## Abstract

We present a general approach for coupling a specific mode in a planar photonic crystal (PC) waveguide to a desired free-space mode. We apply this approach to a W1 PC waveguide by introducing small index perturbations to selectively couple a particular transverse mode to an approximately Gaussian, slowly diverging free space mode. This “perturbative photonic crystal waveguide coupler” (PPCWC) enables efficient interconversion between selectable propagating photonic crystal and free space modes with minimal design perturbations.

© 2011 OSA

## 1. Introduction

Photonic crystals (PCs) exhibit a photonic bandgap that enables localization and manipulation of light at the wavelength-scale. [1]. In two-dimensional (2D) photonic crystals, an important element is the photonic crystal waveguide, which enables compact optical delays [2, 3], slow light and enhanced nonlinear processes [4,5], and efficient connections between various on-chip devices [6–8].

Various techniques exist for coupling to photonic crystal waveguides, including chip-to-fiber butt-coupling [9], directional coupling from a tapered fiber [10, 11], and vertical grating couplers including slanted [12] or focusing lens gratings [13, 14]. Although these gratings can reach high efficiency, they constrain the PC device design as they disrupt the PC lattice and typically affect all waveguides modes. Here, we introduce a general approach for introducing highly non-invasive grating couplers that rely on small perturbations to existing device designs. For instance, in a W1 photonic crystal waveguide in a holey PC membrane, the grating consists only of small dielectric index perturbations to a set of holes near the waveguide region. We call this coupling device a Perturbative Photonic Crystal Waveguide Coupler (PPCWC). Although this approach is general, we focus here on the situation where we introduce extra holes adjacent to the existing air holes. This restricted set of perturbations eliminates fabrication challenges that may be associated with non-circular or very small, isolated perturbations.

Similar directional coupling from photonic crystal cavities by introducing small
perturbation has been explored in previous studies [15–17]. The index perturbation is derived here using coupled mode
theory, under the constraint that it can be easily integrated into the photonic
crystal fabrication. This approach enables a straightforward design of perturbations
that causes selective coupling between a desired PC waveguide mode and a desired
free-space mode (chosen here to be a Gaussian). Figure 1(a) shows the coupling scheme. When light propagates along the
waveguide (the red arrows) and passes through the coupler, the periodic index
perturbations act as a grating and light scatters out of the slab. We use
three-dimensional finite difference time domain (FDTD) simulations to calculate the
waveguide modes for a given index perturbation and to find the efficiency of the
resulting grating coupler. We find that even very small perturbations allow a high
coupling efficiency between the free-space radiative mode and a given waveguide
mode, while causing only very small changes to the propagation of other waveguide
modes. As demonstrated in the end of Section 2, the directional scattering from this
coupler is controllable by detuning the feed frequency or modulating the
perturbation size. In Section 3, we discuss the mode selectivity of our device by
investigating the problem in *k*-space, and in Section 4 we analyze
the transmittance and the modification of the shape and the directionality of the
scattered beam.

## 2. Coupled mode theory

We consider a guided wave in a W1 photonic crystal line defect waveguide and the light scattered at the grating coupler. We derive the coupled mode equations from wave equation

*ɛ*(

*r⃗*) is the relative dielectric constant and

*c*is the speed of light in vacuum. We define

*ɛ*as the relative dielectric constant for the waveguide, which is assumed to be periodic along the propagating direction. The solution of Eq. (1) with

_{w}*ɛ*=

*ɛ*is in form of

_{w}*B⃗ _{w}* are Bloch states having periodicities as the
multiples of the periodicity of

*ɛ*, where

_{w}*k⃗*represents the crystal momentum and

*x̂*is the propagating direction of the waveguide. We desire a scattered beam propagating in free space denoted by

*A⃗*(

*r⃗*).

