Open Access
Add to BibSonomyAbstract
We have investigated an all-optical set/reset and latching operation using a monolithically integrated InP-based semiconductor optical amplifier type Mach-Zehnder interferometer with a feedback loop. In simulation, operation conditions when both set and reset are possible was estimated for input light pulse with a FWHM of 31 and 12.5 ps, and the tolerance of the CW probe light and feedback loop loss becomes large with increasing the input light pulse power. In addition, the loop length could be longer than the distance of the light propagating in one bit pulse because of the longer carrier recovery time than one bit time duration. Moreover, we successfully achieved set/reset operation with 34- and 18-ps wide set/reset pulses.
©2012 Optical Society of America
1. Introduction
Owing to the continuous growth of the Internet, a higher-speed operation of photonic subsystems is required to process large amounts of optical packets. To meet this requirement, electrical processing is parallelized in present network components such as electrical routers; however, increased power consumption may become a significant issue in the near future. An all-optical flip-flop circuit is one of the viable functions, and it can be applied to latching operation used in the label processing and gating pulse generation for control light of the optical gates in all-optical packet switching systems(AOPSs) for realizing higher throughput and lower power consumption due to the elimination of O/E and E/O components, as well as the possibility of bit-rate and modulation format-transparent switching [1–5].
There have been several reports of the all-optical flip-flops; a two-segmented bistable laser [6, 7], a two-mode competition-type bistable laser [8, 9], a polarization switching bistable vertical-cavity surface emitting lasers(VCSEL) [10], and a semiconductor optical amplifier-type Mach-Zehnder interferometer(SOA-MZI) with a feedback loop [11, 12]. The use of a SOA-MZI with a feedback loop has many advantages such as high-speed, polarization insensitivity, wide wavelength bandwidth, and high extinction ratio. Flip-flop operation and its application to an all-optical packet forwarding gate in an optical packet switching system have been demonstrated [13]. However, the widths of the set and reset pulses are limited to about 30ns due to the difficulty in decreasing the length of the feedback loop. Therefore, monolithic integration of an SOA-MZI with a feedback loop is indispensable for improving the set and reset pulse widths.
In this paper, we report the investigation of both analytically and experimentally the monolithically integrated InP-based set/reset and latch using an SOA-MZI with a feedback loop.
This paper is organized as follows; In Section II, the operation principle is explained. In Section III, analytical investigation is described, and the operation conditions for achieving set and reset operations and the operation tolerance for several device and input signal conditions are clarified. In Section IV, the operation performance of the device is investigated experimentally. The operation for ultrashort set and reset light pulses is indicated. In Section V, this paper will be summarized.
2. Operation principle of an all-optical latch with a SOA-MZI with a feedback loop
Figures 1(a) and 1(b) show a schematic and a photograph of the structure of a SOA-MZI with a feedback loop. The device consists of InP-based SOAs(SOA4, SOA5) positioned in the MZI with a monolithically integrated feedback loop. SOAs not positioned in the MZI(SOA1, SOA2, SOA3, SOA6, SOA7, and SOA8) are used for loss compensation. SOA6 in the feedback loop also works for controlling the gain or attenuation. The lengths of SOA1 and SOA3 are 650μm, SOA2 is 700 μm, SOA4 and SOA5 are 2000 μm, and SOA6, SOA7, and SOA8 are 500 μm, respectively. The loop length is 3547 μm, and the time of flight is about 38 ps. It is a custom product of Alphion Corporation, and the device size was 5000μm × 8000μm.
Fig. 1 Structure of a monolithically integrated SOA-MZI with a feedback loop. (a) Schematic (b) photograph.
The operation principle is as follows; In the initial condition, a CW probe light fed back to the loop changes the refractive index in SOA4, and the phase condition in the MZI is maintained under destructive interference. When a single set pulse is fed into the input port (SOA8 side), the refractive index in SOA5 is changed. The phase of the probe light in the MZI arm is then changed by up to “π”, and the output is switched to the ON state completely. Because the probe light in the loop vanishes when the loop length is equal to the light propagating distance in the input pulse width, the “ON” state is maintained. When a 1-bit wide reset pulse is input(SOA1 side), the refractive index in SOA4 is changed and the output is switched to the OFF state. Accordingly, the output pulse width is basically designed to be equal to the interval of the set and reset input timing. However, when the carrier recovery time is longer than the input pulse width, the phase is gradually changed after the set and reset pulses go out from the SOA. Thus, the device could operate as a flip-flop even when the feedback loop is longer than the distance of the light propagating in the pulse width. In this situation, the fabrication tolerance could be relaxed and the device could work at a higher bit rate.
