## Abstract

We demonstrate by numerical simulations of coupled generalized nonlinear Schrödinger equations that the input pulse duration as well as the retarded material response have a crucial impact on the properties of an ultrafast nonlinear optofluidic fiber coupler. This device is composed of two waveguides in close vicinity embedded in a photonic crystal fiber which are filled with a highly nonlinear liquid. We show that in particular the critical peak power above which the coupling between the waveguides is suppressed increases dramatically for short input pulses and long characteristic response times of the liquid. We establish a simple model which describes these effects with high accuracy.

© 2011 Optical Society of America

With the recent advances in preparing selectively liquid-filled photonic crystal fibers (PCFs) [1,2] it has become possible to fabricate new coupling devices [3] based on two or more nearby waveguides filled with highly nonlinear liquids embedded in the photonic structure of a PCF [4] (see Fig. 1a). These devices are of high interest for the studies of the nonlinear coupling behavior since they offer the opportunity to tune the coupling strength, the dispersion, as well as the nonlinear parameter by adjusting the fiber geometry and the used materials accordingly. In addition to these advantages the PCF forms a convenient scaffold which ensures a constant core-to-core distance over several centimeters of propagation length, hardly achievable by usual liquid-filled capillaries put in close vicinity of each other.

The Kerr index of the highly nonlinear liquids, among them CS_{2}, CCl_{4}, nitrobenzene, or toluene, is up to 100 times larger than in the case of fused silica, which reduces the required nonlinear switching power so that low-power nonlinear coupling becomes possible.

Earlier experimental or theoretical investigations of nonlinear fiber couplers have mainly focused on a continuous-wave (cw) input [3, 5] or on nonlinear soliton switching between two waveguides made of fused silica [6, 7]. In the work presented in [8] and [9], the retarded response of fused silica due to the Raman effect and its influence on the coupling was investigated. However, when using highly nonlinear liquids as the carrier medium, their retarded responses differ considerably from each other and that of fused silica in strength as well as in their characteristic time scales [10]. Due to this reason a general analysis of the effects induced by the retarded response without focussing on a specific medium is desirable for the design of an optofluidic coupler.

In this paper we investigate the influence of the input pulse duration and the retarded material response on the properties of an ultrafast nonlinear optofluidic fiber coupler. We find that in particular the critical input peak power *P _{C}*, above which the coupling is suppressed, may increase dramatically, depending strongly on both parameters. We present a simple model capable to explain the findings.

By introducing a hypothetic medium with fixed dispersion and nonlinearity but with a variable retarded response we are able to separate the effects induced by the retarded response from other influences [10]. The nonlinear propagation in two identical nearby waveguides filled with this hypothetic medium is known to be well described by two coupled generalized nonlinear Schrödinger equations (NLSEs) [15]:

*A*(

_{j}*j*= 1,2) being the electric field envelope in the

*j*th waveguide,

*z*the propagation distance, and

*T*the time in the co-moving reference frame. ${\beta}_{k}=\frac{{\partial}^{k}\beta}{\partial {\omega}^{k}}{|}_{{\omega}_{0}}$ denote the dispersion coefficients with

*β*being the propagation constant.

*γ*=

*n*

_{2}

*ω*

_{0}/(

*cA*

_{eff}) represents the nonlinear parameter with

*n*

_{2}as the nonlinear index of refraction,

*ω*

_{0}the central frequency of the pulse,

*c*the vacuum speed of light and

*A*

_{eff}the effective mode area.

*κ*is the linear coupling strength and

*f*

_{j}_{(3–}

_{j}_{)}the cross-phase modulation coefficient. For simplicity we neglect the influence of noise. The NLSEs in this form are well known to accurately simulate the propagation of ultrafast pulses with durations as short as 10 fs, if sufficiently high orders for the dispersion coefficients

*β*are taken into account [15].

_{k}The response of the medium is governed via *R*(*t*), defined by

*h*, which we define with a unitless argument as

_{R}*N*

_{0}being a normalization constant and Θ(

*x*) the Heaviside function that ensures causality. The fraction of the retarded response is given by

*f*.

_{R}This phenomenological design of *R*(*t*) has been shown to represent a good approximation [10] to measured retarded responses of commonly used media (Fig. 1b), allowing us to model all regarded liquids with only one free parameter apart from the fraction *f _{R}* to keep the parameter space small. This free parameter we call

*t*, namely the characteristic time, and use it to scale the time axis.