*A⃗*is also a solution of Eq. (1) with the relative dielectric constant

*ɛ*=

*ɛ*. We assume the field to be a superposition of the waveguide and free space modes:

_{G}Here, *a*(*t*) is the slowly varying component of the
Gaussian beam, and
*b*(*k⃗*,*t*) and
*c*(*k⃗*,*t*) are the
slowly varying components of the forward and backward propagating Bloch modes for
the waveguide, respectively. Substituting Eq. (3) into Eq. (1), we
can obtain the full coupled mode equations describing the interaction of the
coefficients *a*(*t*),
*b*(*k⃗*,*t*), and
*c*(*k⃗*,*t*) through
Δ*ɛ*. In particular, the coupling rate between
*a*(*t*) and
*b*(*k⃗*,*t*) can be
evaluated [7,19] to be

*ɛ*=

_{w}*ɛ*–

_{t}*ɛ*, representing the index perturbations introduced into the already existing structures. The subscript letters

_{w}*w*and

*t*denote the existing waveguide and the target structure.

From Eq. (4), the coupling constant is
determined by *A⃗ ^{*}* ·

*B⃗*and the perturbation term Δ

_{k⃗}*ɛ*. Since the phase of A is nearly constant in the plane of the PC, whereas

_{w}*B⃗*varies rapidly along the direction of propagation, the spatial integral of the dot product

_{k⃗}*A⃗*·

^{*}*B⃗*will vanish. For a non-vanishing coupling constant, the index perturbation Δ

_{k⃗}*ɛ*should have the same periodicity as

_{w}*B⃗*, assuming an interaction region that is large compared to the period

_{k⃗}*a*.

Now we consider a realistic waveguide made by introducing a line defect in a
hexagonal photonic crystal lattice with periodicity *a*, refractive
index of the slab material *n* = 3.6, slab thickness
*h* = 0.7*a*, and hole radius
*r* = 0.3*a*. The full coupling system is
as shown in Fig. 1(a). We used a 3D FDTD
simulation to calculate the waveguide modes. Figure
1(b) shows the dispersion relation of the W1 photonic crystal waveguide.
A fundamental and higher order mode inside the photonic bandgap (PBG), wedged by the
areas shaded in gray, are shown in blue and green, respectively, and are confined
below the light line, represented by the red line. Figures 2(a) and 2(b) show the *x̂* and
*ŷ* electric field components of the fundamental mode,
which we label as *B⃗ _{k⃗}*, at wave-vector

*kx*=

*π*/

*a*, and frequency

*ω*= 0.2436, in units of 2

*πc*/

*a*, where c is the speed of light in free space. As shown previously, the index perturbation must have a period of 2

*π*/

*k⃗*= 2

_{x}*ax̂*to scatter this

*k*-state to the point in

*k*-space, where the in-plane

*k*= 0. Note that because the

_{x}*x̂*and

*ŷ*components of the electric field of the waveguide mode have different symmetries along the

*ŷ*direction, an index perturbation with even symmetry across

*y*= 0 will scatter only the

*ŷ*-component, but not the

*x̂*-component of the field. Similarly, it is possible to scatter only the

*x̂*-component with a anti-symmetric perturbation distribution across the

*y*= 0 line. For the following analysis, we will choose to scatter the

*ŷ*-component. Furthermore, we suppose that we want to scatter into a vertical mode that has a Gaussian profile in the in-plane directions, as shown in Fig. 2(c). Figure 2(d) shows the dot product

*A⃗*·

^{*}*B⃗*, which appears in the integrand in Eq. (4). To find a perturbation that scatters light into a slowly diverging Gaussian beam, we calculate the coupling rate for beams with different divergence angles and compare the coupling constants among different index perturbation designs. We discuss in this paper, the perturbations are formed by enlarging three-hole clusters with periodicity 2

_{k⃗,y}*a*in the

*x̂*-direction. The blue and green lines in Fig. 2(e) represent the coupling constants

*κ*

_{1}and

*κ*

_{2}for Gaussian beams, focused in the plane of the photonic crystal, with waists of 2

*a*and

*a*, respectively, and the red line represents the ratio

*κ*

_{1}/

*κ*

_{2}. The horizontal axis denotes the perturbation size in units of the area of the unit cell of the triangular lattice, and the vertical axes denote the coupling constant (left) and the ratio

*κ*

_{1}/

*κ*

_{2}(right). We observe that a local maximum exists for

*κ*

_{1}/

*κ*

_{2}when the index perturbation size is 0.3244, in units of the fractional area of unit cell, that is, the light scattered in this case diverges less than a well-defined Gaussian beam. The perturbed structure for this case is shown in Fig. 2(f) and the corresponding coupling strength

*κ*

_{1}and the ratio

*κ*

_{1}/

*κ*

_{2}are 0.2228 and 1.815, respectively. As expected, the coupling rate into the Gaussian beam of smaller waist is lower than for the wider, less rapidly diverging beam. The brown shaded area in the right panel of Fig. 2(f) shows the index perturbation.

