## Abstract

In this paper, we propose the use of the invariant based shortcuts to adiabaticity for the analysis of directional couplers. By describing the dynamical evolution of the system using the eigenstates of the invariant through new parameterizations, the system stability against errors in coupling coefficient and propagation constants mismatch is connected with the new parameters, which can be linked back to system parameters through inverse engineering. The merits and limitations of the conventional tapered directional coupler designs with various window functions are obtained through the analysis. We then propose an optimal design of compact directional couplers that is stable against errors in input wavelength and coupling coefficient simultaneously. The designed directional coupler has better tolerance, as compared to the conventional resonant couplers with smooth shape functions of Hamming and Blackman. These results are verified by beam propagation simulations.

© 2016 Optical Society of America

## 1. Introduction

Directional couplers with tolerance to variations in fabrication and wavelength are highly desirable with many interesting and significant applications in integrated optics and optical communications [1–5]. Conventionally, the analysis of directional couplers are based on the solution of coupled mode equations [6], where the spectral properties and fabrication tolerance can be solved for directional couplers having a constant spacing and a uniform cross section. On the other hand, tapered directional couplers with changes in the device geometry can be designed to suppress the side lobes in the device spectra, and coupled-mode theory analysis of such devices leads to a nonlinear differential equation that can only be solved numerically, shedding little insights on the device characteristics beyond some special cases [7–9].

Recently, a family of quantum control protocols called “shortcuts to adiabaticity (STA)” [10] has been exploited in light manipulation in optical waveguide structures [11–18]. For instance, the invariant-based inverse engineering [19] and counter-diabatic driving (or quantum transitionless algorithm) [20, 21] have been directly applied to obtain fast and robust mode conversion/splitting in multimode waveguides [11, 12]. As a matter of fact, invariant-based inverse engineering and quantum transitionless algorithm are mathematically equivalent [21], therefore both approaches are capable of realizing the shortcuts for directional couplers in optical waveguides [13–16]. These applications of various STA protocols to the design of coupled-waveguide devices illustrate the analogies one can draw between quantum mechanics and wave optics [22], and that the robustness of STA protocols can be directly translated to these waveguide devices. In fact, these STA protocols, while originally developed to accelerate “slow” adiabatic passages in various quantum systems, also shed lights on the system dynamics beyond the adiabatic limits. In particular, the inverse engineering, combined with perturbation theory and optimal control, provides a versatile toolbox for designing the optimal shortcuts with respect to different error and perturbation [23–25], in terms of invariant dynamics.

In this paper, we present a systematic STA analysis of directional couplers using the inverse engineering based on the Lewis-Risenfeld invariants. Combined with the perturbation theory, the STA based analysis allows one to handle directional couplers with spatially varying geometry based on the invariant dynamics. Tapered designs using the Hamming and Blackman window functions [7,8] are analyzed with the new approach, showing the merits and limitations of these designs. Finally, we use the invariant based STA to minimize the error sensitivities of directional couplers to wavelength and fabrication errors simultaneously by optimizing the dynamical invariant. The spatially varying coupling coefficient and propagation constant mismatch are inversely engineered to achieve a compact and robust directional coupler. The performance of various directional couplers are compared to demonstrate that the proposed directional coupler has better tolerance, as compared to the common parallel coupler and the directional couplers with various window functions.

## 2. The STA description of power coupling in directional couplers

#### 2.1. The invariant based STA

In this section, we introduce the inverse engineering based on the Lewis-Risenfeld invariant and its connection to power evolution in directional couplers. Consider a directional coupler of length *L*, consisting of two waveguides placed in proximity with propagation constants *β*_{+}(*z*) and *β*_{−}(*z*), see Fig. 1, where the cross-section of the waveguide structure is shown. Under the scalar and paraxial approximation and assuming weak coupling, the changes in the guided-mode amplitudes in the individual waveguides Ψ = [*A*_{+}, *A*_{−}]* ^{T}* with propagation distance is described by coupled-mode equations as,

*id*Ψ/

*dz*= H

_{0}(

*z*)Ψ, that is

*z*) (real) is the coupling coefficient, and Δ ≡ Δ(

*z*) = [

*β*

_{−}(

*z*) −

*β*

_{+}(

*z*)] describes propagation constants mismatch between the waveguides. Replacing the spatial variation

*z*with the temporal variation

*t*, Eq. (1) is equivalent to the time-dependent Schrödinger equation (

*ħ*≡ 1) describing the interaction dynamics of a two-state system driven by a coherent laser excitation [22], in which Ω and Δ are Rabi frequency and detuning, respectively. In optical waveguides, the propagation constant mismatch Δ is linearly dependent on the width difference, and the coupling coefficient Ω can be related to the separation between two waveguides [26]. The conventional coupled mode theory analysis describes the evolution of Ψ using the guided-modes in the individual waveguides as the basis, and the mode coupling behavior is obtained by solving Eq. (1).