3. Simulation of the operation performance of a SOA-MZI with a feedback loop
3.1 Simulation model
First, we investigated the operation conditions of a SOA-MZI with a feedback loop through a numerical simulation. In the simulation, we used a multi-section model with a transfer matrix method (TMM) and rate equations expressed by Eqs. (1)–(5);
where Au(z,t) and Bu(z,t) are the electric field amplitudes in the forward and backward directions, respectively (u = 1,2 represents input signal, and the probe light), α is the α-parameter, Γ is the optical confinement factor, Ni is the carrier density in the i-th section, Su,i is the photon density (u is the same as above) in the i-th section, I is the injection current into the SOA, V is the volume of the active layer of the SOA, e is the electron charge, c1, c2,and c3 are the recombination constants, vg is the group velocity, Across is the cross- sectional area of the active layer, Eu is the photon energy, gm,i(Ni, λu) is the material gain in the i-th section for carrier density Ni and wavelength λu, gp is the peak gain, p and c are polynomial coefficients, gi is the net gain coefficient in the i-th section, λp is the gain peak wavelength, ε is the nonlinear gain compression coefficient, and αactive, αclad, and αscat are optical losses in the active and cladding layers and the scattering loss, respectively. The α-parameter is given bywhere dn/dN is the differential refractive index and dg/dN is the differential gain coefficient. Several parameters in the gain profile of the SOA were obtained by fitting the gain profile of the SOA with simulation profile. We used a model of gain profile of the SOA that took the variation of gain spectrum with carrier density into consideration [14]. The SOA parameters used for the simulation are summarized in Table 1 .
Table 1. Parameters used in simulations.
Also in simulations, the set and reset pulses are injected into the SOA-MZI with a feedback loop. We investigated the operation performance of the SOA-MZI with a feedback loop under various optical powers of the CW probe light and the attenuator losses.
The full width at half maximum(FWHM) of the input pulse of 31 and 12.5ps with wavelength of 1550 nm, and the CW probe light wavelength was 1555 nm. In case of a FWHM of 31ps, the peak power of set and reset pulses were varied from + 3dBm to + 9dBm. The CW probe light power was ranged from −9dBm to 0dBm. The loop attenuation was changed between 15 and 25dB. In case of a FWHM of 12.5ps, the peak power of set and reset pulses of + 6dBm, and + 12dBm were used. The CW probe light power was ranged from
−12dBm to −6dBm. The loop attenuation was changed between 20 and 30dB. The light propagating time in the feedback loop was varied from 10 to 40 ps. In addition, the bias currents of SOAs in the MZI were set to be 400 mA, and the carrier recovery time ranged from 20 to 100 ps, depending on CW probe light power.
3.2 Simulation results
In the first stage of simulations, the case of input pulse width of 31ps was investigated. The simulation results for the peak power of the set light pulse of + 3, + 6, and + 9dBm are plotted in Figs. 2 , 3 , 4 , respectively. In each figure, the results for a reset pulse peak power of + 3, + 6, and + 6 dBm are represented in (a) to (c). In the figures, blue rectangle, red circle, and green triangle represent the operation conditions when reset is not possible, both set and reset are possible, and set is not possible, respectively. As can be seen in the figures, we can find the simultaneous set and reset operation conditions in each case. However, the tolerance of the CW light power and loop attenuation increase with increasing the peak power of input pulses. Figures 4(d), 4(e), and 4(f) shows the waveforms of the simulation under the conditions indicated in (1), (2), and (3) in Fig. 4(c). When the loss in the feedback loop is too large, the device cannot be reset as indicated in Fig. 4(d). On the other hand, for low loss, the set operation is not possible as shown in Fig. 4(f). The tolerance of feedback power and CW probe light power are both found to be about 2.5dB as indicated in Fig. 4(c).