We check our theoretical predictions using another 3D FDTD simulation of the coupling
efficiency of a grating consisting of the index perturbation as shown in Fig. 2(f). The fundamental waveguide mode at
frequency *ω* = 0.2436 is excited by a Gaussian beam
source at one end of the PC waveguide. The *ŷ*-component of
the electric field in the plane of the PC is shown in Fig. 3(a), which contains five periods of the index
perturbation shown in Fig. 2(f). The
perturbation couples the incident waveguide mode upward, resulting in reduced
waveguide transmission (< 0.5%) on the right side of the
perturbation. In Fig. 3(b), we consider the
effect of this perturbation on a higher-frequency mode at *ω*
= 0.3078, which also has even symmetry across the *y*
= 0 axis, but which has a periodicity of *a* in the direction
of propagation. Equation (4) predicts
that if Δ*ɛ* has a periodicity of
2*a*, the coupling constant *κ _{ba}*
will vanish in this case. This result is confirmed by a high in-plane transmission
after the coupler (> 99%), as shown in Fig. 3(b). We will discuss this mode selective property of
the coupler in greater detail in Section 3.

Figure 4 displays the energy density of the
*y⃗* component of the evanescent field and the upward
component of the scattering field from the coupler. Figure 4(a) is a snapshot viewed in the vertical cross-section through a
plane passing through the middle of waveguides, and Fig. 4(b) is the three dimensional view of the
*E _{y}* energy density isosurface. We observe a strong
radiated field originating from the 5-period long perturbed region.

As observed in Fig. 4, the radiation exits at 43° from the vertical. This direction depends on the interference of the fields scattered by discrete perturbations, as in a line grating. The interference is governed by relative amplitudes and phases of the fields at the positions of the perturbations. The scattered field can be approximated as a sum over all perturbations:

*r⃗*,

_{i}*u⃗*(

*r⃗*) denotes the spatial profile of the radiated field,

*k⃗*

_{0}is the

*k*-vector in free space,

*k⃗*= (

_{w}*π*/

*a*+Δ

*k*)

_{w,x}*x̂*is the crystal momentum of the perturbed PC waveguide where Δ

*k*indicates a possible shift in the

_{w,x}*k*-vector due to a slightly different dispersion in the perturbation region.

*r⃗*is th position of the

_{i}*i*-th scattering point, and

*r⃗*

_{1}is the position of the start of the coupler. The field dependence in Eq. (5) includes the decrease of the field amplitude along the coupler,

*α*, and the phase difference of the waveguide mode between perturbations. Because the perturbations are

*r⃗*have a periodicity of 2

_{i}*ax̂*, the phase difference is equivalent to 2

*a*Δ

*k*. Equation (5) therefore predicts a shifting of the diffraction angle with wavelength, as given by the waveguide’s dispersion relation. We observe this dependence in numerical simulations. In Fig. 5(a), the frequency of the incident waveguide mode is tuned from

_{w,x}*ω*= 0.2960 to

*ω*= 0.3120 as the scattering angle shifts from −4° to 20°.

The coupled mode theory discussed above assumes a weak perturbation to the waveguide mode. However, the mode intensity drops as the field propagates through the coupler. This decrease in mode intensity should be taken into account to improve the efficiency of the coupler. Figure 5(b) shows the structure of a 5-period long coupler in which we now include an increasing perturbation size to compensate for the decreasing field intensity. Using Eq. (5,6) to estimate the field intensity profile along the waveguide, we simulate a 5-period long coupler with perturbation sizes multiplied by factors of 0.8589, 0.8973, 0.9059, 0.9731, and 1. Figure 5(c) shows the resulting intensity distribution of the scattered field viewed in a vertical plane through the center of the waveguides. We now observe a more uniform scattering along the length of the perturbation coupler, which results in higher mode overlap with the vertical Gaussian beam.

