1. Introduction

Ultrafast all-optical switching is an important technique in future ultrahigh speed optical communication systems. Several methods of pulse switching have been proposed over the years. An optical fiber or semiconductor optical amplifier (SOA) has been used as the nonlinear optical device for ultrafast all optical switching in a nonlinear optical loop mirror (NOLM) [1,2] and Mach-Zehnder interferometer [3]. All optical switching by use of cross phase modulation (XPM) and four wave mixing (FWM) has also been demonstrated [4,5].

The interaction phenomena occur when the optical pulses collide in the optical fibers. Previously, the soliton trapping and trapped pulse generation were discovered as trapping phenomena between two orthogonally polarized components in birefringent optical fibers [6,7]. Recently, we discovered the phenomenon of pulse trapping by a femtosecond soliton pulse across zero dispersion wavelength [8,9]. The optical pulse in the normal-dispersion region is trapped by an ultrashort soliton pulse in the anomalous dispersion region. The wavelength of the trapped pulse is shifted to satisfy the group-velocity matching and the soliton and trapped pulses copropagate along the fiber. It is expected that the phenomenon of pulse trapping will be useful for wavelength control and optical switching of optical pulses in the normal dispersion region.

In this paper we propose ultrafast all optical switching by use of pulse trapping across zero dispersion wavelength. The high repetition rate signal pulse train in the normal dispersion region is demultiplexed by the soliton pulse in the anomalous dispersion region. In this technique, the ultrafast all optical switching can be demonstrated using a few meter long fiber and wavelength filter. The spectrogram of optical switching is directly observed using the cross-correlation frequency resolved optical gating (X-FROG) technique [10]. All optical switching is experimentally demonstrated for a 0.67 THz pulse train. To the best of our knowledge, this is the fastest optical switching in optical fibers. Almost ~100 % extinction ratio is obtained for pulse switching. The optical switching is numerically demonstrated using strict coupled nonlinear Schrödinger equations and the characteristics of pulse switching are analyzed. The numerical results are almost in agreement with the experimental ones. In the numerical analysis, ultrafast switching as fast as 1 THz is demonstrated.

2. Experimental

Experimental setup is shown in Fig. 1. As the pump light source, a passively mode-locked Erdoped fiber laser is used. It generates about 110 fs sech² like ultrashort pulses at repetition frequency of 48 MHz. The center wavelength is about 1.56 µm. The output of the laser is divided into two optical axes. One of them is coupled into diameter reduced type polarization maintaining fiber (PMF) and the wavelength tunable soliton pulse is generated [11]. At the output of PMF, only the wavelength tunable soliton pulse is selected and is used as the control pulse. The temporal width is 110 fs at full width at half maximum (FWHM) and the pedestal free transform limited sech² pulse is obtained. The other pulse is coupled into cascadely spliced polarization maintaining dispersion-shifted fiber (PM-DSF). In this fiber, the first PMDSF is fusion spliced to the same kind of second fiber whose birefringent axis is 45 degree inclined from that of the first one. At the fiber input, the polarization direction of the input pulse is inclined from the birefringent axis by 45 degree. The orthogonally polarized components propagate independently due to the birefringence and they generate the soliton and anti-stokes pulses, respectively [12]. In the second fiber, the propagating pulses are divided into two polarization components due to the birefringence and we can generate the train of four pulses. At the output of PM-DSF, only the generated train of anti-stokes pulses is picked off using the high pass filter and is used as the signal pulse train.

 

Fig. 1. Experimental setup of ultrafast all optical switching using pulse trapping by ultrashort soliton pulse across zero-dispersion wavelength. LPF: low pass filter; HPF: high pass filter; BS: beam splitter.

Download Full Size | PDF

The temporal difference between the signal and soliton pulses is controlled by the corner mirror. Then the soliton pulse and signal pulse train are coupled into 10-m-long polarization maintaining highly nonlinear dispersion shifted fiber (PM-HN-DSF) [13]. The mode field diameter is 3.7 µm, the magnitude of nonlinearity γ=20 W^-1km^-1, and the second- and thirdorder dispersions are β₂=-2 ps²/km and β₃=0.01 ps³/km at a wavelength of 1.55 µm. In this fiber, we can demonstrate the pulse switching by use of pulse trapping. The output pulses are observed using the optical spectrum analyzer and cross-correlation frequency resolved optical gating (X-FROG) system [10].