The idea of the invariant based STA is that one can describe the evolution of Ψ using the eigenstates of the dynamical invariant **I**(*z*) with the invariant satisfying *∂ _{z}*

**I**+ (1/

*i*)[

**I**,

**H**

_{0}] = 0 [27]. And the eigenstates of the invariant are decoupled during system evolution. We can parameterize the eigenstates of the invariant as

*ψ*

_{0}(

*z*)|

*ψ*

_{⊥}(

*z*)〉 = 0)

*ψ*

_{0}(

*z*)〉 〈

*ψ*

_{0}(

*z*)| (|

*ψ*

_{⊥}(

*z*)〉 〈

*ψ*

_{⊥}(

*z*)|) is a dynamical invariant [23–25].

To describe power coupling between two waveguides along |*ψ*_{0}(*z*)〉, up to a phase factor, from the initial states
$|{\psi}_{0}(0)\u3009=|2\u3009\equiv \left(\begin{array}{c}0\\ 1\end{array}\right)$ to the final state
$|{\psi}_{0}(L)\u3009=|1\u3009\equiv \left(\begin{array}{c}1\\ 0\end{array}\right)$, we should set the following boundary conditions at *z* = 0 and *z* = *L*:

*θ*, satisfying the boundary conditions, to achieve power coupling. The freedom allows us to combine the inverse engineering and optimal control to select the most robust state dynamics in presence of various noise and systematic errors [23–25].

#### 2.2. Perturbation theory

In optical waveguides, there exist both input wavelength and coupling coefficient errors from fabrication at the same time, which can be described by the systematic errors in detuning and Rabi frequency, namely, H′ = (*ħ*/2)(*η*_{Ω}Ω*σ _{x}* +

*η*

_{Δ}

*σ*), with Pauli matrix

_{z}*σ*and the error amplitude

_{x,z}*η*

_{Ω,Δ}.

In the error-free case, we consider the family of protocols that results in perfect state transfer from |2〉 to |1〉: the unperturbed solution, |Ψ(*z*)〉 = |*ψ*_{0}(*z*)〉, satisfying Eq. (7). By assuming the ideal, unperturbed Hamiltonian, H_{0}, given by Eq. (1), we consider H′ as the perturbation in our perturbation theory calculation. Finally, the efficiency of the system to be in state |1〉 at *z* = *L* (coupling efficiency) is calculated as,
$P={\left|\u30081\left|\mathrm{\Psi}(L)\right|\u3009\right|}^{2}\approx 1-{\left|{\int}_{0}^{L}dz\u3008{\psi}_{\perp}(z)\left|{H}^{\prime}\right|{\psi}_{0}(z)\u3009\right|}^{2}$ [16,23], which turns out to be

*q*

_{Δ,Ω}are the error sensitivities, defined as

## 3. Applications to device analysis

#### 3.1. Resonant coupler (Δ = 0) with uniform spacing (Ω = *constant*)

First of all, we shall discuss the simplest case of directional coupler with no propagation constants mismatch and a uniform spacing. With Δ = 0, we obtain *β* = *π*/2 from Eq. (5). To obtain a constant Ω, the function of *θ* is chosen as

*s*=

*z/L*, which results in Ω =

*π/L*. Clearly, such coupler works like the flat

*π*pulse in quantum optics, which is the shortest design but sensitive to the parameter variations.

#### 3.2. Resonant coupler (Δ = 0) with tapered coupling (Ω(*z*))

To suppress side lobes in the simple directional couplers, several window functions, such as Hamming and Blackman [7, 8], are used in directional couplers with tapered coupling coefficients. With the right boundary conditions, the Hamming and Blackman functions of *θ*(*z*) are given by

When considering solely the coupling coefficient variations, described by the error in Rabi frequency, we have the efficiency for the resonant coupling (Δ = 0) as follows [23],

*θ*, the stability with respect to the coupling efficient error can’t be improved for resonant coupler (

*β*=

*π*/2), since the “pulse area” needs to be fulfilled. In this case, Eq. (9) gives

*q*

_{Ω}=

*π*

^{2}/4, and the efficiency is calculated as $P\approx 1-{\pi}^{2}{\eta}_{\mathrm{\Omega}}^{2}/4$, which is consistent with Eq. (14), when

*η*

_{Ω}is perturbative.