## 3. Mode selectivity

It is clear from the coupled mode picture that the out-coupling rate from the
waveguide depends on the in-plane periodicity and the symmetry of the waveguide
mode; for example, as was shown in Fig. 3,
the in-plane transmission after the coupler is suppressed for the fundamental mode,
but not for the higher-order mode at *ω* = 0.3078. We
now analyze this mode selectivity more closely in *k*-space. Figure 6(a) compares the dispersion relations
for the fundamental mode (the blue line) and a higher mode (the green line) inside
the photonic band gap, having frequencies of *ω _{A}*
= 0.2436(2

*πc*/

*a*) and

*ω*= 0.3151 at

_{B}*k*=

_{x}*π*/

*a*. For the structure with index perturbations (the right panel), the dispersion is raised slightly because of the decrease in the effective refractive index. The periodicity of 2

*a*scatters modes near

*k*=

_{x}*π*/

*a*by a lattice vector of Δ

*k*=

_{x}*π*/

*a*to the point where

*k*= 0. Since these components are above the light line, they leak out of the PC plane and result in higher loss rate of the waveguide mode.

_{x}In Fig. 6(b), the blue line shows the loss
rate, *γ _{r}*, normalized by the loss rate of the
fundamental waveguide mode in the unperturbed structure, through different
perturbation patterns. The green line shows

*γ*for the higher-order waveguide mode at

_{r}*ω*= 0.3151. As the perturbation size increases, the loss rate of the fundamental mode increases by over two orders of magnitude, while the loss rate of the higher-mode remains roughly constant. This result demonstrates a high degree of selectivity of the perturbation coupler between different kinds of waveguide modes. Figure 6(c) shows the frequency increase of the fundamental and higher order mode (the blue and green lines).

## 4. Transmittance and beam shape modification

The efficiency of the coupler is estimated by calculating the transmittance from the waveguide mode to a target free space Gaussian spatial mode, which includes both upward and downward components. We approximate the efficiency by

*S*is the transmission out of the PC plane and the fraction quantifies the mode overlap of the upward(or downward)-traveling component of the radiated field with the target Gaussian beam profile. The scattering rate of the couplers with different index perturbations are shown in Fig. 7 (the red solid line). The perturbations not only scatter the incident waveguide mode into the Gaussian mode, but also induce backward reflection (the gray solid line) and other in-plane leakage (the gray dotted line). The maximal scattering rate is reached when

*r*

_{0}= 0 and the perturbation size in the sides is 0.4466, in units of the fractional area of unit cell. As shown in Fig. 7, the scattering rate doubles when the perturbation size increases about 40% (the region A), but slightly decreases as the radius of the central holes increases (the region B). If a perturbation is introduced only at the center of the waveguide (the region C in Fig. 7), then the incident light is mostly reflected and only weakly scattered out of plane.

When coupling to free space optics, and especially for single mode fiber coupling,
the scattered field should have a large overlap with a Gaussian mode. In the case
with the maximal scattering rate of 0.6989, we calculate an overlap of 89%
with a Gaussian beam focused in the plane of the PC. This indicates that the coupler
allows for both strong scattering and a highly Gaussian-like profile. The resulting
transmittance through the coupler is *T _{c}* = 0.6989
× 89% = 62%.

It is possible to decrease the beam divergence to allow for coupling to lower-NA
lenses. By increasing the extent of the PPCWC in both *x̂*
and *ŷ* (transverse in-plane direction), we expand the
coupling region and decrease the output beam divergence. This modified index
perturbation is shown in Fig. 8(a), again
optimized for the mode at *ω* = 0.2436. Figures 8(b) and 8(c) show the intensity
distributions of beams scattered from a small coupler (Fig. 3) and a broadened one (Fig. 8(a)), viewed in
*y*–*z* plane through the centers of the
couplers. Figures 8(d) and 8(e) compare the
divergence of the radiated beams in both cases by fitting a Gaussian beam profile to
the data. We assume 2*θ* =
2*W*_{0}/*z*_{0}, where
2*θ* is the divergence angle of the beam and
*W*_{0} and *z*_{0} are the beam
waist and depth of the focus, respectively. In the case of the broadened coupler,
the divergence is decreased by 20% from that of the original smaller
coupler.

