1.33 and 1.335. (b) Fractional frequency shifts as a function of center frequency under ambient refractive index changes from 1.33 to 1.335, for modes
Al(

),
A2(

), and
A3(

). (c) – (e) Time-averaged
E-field intensity for (c) mode A1 at ~0.216 (
c/
a), (d) mode A2 at ~0.227 (
c/
a), and (e) mode A3 at ~0.238 (c/a).
4. Ambient refractive index sensing
Here we examine effects of small refractive index changes on the surface wave transmission. We tune the refractive index of the ambient cladding and the PC holes lattice, with an eye on potential applications in microfluidic sensing [
16–18
D. Erickson, T. Rockwood, T. Emery, A. Scherer, and D. Psaltis, “Nanofluidic tuning of photonic crystal circuits,” Opt. Lett.
31, 59–61 (2006). [CrossRef] [PubMed]
]. We consider the refractive index change in water (e.g. due to a presence of targeted biochemical molecules). We vary the ambient refractive index from
n
c
= 1.33, with
Δnc
= 0.005. We assume surface waveguide of periodicity
a = 0.36 μm and
L = 32
a. We use the Fabry-Perot spectra to identify the wavelength shifts.
Figure 7(a) shows the transmission spectra around mode A1 zone-edge frequency with two ambient refractive indices. Mode A1 spectra redshifts by ~1.3 nm as the ambient refractive index rises from 1.33 to 1.335. It is worth mentioning that the FP resonance wavelengths redshifts and the zone-edge wavelength redshift with such an ambient refractive change are comparable to the reported resonance wavelength shifts in a PC-based microcavity [
18
E. Chow, A. Grot, L. W. Mirkarimi, M. Sigalas, and G. Girolami, “Ultracompact biochemical sensor built with two-dimensional photonic crystal microcavity,” Opt. Lett.
29, 1093–1095 (2004). [CrossRef] [PubMed]
].
Figure 7(b) shows the Δ
nc
induced fractional frequency shifts |Δ(
c/a)| / (
c/a) as a function of center frequency (at selected transmission peaks) for modes A1, A2, and A3. The fractional frequency shift increases with the center frequency (with narrowing margin widths). Mode A1 exhibits largest fractional frequency shifts (~1.0×10
-3) to ambient refractive index variations among the three modes.
Figures 7(c)–
7(e) show the time-averaged
E-field intensity distribution for modes A1, A2 and A3. The
E-field intensity filling factor in the ambient region (integrated
E-field intensity normalized by the computation area as shown) for mode A1 is ~58.1%, and for modes A2 and A3 are ~51.1% and ~50.6%, respectively.
5. Asymmetric 4-port 3-channel bi-directional coupler design
In order to further illustrate potential functionalities of multimode surface waveguides, we show an initial design of a surface waveguide side-coupled to a conventional strip waveguide as a 4-port multi-wavelengths bi-directional coupler [see schematic in
Fig. 1(c)]. The design is motivated by potential passive optical network (PON) applications, with multiplexed wavelength channels of about 1.31 μm, 1.49 μm, and 1.55 μm. We assume silicon photonic devices with a device refractive index of 3.5 and air in the cladding and PC holes. We adopt the PC periodicity
a = 0.33 μm, margin of 0.23 μm (
m ~ 0.7
a), and a strip waveguide of
w = 0.25 μm. The lateral interaction length is 15.18 μm (
L = 46
a) and the coupling air gap distance is 0.2 μm.