Regarding the input wavelength fluctuation, we consider it as the systematic error in detuning through the propagation constants mismatch. So the efficiency can be obtained by $P\approx 1-{\eta}_{\mathrm{\Delta}}^{2}{q}_{\mathrm{\Delta}}$, where Eq. (10) gives

with*β*=

*π*/2 and

*γ*= 0 for resonant couplers. In principle,

*q*

_{Δ}can be nullified by a sudden jump from

*θ*(0) = 0 to

*θ*(

*L*) =

*π*. This, however, results in infinite coupling coefficient Ω as shown by Eq. (4). So, smooth window functions such as the Hamming and the Blackman functions can improve the stability with respect to input wavelength variations, since the value of the integral in Eq. (16) is reduced as compared to using the linear function in Eq. (11). The STA analysis clearly shows that the window functions for

*θ*, see Eqs. (12) and (13), only improve the stability against input wavelength error but not coupling coefficient error.

#### 3.3. Design of an optimal coupler

In what follows, we shall design a directional coupler with higher stability with respect to two errors simultaneously using the STA. We shall design the function *θ* and *β* to nullify both *q*_{Ω} and *q*_{Δ}, which guarantees the best tolerance with respect to the variation of input wavelength and coupling coefficient. For simplicity, we choose *θ _{l}* (

*z*) without loss of generality, and impose [25]

*c*

_{1}and

*c*

_{2}. Numerical calculation gives

*c*

_{1}= −1.12 and

*c*

_{2}= 0.84, resulting in

*q*

_{Δ}= 0 and

*q*

_{Ω}= 0, and the value of following integral is

*π*, see Eq. (15). Once the parameters

*c*

_{1}and

*c*

_{2}are fixed, one can solve for

*β*by Eq. (6). Actually, one can also use the Hamming and Blackman functions of

*θ*to achieve the

*c*

_{1}and

*c*

_{2}for optimization, see Table 1. Finally, the detuning and Rabi frequency can be inversely designed by Eqs. (4)–(6), and the optimal directional coupler is thus achieved. To clarify how it works, we choose the simpler case of linear function

*θ*(

_{l}*z*), as an example in the following beam propagation simulations.

## 4. BPM simulation and discussion

Next, we turn to perform the numerical simulation for polymer waveguides using beam propagation method (BPM) [28]. The cross-section of the polymer channel waveguide structure is shown in Fig. 1, where the waveguide spacing *D* and widths *W _{L,R}* are adjusted along the propagation direction to satisfy the designed set of Ω and Δ. In detail, the relation between Δ and the width difference

*δW*=

*W*−

_{L}*W*can be approximated by a linear relation [1], and the relation between Ω and waveguide separation

_{R}*D*in a symmetric coupler is well fitted by the exponential relation Ω = Ω

_{0}exp[−

*κ*(

*D*−

*D*

_{0})] [26]. As a consequence, the waveguide parameters can be achieved by such mapping. The length of the devices is set at

*L*= 2000

*μm*, and the central wavelength is set at 1.55

*μm*. For the resonant couplers, the waveguide width is

*W*=

_{L}*W*= 2

_{R}*μm*and

*δW*= 0. Also the waveguide separation

*D*is obtained from Ω by using different

*θ*, see Fig. 2(a). For the optimal design, we obtain the parameters,

*D*and

*W*by using the inverse engineering, as shown in Fig. 2(b).