 

Fig. 2. Optical spectra of output pulses from 10-m-long PM-HN-DSF for all optical switching by use of pulse trapping. The blue and red lines show the optical spectra when the optical power of the soliton pulse is 0.5 and 1.8 mW, respectively.

Download Full Size | PDF

Figure 2 shows the observed optical spectra at the input and output of PM-HN-DSF. At the fiber input, the wavelengths of the soliton and signal pulses are set to be 1.65 µm and 1.4 µm, respectively. The average fiber input power of signal pulse train is 60 µW and that of soliton pulse is varied from zero to 1.8 mW. When the fiber input power of the soliton pulse exceeds the threshold power of the pulse breakup, the wavelength of the soliton pulse is shifted toward the longer wavelength side due to the soliton self-frequency shift (SSFS) in the propagation along the fiber [14]. For the signal pulse, when the initial temporal separation between soliton and signal pulses is adjusted, only one of the pulses in the pulse train is trapped by the Raman shifted soliton pulse [8,9]. The wavelength of the trapped pulse is blue shifted to satisfy the group velocity matching condition. In Fig. 2, the wavelength of the trapped pulse is shifted to be 1300 nm and those of the untrapped pulses are unchanged. The magnitude of wavelength shift of trapped pulse is increased as the fiber input power of the soliton pulse is increased.

Figure 3 shows the observed spectrogram of output pulses from 10-m-long PM-HN-DSF. The horizontal axis represents the time and the vertical one represents the wavelength of observed sum frequency generation (SFG) signals, respectively. When the pulse trapping does not occur, the four pulses with temporal separation of 1.5 ps are clearly observed. The linear chirp due to the normal dispersion is observed obviously. Figure 3(b) shows the spectrogram of output pulses when only the second pulse is trapped by the soliton pulse. The trapped pulse copropagates with the soliton pulse and the leading part is overlapped with the trailing part of the soliton pulse. The wavelength of the trapped pulse is blue shifted due to the cross phase modulation. We can see that the trapping efficiency is as high as ~100%. By adjusting the initial temporal difference between the soliton pulse and signal pulses, we can trap only the arbitrary one pulse among the four pulses. As shown in Fig. 2, since the optical spectra of trapped pulse is distinctly separated from those of the untrapped signal pulses, we can pick off only the trapped pulse from the pulse train.

It is interesting to note that although the soliton pulse also overlaps with the third and fourth pulses, these pulses are not trapped by the soliton pulse. The reason is considered as follows. The soliton pulse is rapidly red-shifted in the propagation along the fiber. The trapping efficiency depends on the group velocity difference between the soliton and signal pulses [8]. When the soliton pulse overlaps with the third signal pulse, the magnitude of group velocity difference has been increased and the third pulse is not trapped by the soliton pulse. It is estimated that the temporal and wavelength windows of the pulse trapping are less than 1 ps and 20 nm. Thus we can trap only one pulse whose temporal and wavelength conditions are satisfied. The precise characteristics of pulse trapping are now under investigation.

 

Fig. 3. Observed spectrogram of output pulses from PM-HN-DSF using X-FROG technique when one of the signal pulses are (a) not trapped and (b) trapped.

Download Full Size | PDF

3. Numerical

Next, we have numerically analyzed the characteristics of all optical switching using pulse trapping. We use the strict coupled nonlinear Schrödinger equations [9] and simulate the pulse switching. In the numerical simulation, the parameters are set to be the same as those in the experiment. The 100 fs transform limited sech² pulse is assumed for the soliton pulse. For the signal pulses, 400 fs transform limited sech2 pulses with temporal separation of 1.5 ps are assumed for simplicity. The center wavelength of the soliton pulse is 1.65 µm and those of the signal pulses are set to be 1.4 µm at the fiber input. The peak powers are 175 W for the soliton pulse and 1 W for the signal pulses. In this simulation, we assume that the center of the second signal pulse (trapped pulse) is delayed from that of the soliton pulse temporally by 250 fs at the fiber input.