Fig. 2 Operation status of the SOA-MZI with a feedbacks of the SOA-MZI with a feedback loop with a set pulse width of 31ps and a peak power of + 3dBm. Each case indicates a reset pulse peak power of (a) + 3dBm (b) + 6dBm, and (c) + 9dBm, respectively. Blue rectangle, red circle, and green triangle represents the operation conditions when reset is not possible, both set and reset are possible, and set is not possible.
Fig. 3 Operation status of the SOA-MZI with a feedbacks of the SOA-MZI with a feedback loop with a set pulse width of 31ps and a peak power of + 6dBm. Each case indicates a reset pulse peak power of (a) + 3dBm (b) + 6dBm, and (c) + 9dBm, respectively. Blue rectangle, red circle, and green triangle represents the operation conditions when reset is not possible, both set and reset are possible, and set is not possible.
Fig. 4 Operation status of the SOA-MZI with a feedbacks of the SOA-MZI with a feedback loop with a set pulse width of 31ps and a peak power of + 9dBm. Each case indicates a reset pulse peak power of (a) + 3dBm (b) + 6dBm, and (c) + 9dBm, respectively. Waveforms of (d), (e), and (f) indicate the signal under the conditions of (1), (2), and (3) in (c). Blue rectangle, red circle, and green triangle represents the operation conditions when reset is not possible, both set and reset are possible, and set is not possible.
In the second stage of simulations, the case of input pulse width of 12.5ps was investigated. Figures 5 and 6 indicate the simulation results under the condition of the peak power of set/reset pulse of + 6 and + 12dBm, respectively. In case of + 6dBm input pulses, the operation condition when both set and reset are possible becomes narrow in CW probe light power and loop attenuation, although this condition is suitable for input pulse width of 31ps, because the pulse width becomes narrower and the peak power of input pulses should be increased for device operation. In case of + 12dBm input pulses, the operation range becomes large, and the tolerance of both the CW probe light power and the loss in the feedback loop are more than 1dB. In Figs. 5(a) to 5(c), loop length also varied from 10ps to 30ps. In Figs. 6(a) to 6(d), it varied from 10ps to 40ps. In principle, the loop length should be equal to the light propagation distance in one bit duration as mentioned. However, the carrier recovery time was ranged longer than the input pulse width, then the loop length can be tolerable in some range. As can be seen in Fig. 6, 40ps loop length could be achievable, that is almost three times longer than the light propagation distance in FWHM of input pulses. As indicated in Fig. 1, the fabricated device has a loop length of 38ps. Thus, this device has the potential for operating for input pulse with a FWHM of 12.5ps, which corresponds to the data at 40Gbps.
Fig. 5 Operation status of the SOA-MZI with a feedback of the SOA-MZI with a feedback loop with a set pulse width of 12.5ps and a peak power of + 6dBm. Each case indicates a loop delay of (a) 10ps (b) 20ps, and (c) 30ps, respectively. Blue rectangle, red circle, and green triangle represents the operation state of reset not possible, both set and reset possible, and set not possible.
Fig. 6 Operation status of the SOA-MZI with a feedback of the SOA-MZI with a feedback loop with a set pulse width of 12.5ps and a peak power of + 12dBm. Each case indicates a loop delay of (a) 10ps (b) 20ps, (c) 30ps, and (d) 40ps, respectively. Blue rectangle, red circle, and green triangle represents the operation state of reset not possible, both set and reset possible, and set not possible.
4. Experimental investigation of a SOA-MZI with a feedback loop
Next, we experimentally demonstrate the AOFF operation using a monolithically integrated SOA-MZI with a feedback loop. The experimental setup is shown in Fig. 7 . The input signal had an FWHM of 34ps. It was generated by modulating CW light from a tunable laser (T-LD1) with a LiNbO3 intensity modulator (LN) and an electroabsorption modulator (EAM) driven by a pulse pattern generator at a repetition frequency of 10GHz. A 256-bit signal period was used. The CW probe and set and reset pulses were injected into the AOFF from the ports connected with SOA2, SOA1 and SOA8, respectively. The CW probe light and the set and reset peak powers were set at 4.0, 5.9, and 4.4dBm, respectively. In addition, wavelengths of the probe light and set/reset pulses were set at 1555 and 1550nm. The bias currents for each SOA and phase shifter (PS) are shown in the inset of Fig. 7. The reset pulse was delayed from the set pulse by 16ns.