Unfortunately, the out-of-plane scattering from the PPCWC is symmetrically upward and
downward, so both contribute to the total scattering rate. While large
*T _{c}* can be achieved by the methods described
above, the overall efficiency of the PPCWC is so far limited to 50%. The
upward scattering efficiency is defined as the directionality,

*P*and

_{U}*P*are the powers radiated in the upward and downward directions, respectively. One method of increasing

_{D}*D*is to redirect the downward scattered field with a mirror. This may be achieved with a DBR structure, separated from the PC slab by some distance,

*d*. When

*d*= 0.4667

*h*, the downward radiated light reflected by the mirror interferes with the upward radiated light constructively and a local maximum of

*D*= 70% occurs. When we move the mirror further from the slab (

*d*> 0.6

*h*), there is minimal interference between two beams due to divergence of the beams. The upward scattering rate is close to the total scattering rate in this case, and we achieve

*D*= 95%. More sophisticated methods may be used to increase directional scattering, such as only partial etching of the PPCWC. This has been successfully employed for photonic crystal cavities [15,17] and channel waveguides [20].

## 5. Conclusions

We have designed and implemented a method for selectively coupling a planar photonic crystal waveguide mode to a target free space mode. The method proceeds as follows: (1) one chooses a waveguide mode to out-couple, (2) one chooses an index perturbation to couple the mode that has the appropriate periodicity and symmetry given by coupled mode theory, and finally, (3) one optimizes the perturbation for the desired out-coupled beam parameters. To demonstrate this approach, we focused on the problem of coupling a W1 waveguide mode to a slowly diverging Gaussian mode. We were able to ensure that the target waveguide mode was coupled with high rate, selectivity, and overlap with the target Gaussian mode.

This perturbative photonic crystal waveguide coupler (PPCWC) not only enables efficient scattering of selected waveguide modes, but also allows modifications in the shape and the direction of scattering. The angle of the scattered field can be shifted by detuning the mode frequency with respect to the lattice constant or introducing some modulation to the perturbation size. By breaking the vertical symmetry of the PC or using reflectors under the PC slab, it is possible to minimize downward propagation, directing most vertical scattering upward. The PPCWC enables a straightforward integration of vertical in/out- couplers which can be employed in a range of application from optical interconnects to photonic crystal light sources. Furthermore, the spectrally resolved light coupler could find application in light trapping for thin-film solar photovoltaics.

## Acknowledgments

C.C.T. was partially supported by the Center for Re-Defining Photovoltaic Efficiency Through Molecule Scale Control, an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under Award Number DE-SC0001085. J.M. was supported by the DARPA Information in a Photon program, through grant W911NF-10-1-0416 from the Army Research Office.

## References and links

**1. **J. D. Joannopoulos, S. G. Johnson, J. N. Winn, and R. D. Meade, *Photonic Crystal: Molding the Flow of Light*,
2nd ed., (Princeton University
Press, 2008).

**2. **T. Baba, “Slow light in photonic
crystals,” Nature Photonics **2**, 465–473
(2008). [CrossRef]

**3. **H. Gersen, T. J. Karle, R. J. P. Engelen, W. Bogaerts, J. P. Korterik, N. F. van Hulst, T. F. Krauss, and L. Kuipers, “Real-space observation of ultraslow
light in photonic crystal waveguides” Phys.
Rev. Lett. **94**, 073903 (2005). [CrossRef] [PubMed]

**4. **J. F. McMillan, M. Yu, D. L. Kwong, and C. W. Wong, “Observation of four-wave mixing in
slow-light silicon photonic crystal waveguides,”
Opt. Express **18**, 15484–15497
(2010). [CrossRef] [PubMed]