_{L,R}Figure 3 displays the BPM simulations for resonant couplers with fixed and tapered coupling and the optimal coupler, where the geometry of the designed directional couplers are shown. The power is completely coupled from one waveguide to the other. The coupling efficiencies versus different errors are shown in Fig. 4. The designed optimal directional coupler has better tolerance, as compared to the resonant couplers. Furthermore, the numerical simulation verifies the analysis on the use of window functions in section 3.2. In Fig. 4(a), using different *θ* functions actually improves the stability, which agrees with the theoretical analysis, also see Eq. (16). And in Fig. 4(b), it is clear that the stability doesn’t change with the chosen window function. Remarkably, for the optimal directional coupler, the bandwidth is increased from 1.35 *μm* to 1.85 *μm*, for coupling efficiencies greater than 95 %. The spacing error window is also increased, with greater than 95 % coupling efficiency from −0.80 *μm* to 0.50 *μm*. These results show significant improvement over conventional designs with a minimized spacing error and a flat wavelength window. The performance of the devices subject to variations in the waveguide widths is also investigated in Fig. 4(c). Width variations lead to errors in both Ω and Δ simultaneously in the optimal coupler. For the optimal coupler, the fabrication tolerance for greater than 95 % coupling efficiency is from +0.40 *μm* down to −1.0 *μm*. For the resonant couplers, the fabrication tolerance in width variation is from around −0.30 *μm* to 0.60 *μm*, with better tolerance on the positive variation side, because Ω is increased due to larger modal overlap. The insertion loss of the optimal coupler is 0.2 *dB* larger than that of the resonant couplers due to the sharp corners seen in Fig. 3(a). These sharp turns also limit the minimum device length achievable with the STA technique, as a shorter device length would lead to shaper turns in the resulting coupler.

## 5. Conclusion

In conclusion, we have systematically introduced the STA as an new way for the analysis of directional couplers. New insights on the merits and limitations of the conventional tapered directional coupler designs with various window functions are obtained through the analysis. It is shown that the use of window functions only improves the device stability against wavelength errors. We also applied the STA to design a compact and optimal directional couplers against spacing and wavelength errors simultaneously. BPM simulations demonstrate that the designed directional coupler is compact with better robustness, as compared to the conventional couplers, even with smooth window functions of Hamming and Blackman. The STA technique can be applied in general to weakly coupled waveguide systems described by the coupled mode theory. For applications to high index contrast waveguides such as silicon photonic waveguides [29], care must be taken to ensure that the device is working in the weak coupling regime.

## Funding

NSFC (11474193); Shuguang Program (14SG35); MOST of Taiwan (104-2221-E-006 -212, 105-3113-M-110-001, 105-2221-E-006-151-MY3); Research Fund for the Doctoral Program (2013310811003); Program for Eastern Scholar.

## References and links

**1. **A. Syahriar, V. M. Schneider, and S. Al-Bader, “The design of mode evolution couplers,” J. Lightwave Technol. **16**, 1907–1914 (1998). [CrossRef]

**2. **T. A. Ramadan, R. Scarmozzino, and R. M. Osgood, “Adiabatic couplers: design rules and optimization,” J. Lightwave Technol. **16**, 277–283 (1998). [CrossRef]

**3. **R. R. A. Syms, “The digital directional coupler: improved design,” IEEE Photonics Technol. Lett. **4**, 1135–1138 (1992). [CrossRef]

**4. **G. T. Paloczi, A. Eyal, and A. Yariv, “Wavelength-insensitive nonadiabatic mode evolution couplers,” IEEE Photonics Technol. Lett. **16**, 515–517 (2004). [CrossRef]

**5. **X. Sun, H.-C. Liu, and A. Yariv, “Adiabaticity criterion and the shortest adiabatic mode transformer in a coupled-waveguide system,” Opt. Lett. **34**, 280–282 (2009). [CrossRef] [PubMed]

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

**7. **R. C. Alferness, “Optical directional couplers with weighted coupling,” Appl. Phys. Lett. **35**, 260–262 (1979). [CrossRef]

**8. **R. C. Alferness and P. S. Cross, “Filter characteristics of codirectional coupled waveguides with weighted coupling,” IEEE. J. Quantum Electron. **14**, 843–847 (1978). [CrossRef]

**9. **B. E. Little, “Filter synthesis for coupled waveguides,” J. Lightwave Technol. **15**, 1149–1155 (1997). [CrossRef]

**10. **E. Torrontegui, S. Ibáñez, S. Martínez-Garaot, M. Modugno, A. del Campo, D. Guéry-Odelin, A. Ruschhaupt, X. Chen, and J. G. Muga, “Shortcuts to adiabaticity,” Adv. At. Mol. Opt. Phys. **62**, 117–169 (2013). [CrossRef]

**11. **T.-Y. Lin, F.-C. Hsiao, Y.-W. Jhang, C. Hu, and S.-Y. Tseng, “Mode conversion using optical analogy of shortcut to adiabatic passage in engineered multimode waveguides,” Opt. Express **20**, 24085–24092 (2012). [CrossRef] [PubMed]