The experimental results are shown in Fig. 8 . The waveforms of the set, reset pulses, and the output are indicated in Figs. 8(a) to 8(c). As can be seen in Fig. 8(c), set/reset and latching operation using a monolithically integrated SOA-MZI with a feedback loop was successfully achieved. An extinction ratio of around 6dB and a guard time of around 16ns, which is equal to the interval between the set and reset input timings, were observed. The switching times for the set and reset operations required to reach a stable power level after a several cycles of feedback inputs are about 300 and 500ps, respectively.
Fig. 8 Waveforms of all-optical set/reset and latching operation. (left) experimental results (a) set pulse (b) reset pulse (c) output of a SOA-MZI with a feedback loop (right) simulation results (d) set pulse (e) reset pulse (f) output of a SOA-MZI with a feedback loop
We also simulated the device operation using the same method mentioned in Section III. The parameters selected for the simulation are similar to the values used in the experiments. The simulation results are in good agreement with the experimental results as shown in Figs. 8(d) to 8(f).
Next, we changed the CW probe light power, and the tolerance of the set/reset operation was investigated. The results are shown in Figs. 9(a) to 9(e). The CW light power was changed from + 0.4dBm to + 4.9dBm. As can be seen in the figures, the device cannot be set under insufficient CW probe light power, and this agree well with simulation as indicated in Fig. 4(f). The tolerance against the CW probe light power is about 1dB. We also varied the driving current of SOA6, the SOA in the feedback loop, between 5.0mA and 25.0mA. The results are shown in Figs. 10(a) to 10(d). From these figures, set operation is not possible under small loop attenuation. On the other hand, reset operation is not possible under large loop attenuation. These also agree well with simulation qualitatively.
Fig. 9 Output waveforms of a SOA-MZI with a feedback loop with a CW probe power of (a) + 4.9dBm, (b) + 4.4dBm, (c) + 3.9dBm, (d) + 3.4dBm, and (e) + 0.4dBm.
Fig. 10 Output waveforms of a SOA-MZI with a feedback loop with a SOA6 bias current of (a) 5.0mA, (b) 10.0mA, (c) 20.0mA, and (d) 25.0mA.
Finally, we investigated the set/reset operation of the device for input pulse with a FWHM of 18ps. The experimental setup is almost the same as Fig. 7. Set pulses were generated by modulating CW probe light from a tunable laser (T-LD1) with a LiNbO3 intensity modulator (LN-IM) and an electroabsorption modulator (EAM) driven by a pulse pattern generator driven at 40GHz. A 512-bit signal period was used. The CW probe light and the set and reset peak powers were set at 5.8, 13.0, and 13.0 dBm, respectively. In addition, wavelengths of the CW probe light and set/reset pulses were set at 1555 and 1550 nm, respectively. The reset pulse was delayed from the set pulse by 10 ns.
The experimental results are shown in Figs. 11(a) to 11(f). The waveforms of the set pulse (Fig. 11(a)), reset pulses (Fig. 11(b)), and output (Fig. 11(c)) are shown. As indicated in these figures, the device could work with an input pulse width of 18 ps, which is half of the light propagating time of the feedback loop with a length of 38 ps. An extinction ratio of 5.5 dB was observed. Magnified images of the set/reset pulses, leading and trailing edges are also shown in Figs. 11(d), 11(e) and 11(f), respectively. The switching times for the set and reset operations required to reach a stable power level were about 200 and 400 ps, respectively.
Fig. 11 Experimental results for a input pulse width of 18ps. (a) Set pulse (b) reset pulse (c) output from a SOA-MZI with a feedback loop (d) magnified image of a set/reset pulse (e) leading edge (f) trailing edge.