**5. **M. Soljacic, S. G. Johnson, S. Fan, M. Ibanescu, E. Ippen, and J. D. Joannopoulos, “Photonic-crystal slow-light enhancement
of nonlinear phase sensitivity,” J. Opt.
Soc. Am. B **19**, 2052–2058
(2002). [CrossRef]

**6. **H. Takano, B. S. Song, T. Asano, and S. Noda, “Highly efficient multi-channel drop
filter in a two-dimensional hetero photonic
crystal,” Opt. Express **14**, 3491–3496
(2006). [CrossRef] [PubMed]

**7. **E. Waks and J. Vuckovic, “Coupled mode theory for photonic
crystal cavity-waveguide interaction,” Opt.
Express **13**, 5064–5073
(2005). [CrossRef] [PubMed]

**8. **S. Noda, A. Chutinan, and M. Imada, “Trapping and emission of photons by
single defect in a photonic bandgap structure,”
Nature (London) **407**, 608–610
(2000). [CrossRef]

**9. **M. Notomi, A. Shinya, S. Mitsugi, E. Kuramochi, and H. Ryu, “Waveguides, resonators, and their
coupled elements in photonic crystal slabs,”
Opt. Express **12**, 1551–1561
(2004). [CrossRef] [PubMed]

**10. **P. Bienstman, S. Assefa, S. G. Johnson, J. D. Joannopoulos, G. S. Petrich, and L. A. Kolodziejski, “Taper structures for coupling into
photonic crystal slab waveguides,” J. Opt.
Soc. Am. B **20**, 1817–1821
(2003). [CrossRef]

**11. **P. E. Barclay, K. Srinivasan, and O. Painter, “Design of photonic crystal waveguides
for evanescent coupling to optical fiber tapers and integration with high-Q
cavities,” J. Opt. Soc. Am. B **20**, 2274–2284
(2003). [CrossRef]

**12. **B. Wang, J. Jiang, and G. P. Nordin, “Compact slanted grating
couplers,” Opt. Express **12**, 3313–3326
(2004). [CrossRef] [PubMed]

**13. **A. Mizutani, Naoki Ikeda, Y. Watanabe, N. Ozaki, Y. Takata, Y. Kitagawa, F. Laere, R. Baets, Y. Sugimoto, and K. Asakawa, “Planar focusing lens grating for
vertical coupling on 2D photonic crystal slab
waveguide,” in Lasers and Electro-Optics
Society, 2006. LEOS 2006. 19th Annual Meeting of the IEEE, pp.
843–844
(2006).

**14. **A. Faraon, I. Fushman, D. Englund, N. Stoltz, P. Petroff, and J. Vuckovic, “Dipole induced transparency in
waveguide coupled photonic crystal cavities,”
Opt. Express **16**, 12154–12162
(2008). [CrossRef] [PubMed]

**15. **M. Toishi, D. Englund, A. Faraon, and J. Vuckovic, “High-brightness single photon source
from a quantum dot in a directional-emission
nanocavity,” Opt. Express **17**, 14618–14626
(2009). [CrossRef] [PubMed]

**16. **N. Tran, S. Combrie, and A. De Rossi, “Directive emission from high-Q photonic
crystal cavities through band folding,”
Phys. Rev. B **79**, 041101 (2009). [CrossRef]

**17. **Y. Tanaka, M. Tymczenko, T. Asano, and S. Noda, “Fabrication of two-dimensional photonic
crystal slab point-defect cavity employing local three-dimensional
structures,” Japanese J. Appl.
Phys. **45**, 6096–6102
(2006). [CrossRef]

**18. **A. Yariv, *Optical Electronics in Modern Communications*,
5th ed., (Oxford University
Press, New York,
1997).

**19. **A. Yariv, “Coupled-mode theory for guided-wave
optics,” IEEE J. Quantum Electron. **9**, 919–933
(1973). [CrossRef]

**20. **X. Chen, C. Li, C. K. Y. Fung, S. M. G. Lo, and H. K. Tsang, “Apodized waveguide grating couplers for
efficient coupling to optical fibers,” IEEE
Photon. Technol. Lett. **20**, 1156–1158
(2010). [CrossRef]