Figure 4 shows the variation of the optical spectra for optical switching. The soliton pulse is red-shifted due to the SSFS and the trapped pulse is blue-shifted to satisfy the group velocity matching. For the signal pulse, only the second pulse among the four pulses is trapped by the soliton pulse and the optical spectrum of the trapped pulse is distinctly blueshifted through the cross phase modulation and is separated from those of untrapped ones.

 

Fig. 4. Numerically obtained optical spectra for all optical switching. The blue and red lines show the optical spectra at the input and output of 15-m-long PM-HN-DSF, respectively.

Download Full Size | PDF

 

Fig. 5. (236KB) Numerically calculated spectrogram for optical switching for 0.67 THz pulse train. In order to clarify the behavior of pulse trapping, the signal pulses are enlarged.

Download Full Size | PDF

Figure 5 shows the numerical results of all optical switching using pulse trapping. The spectrogram is obtained from the numerical results under assumption of the polarization gate (PG)-FROG measurement [15]. The observed spectrogram is almost the same as that of the X-FROG measurement used for the experiment. In the PG-FROG measurement, we can observe the actual wavelength of detected signals. The soliton pulse at fiber input is assumed to be the probe pulse. The numerical results are well in agreement with the experimental ones. We can see that the second signal pulse is trapped by the soliton pulse and they copropagate along the fiber. The leading part of the signal pulse is overlapped with the trailing edge of the soliton pulse and the condition of group velocity matching is satisfied. The high conversion efficiencies are obtained for both SSFS and pulse trapping. From Fig. 5, we can see that when the soliton pulse overlaps with the third and fourth pulses in the propagation along the fiber, these pulses suffer the cross phase modulation and the spectrum shapes are modulated. However, the spectral shapes of these pulses perfectly recover after the passage of the soliton pulse.

 

Fig. 6. (167KB) Spectrogram of optical switching when the second signal pulse does not exist. The signal pulses are enlarged.

Download Full Size | PDF

Figure 6 shows the spectrogram when the second signal pulse does not exist. In this case, although the third and fourth pulses suffer the cross phase modulation and the spectral shapes are obviously changed when the temporal overlapping with the soliton pulse occurs, no optical component is trapped by the soliton pulse. From Figs. 5 and 6, we can confirm that the optical pulse is trapped with high extinction ratio when the second pulse exists and nothing is trapped when the second one does not exist. Thus we can demonstrate the high performance all optical switching by use of pulse trapping. When the temporal separation between the signal pulses is changed to be 1 ps, all optical switching is also demonstrated numerically. This corresponds to 1 THz ultrafast all optical switching.

4. Conclusion

In this paper, we have demonstrated all optical switching by use of pulse trapping across zero dispersion wavelength both experimentally and numerically. A train of four pulses with temporal separation of 1.5 ps is generated by the use of birefringence in PM fibers and is used as the signal pulses. In the optical fiber, only one of the signal pulses is trapped by the soliton pulse and the wavelength of the trapped pulse is blue shifted. The spectrogram of optical switching is observed by use of the X-FROG technique and the optical switching is confirmed directly. The repetition frequency corresponds to 0.67 THz. To the best of our knowledge, this is the fastest all optical switching in optical fibers. We can pick off this trapped pulse easily by use of a wavelength filter such as a fiber Bragg grating. It is interesting to note that although the other pulses are also overlapped with the soliton pulse, they are not trapped except for the trapped pulse. In the numerical simulation, the coupled strict nonlinear Schrödinger equations are used and all optical switching is demonstrated precisely. The numerical results are almost in agreement with the experimental ones. The all optical switching for a 1 THz pulse train is also confirmed and we can demonstrate ultrafast all optical switching by use the of pulse trapping.