Fig. 8. (a) Simulated spectra of the 4-port 3-channel bi-directional coupler. Port 2: black line, port 3: blue line, and port 4: red line. (b) PWE-calculated dispersion diagram of surface modes A1 (red) and A2 (blue), and of the strip waveguide (purple). Two intersections at (k1
ω
1) and (k2
, ω
2) correspond to the two resonant coupling. (c) - (e) Time-averaged steady-state H-field intensity patterns for (c) forward coupling to port 3 (λ = 1.3114 μm), (d) throughput transmission to port 2 (λ = 1.49 μm), and (e) backward coupling to port 4 (λ = 1.5494 μm). Dashed-line windows denote the coupling regions for Fourier transform analysis. (f), (g) Fourier-transformed k-space representation at (f) λ = 1.3114 μm, and (g) λ = 1.5494 μm. (h) Schematic extended dispersion diagram (1st and part of 2nd BZ’s). Strip waveguide mode projected wavevector k
WG2
matches with surface mode projected wavevector kA2
. Strip waveguide mode projected wavevector kWG1
matches with counter-propagating surface mode projected wavevector kA1
. L.L. is light line.
Figure 8(a) shows the FDTD-simulated spectra of output ports 2–4. At throughput port 2, the spectrum shows an overall wide-band transmission except for two pronounced transmission dips around 1.31 μm (3-dB bandwidth ~ 10 nm) and 1.55 μm (3-dB bandwidth ~ 26 nm). The forward coupling (port 3) shows a single transmission peak around 1.31 μm, whereas the backward coupling (port 4) shows a single reflection peak around 1.55 μm.
Figure 8(b) shows the PWE-calculated dispersion diagram of surface modes A1 and A2, and of the strip waveguide. The dispersion lines cross at (
k1
ω
1) ≈ (0.4406 (2
π/
a), 0.2138
c/
d) and (
k2
,
ω
2) ≈ (0.3775 (2
π/
a), 0.2516
c/a), suggesting phase matching between the surface modes and the strip waveguide mode. The transmission dips wavelengths correspond well to the two intersections normalized frequencies.
Figures 8(c)–
8(e) show the FDTD-simulated steady-state
H-field intensity patterns at wavelengths 1.3114 μm (forward coupling), 1.49 μm (throughput transmission), and 1.5494 μm (backward coupling). It is noteworthy that the waveguide-coupled surface mode field intensities suggest field enhancement.
We apply spatial Fourier transform analysis to the coupling regions (denoted by dashed-line windows) in order to extract k-space information of the forward and backward coupling modes.
Figures 8(f) and
8(g) show the Fourier-transformed k-space representations at
λ = 1.3114 μm and
λ= 1.5494 μm. For the forward coupling mode, we only discern one major k-space component in the 2
nd BZ with
kz
~ 0.624 (2
π/
a), representing co-propagating phase-matched strip waveguide mode and mode A
2. We remark that the
kz
value corresponds to ~0.376 (2
π/a) in the 1
st BZ. For the backward coupling mode, we discern one major k-space component in the 1
st BZ with
kz
~ 0.445 (2
π/a), representing forward propagating strip waveguide mode, and another one in the 2
nd BZ in the negative domain with
kz
~ -0.553 (2
π/a), representing backward propagating mode A1. We note that the kz value in the 2
nd BZ corresponds to ~0.447 (2
π/a) in the 1
st BZ.
We explain the bi-directional two-mode coupling using a schematic extended band diagram spanning the full 1
st and part of 2
nd BZ’s, as shown in
Fig. 8(h). According to the Fourier-transformed k-space, modes A1 and A2 dispersion lines are in the 2
nd BZ, whereas the waveguide dispersion line extends from the 1
st to the 2
nd BZ. We see that the forward propagating waveguide mode at
ω
2 phase matches with mode A
2 (
kWG2
=
kA2
), suggesting directional coupling. In contrast, the forward propagating waveguide mode at
ω
1 only phase matches with mode A1 with an additional momentum of 2π/a (
kWG1
= 2
π/a +
kA1
), suggesting Bragg-reflection assisted counter-directional coupling.
6. Conclusions
In conclusion, we have numerically studied surface modes in a two-dimensional photonic crystal surface waveguide with a flat dielectric margin and a finite length. We examined the multiple guided modes inside the photonic band gap as a function of the margin width spanning from 0.1a to 2.0a. Our numerical analysis based on plane wave expansion method suggests that some high-order modes demonstrate group refractive indices exceeding 50 with a relatively wide k-span exceeding 0.1 (2π/a). We used two-dimensional FDTD to evaluate the finite-length surface guided mode transmission and reflection and their mode-field patterns. We apply spatial Fourier analysis on the simulated mode-field patterns and verify the flat bands in surface wave dispersion curves. For ambient index variation of Δnc
/nc
= 0.005/1.33, wavelength redshifts of |Δ(c/a)| / (c/a) ~ 1×10-3 can be attained.
The surface waveguide provides one interface for side-coupling to other photonic components or provides a useful platform for more sophisticated device architectures. We demonstrated a simple example with the surface waveguide laterally coupled to a conventional strip waveguide. Such coupling forms a 4-port bi-directional coupler with potential to realize three wavelengths multiplexer/demultiplexer (e.g. for passive optical networks (PON) applications). We envision that the surface waveguide can be incorporated with other photonic devices in order to realize further functionalities such as dispersion control, buffering, and (biochemical) sensing.