_{R}The PCF in which we embed the hypothetic medium has a hole diameter of 2.7 *μ*m and a hole-to-hole distance of Λ = 5.6 *μ*m. This leads to a waveguide separation of
$\sqrt{3}\Lambda =9.7\mu \text{m}$. For the calculation of the dispersion coefficients [15] we use the Kerr-corrected refractive indices of CCl_{4} [16], a central wavelength of *λ*_{0} = 1040 nm and include the fiber geometry. This results in *β*_{2} = 5.5787 × 10^{−27} s^{2}/m, *β*_{3} = 9.6736 × 10^{−42} s^{3}/m, *β*_{4} = −3.8758 ×10^{−57} s^{4}/m, and *β*_{5} = 1.6628 × 10^{−72} s^{5}/m for the linear coefficients. Since *β*_{2} is positive, we operate in the normal dispersion regime, and soliton formation is not expected. With the Kerr index of CCl_{4} being *n*_{2} = 15 × 10^{−20} m^{2}/W [17], we obtain for the nonlinear parameter *γ* = 0.0265 (Wm)^{−1}. The coupling strength *κ* is set to 174.53 m^{−1}, leading to a linear coupling length *L _{C}* =

*π*/2

*κ*= 9 mm [3]. The propagation length

*z*

_{0}is set to 4

*L*= 36 mm. The cross-phase modulation coefficient becomes

_{C}*f*

_{j}_{(3–}

_{j}_{)}= 8 ×10

^{−4}[18].

In principle, the waveguides can also be operated in the anomalous dispersion regime. However, in that case due to the high nonlinearity of the regarded liquids, soliton fission and supercontinuum generation is likely. A full analysis of the propagation and switching behavior in the anomalous dispersion regime including the supercontinuum generation requires a more detailed discussion which we will present in a future publication.

To solve Eq. 1 we implement the symmetrized split-step Fourier method [15], which is well suited for the pulse durations and spectral widths we expect for the results. As the initial pulse we choose a sech-pulse given by

*P*

_{0}denotes the peak power and

*T*

_{0}the pulse width. As central wavelength we choose

*λ*

_{0}= 1040 nm for the following simulations.

Fig. 2 shows a selection of evolutions of the pulse energy contained in waveguide 1, normalized to the input pulse energy, for different input pulse durations and retarded response parameters. Common for all plots is the following behavior: Starting from low input powers, the coupling length increases for increasing input peak power. This corresponds to a decreasing coupling strength, which can be explained by a higher confinement and hence lower overlap of the fields. This higher confinement is caused by the increasing index contrast between the fiber core and cladding due to the Kerr effect. Once a certain input peak power is reached, namely the critical power *P _{C}*, the coupling is suppressed and the pulse energy remains in waveguide 1. In case of cw input, the critical power can be calculated by

*P̃*= 4

_{C}*κ*/

*γ*≈ 26.4 kW [3], which in contrast to the results of Fig. 2 is independent of

*T*

_{0}and the retarded response. We observe a shift to higher values of

*P*if the pulse duration

_{C}*T*

_{0}is decreased. Furthermore, the shift of the critical power to higher values becomes stronger for a more distinctive retarded response, i.e., for increasing

*f*and

_{R}*t*.

_{R}Performing more simulations with varying input pulse durations and retarded response parameters, we show in Fig. 3a the critical powers *P _{C}* extracted manually from the simulation results and the corresponding evolution plots. Clearly a strong deviation from the cw-case is observed.

*P*may exceed the cw-case

_{C}*P̃*by a factor of more than 3 for short pulses and high retardation.

_{C}To quantify our findings we expand the cw-case critical power by taking the retarded response into account:

Inserting Eq. 2, we obtain for the correction divisor

*τ*′ =

*t*′/

*T*

_{0},

*τ*=

*T*/

*T*

_{0}, and

*τ*=

_{R}*t*/

_{R}*T*

_{0}. We define the maximum of

*D*(

_{C}*τ*) as the appropriate correction divisor so that

*P*=

_{C}*P̃*/

_{C}*D*.

_{C}*D*can be derived easily by numerical integration. We note that

_{C}*t*and

_{R}*T*

_{0}enter

*D*only as ratio

_{C}*τ*. For

_{R}*f*= 0,

_{R}*t*= 0, or

_{R}*T*

_{0}→ ∞, the correction divisor becomes

*D*= 1, and hence the cw-case is reproduced. Fig. 3b shows

_{C}*P*obtained by this simple model, which exhibits an excellent agreement with the critical powers extracted by the simulations for a small retarded contribution, and gives still a good qualitative agreement for larger retarded contributions. We attribute the occurring deviations to the fact that our model does not include dispersion terms and self-steepening, and that it takes into account only the input field envelope at

_{C}*z*= 0.The correction divisor

*D*is plotted in dependence of

_{C}*t*/

_{R}*T*

_{0}and

*f*in Fig. 4.