**12. **S.-Y. Tseng and X. Chen, “Engineering of fast mode conversion in multimode waveguides,” Opt. Lett. **37**, 5118–5120 (2012). [CrossRef] [PubMed]

**13. **S.-Y. Tseng, “Counterdiabatic mode-evolution based coupled-waveguide devices,” Opt. Express **21**, 21224–21235 (2013). [CrossRef] [PubMed]

**14. **S.-Y. Tseng and Y.-W. Jhang, “Fast and robust beam coupling in a three waveguide directional coupler,” IEEE Photon. Technol. Lett. **25**, 2478–2481 (2013). [CrossRef]

**15. **S.-Y. Tseng, R.-D. Wen, Y.-F. Chiu, and X. Chen, “Short and robust directional couplers designed by shortcuts to adiabaticity,” Opt. Express **22**, 18849–18859 (2014). [CrossRef] [PubMed]

**16. **S.-Y. Tseng, “Robust coupled-waveguide devices using shortcuts to adiabaticity,” Opt. Lett. **39**, 6600–6603 (2014). [CrossRef] [PubMed]

**17. **S. Martínez-Garaot, S.-Y. Tseng, and J. G. Muga, “Compact and high conversion efficiency mode-sorting asymmetric Y junction using shortcuts to adiabaticity,” Opt. Lett. **39**, 2306–2309 (2014). [CrossRef] [PubMed]

**18. **X. Chen, H.-W. Wang, Y. Ban, and S.-Y. Tseng, “Short-length and robust polarization rotators in periodically poled lithium niobate via shortcuts to adiabaticity,” Opt. Express **22**, 24169–24178 (2014). [CrossRef] [PubMed]

**19. **X. Chen, A. Ruschhaupt, S. Schmidt, A. del Campo, D. Guéry-Odelin, and J. G. Muga, “Fast optimal frictionless atom cooling in harmonic traps: shortcut to adiabaticity,” Phys. Rev. Lett. **104**, 063002 (2010). [CrossRef] [PubMed]

**20. **X. Chen, I. Lizuain, A. Ruschhaupt, D. Guéry-Odelin, and J. G. Muga, “Shortcut to adiabatic passage in two- and three-level atoms,” Phys. Rev. Lett. **105**, 123003 (2010). [CrossRef] [PubMed]

**21. **X. Chen, E. Torrontegui, and J. G. Muga, “Lewis-Riesenfeld invariants and transitionless quantum driving,” Phys. Rev. A **83**, 062116 (2011). [CrossRef]

**22. **S. Longhi, “Quantum-optical analogies using photonic structures,” Laser Photonics Rev. **3**, 243–261 (2009). [CrossRef]

**23. **A. Ruschhaupt, X. Chen, D. Alonso, and J. G. Muga, “Optimally robust shortcuts to population inversion in two-level quantum systems,” New J. Phys. **14**, 093040 (2012). [CrossRef]

**24. **X.-J. Lu, X. Chen, A. Ruschhaupt, D. Alonso, S. Guérin, and J. G. Muga, “Fast and robust population transfer in two-level quantum systems with dephasing noise and/or systematic frequency errors,” Phys. Rev. A **88**, 033406 (2013). [CrossRef]

**25. **D. Daems, A. Ruschhaupt, D. Sugny, and S. Guérin, “Robust quantum control by a single-shot shaped pulse,” Phys. Rev. Lett. **111**, 050404 (2013). [CrossRef] [PubMed]

**26. **K. Okamoto, *Fundamentals of Optical Waveguides* (Academic, 2006).

**27. **H. R. Lewis and W. B. Riesenfeld, “An exact quantum theory of the time-dependent harmonic oscillator and of a charged particle in a time-dependent electromagnetic field,” J. Math. Phys. **10**, 1458–1473 (1969). [CrossRef]

**28. **K. Kawano and T. Kitoh, *Introduction to Optical Waveguide Analysis: Solving Maxwell’s Equations* (John Wiley and Sons, 2001). [CrossRef]

**29. **T.-H. Pan and S.-Y. Tseng, “Short and robust silicon mode (de)multiplexers using shortcuts to adiabaticity,” Opt. Express **23**, 10405–10412 (2015). [CrossRef] [PubMed]