_{R}To summarize, we have shown by numerical simulations that the input pulse duration as well as the retarded material response have a large influence on the properties of an ultrafast nonlinear optofluidic fiber coupler. In particular, the critical power may have to be corrected by a factor larger than 3 for commonly used laser sources and liquids. These findings are supported by a simple model and might aid further investigations on design and experimental realization of an ultrafast nonlinear optofluidic fiber coupler.

## Acknowledgments

The authors would like to thank the Landesgraduiertenförderung of Baden-Württemberg and DFG (Open Access Publishing) for support of this work.

## References and links

**1. **M. Vieweg, T. Gissibl, S. Pricking, B. J. Eggleton, D. C. Wu, B. T. Kuhlmey, and H. Giessen, “Ultrafast nonlinear optofluidics in selectively liquid-filled photonic crystal fibers,” Opt. Express **18**, 25232–25240 (2010). [CrossRef] [PubMed]

**2. **B. Kuhlmey, B. J. Eggleton, and D. K. Wu, “Fluid-filled solid-core photonic bandgap fibers,” J. Lightwave Technol. **27**, 1617–1630 (2009). [CrossRef]

**3. **St. M. Jensen, “The nonlinear coherent coupler,” *IEEE Trans. Microwave Theory Tech.*30, 1568–1571(1982). [CrossRef]

**4. **P. St. J. Russell, “Photonic crystal fibers,” Science **17**, 358–362 (2003). [CrossRef]

**5. **Y. V. Kartashov and V. A. Vysloukh, “Switching management in couplers with biharmonic longitudinal modulation of refractive index,” Opt. Lett. **34**, 3544–3546 (2009). [CrossRef] [PubMed]

**6. **S. Trillo, S. Wabnitz, E. M. Wright, and G. I. Stegeman, “Soliton switching in fiber nonlinear directional couplers,” Opt. Lett. **13**, 672–674 (1988). [CrossRef] [PubMed]

**7. **P. L. Chu, G. D. Peng, and B. A. Malomed, “Analytical solution to soliton switching in nonlinear twin-core fibers,” Opt. Lett. **18**, 328–330 (1993). [CrossRef] [PubMed]

**8. **B. A. Malomed, I. M. Skinner, and R. S. Tasgal, “Solitons in a nonlinear optical coupler in the presence of the Raman effect,” Opt. Commun. **139**, 247–251 (1997). [CrossRef]

**9. **Y. Wang and W. Wang, “Study of ultrafast pulse coupling dynamics considering retarded nonlinear response and self-steepening effects,” J. Lightwave Technol. **24**, 1041–1047 (2006). [CrossRef]

**10. **S. Pricking and H. Giessen, “Generalized retarded response of nonlinear media and its influence on soliton dynamics,” Opt. Express **19**, 2895–2903 (2011). [CrossRef] [PubMed]

**11. **R. Zhang, J. Teipel, and H. Giessen, “Theoretical design of a liquid-core photonic crystal fiber for supercontinuum generation,” Opt. Express **14**, 6800–6812 (2006). [CrossRef] [PubMed]

**12. **P. Wiewior and C. Radzewicz, “Dynamics of molecular liquids studied by femtosecond optical Kerr effect,” Opt. Appl. **30**, 103–120 (2000).

**13. **K. Itoh, Y. Toda, R. Morita, and M. Yamashita, “Coherent optical control of molecular motion uUsing polarized sequential pPulses,” Jpn. J. Appl. Phys. **43**6448–6451 (2004). [CrossRef]

**14. **T. F. Laurent, H. Hennig, N. P. Ernsting, and S. A. Kovalenko, “The ultrafast optical Kerr effect in liquid fluoroform: an estimate of the collision-induced contribution,” Phys. Chem. Chem. Phys. **2**, 2691–2697 (2000). [CrossRef]

**15. **G. P. Agrawal, *Nonlinear Fiber Optics* (Academic Press, 1995).

**16. **H. H. Marvin, “The selective transmission and the dispersion of the liquid chlorides,” Phys. Rev. **34**, 161–186 (1912).

**17. **P. P. Ho and R. R. Alfano, “Optical Kerr effect in liquids,” Phys. Rev. A **20**, 2170–2187 (1979). [CrossRef]

**18. **K. Okamoto, *Fundamentals of Optical Waveguides* (Academic Press, 2005).