7. Conclusion
In conclusion, we have investigated the operation performance of a monolithically integrated InP-based SOA-MZI with a feedback loop. The device could work for input set and reset pulses with a FWHM from 31ps to 12.5ps through simulations. The simultaneous set and reset operation conditions have the tolerance within more than 1dB against CW probe light power and loss attenuation in the feedback loop. The loop length could be designed to be three time longer than the light propagation distance in one bit duration because of the slow carrier recovery time. In experiments, we have confirmed that the operation dependence against CW probe light power and loss attenuation agree well with simulation ones qualitatively. Set/reset and latching operation was achieved for input pulse with a FWHM of 18ps.
Acknowledgments
We would like to thank Prof. Emeritus K. Iga, Prof. F. Koyama and Assoc. Prof. T. Miyamoto for their encouragements and discussions. This work was partially supported by “Innovation for New-Generation Optical Communications - Based on Photonic Device Breakthrough –“ by Ministry of education, culture, sports, science, and technology Grant in Aid # 17068007
References and links
1. D. Blumenthal, P. Prucnal, and J. Sauer, “Photonic packet switches: Architectures and experimental implementations,” Proc. IEEE 82(11), 1650–1667 (1994). [CrossRef]
2. S. J. B. Yoo, H. J. Lee, Z. Pan, J. Cao, Y. Zhang, K. Okamoto, and S. Kamei, “Rapidly switching all-optical packet routing system with optical-label swapping incorporating tunable wavelength conversion and a uniform-loss cyclic frequency AWGR,” IEEE Photon. Technol. Lett. 14(8), 1211–1213 (2002). [CrossRef]
3. R. Takahashi and H. Suzuki, “1-Tb/s 16-b all-optical serial-to-parallel conversion using a surface-reflection optical switch,” IEEE Photon. Technol. Lett. 15(2), 287–289 (2003). [CrossRef]
4. N. Wada, H. Harai, and F. Kubota, “Optical packet switching network based on ultra-fast optical code label processing,” IEICE Trans. Electron. E87-C(7), 1090–1096 (2004).
5. H. Furukawa, N. Wada, and T. Miyazaki, “640 Gbit/s (64-wavelength 10 Gbit/s) data-rate wide-colored NRZ-DPSK optical packet switching and buffering demonstration,” J. Lightwave Technol. 28(4), 336–343 (2010). [CrossRef]
6. H. Kawaguchi, Bistabilities and Nonlinearities in Laser Diode (Artech House Optoelectronics Library, Boston, MA, 1994).
7. H. Uenohara, Y. Kawamura, and H. Iwamura, “Long wavelength multiple quantum well voltage-controlled bistable laser diodes,” IEEE J. Quantum Electron. 31(12), 2142–2147 (1995). [CrossRef]
8. K. Takeda, M. Takenaka, T. Tanemura, and Y. Nakano, “Experimental study on wavelength tenability of all-optical flip-flop based on multimode-interference bistable laser diode,” IEEE Photon. J. 1(1), 40–47 (2009). [CrossRef]
9. R. Kumar, K. Huybrechts, L. Liu, T. Spuessens, G. Roelkens, E.-J. Geluk, T. de Vries, P. Regreny, D. Van Thourhout, R. Bates, and G. Morthier, “An ultra-small, low-power all-optical flip-flop memory on a silicon chip,” in Conference on Optical Fiber Communication Conference2010(OFC2010), Technical digest (CD) (Optical Society of America 2010), paper OTuN7.
10. T. Mori, Y. Yamayoshi, and H. Kawaguchi, “Low-switching-energy and high-repetition-frequency all-optical flip-flop operations of a polarization bistable vertical-cavity surface-emitting laser,” Appl. Phys. Lett. 88(10), 101102 (2006). [CrossRef]
11. R. Clavero, F. Ramos, J. M. Martinez, and J. Marti, “All-optical flip-flop based on a single SOA-MZI,” IEEE Photon. Technol. Lett. 17(4), 843–845 (2005). [CrossRef]
12. K. Vyrsokinos, P. Bakopoulos, D. Fitsios, T. Alexoudi, D. Apostolopoulos, H. Avramopoulos, A. Miliou, and N. Pleros, “All-optical T flip flop using a single SOA-MZI and a feedback loop,” in Proceeding of 37th European Conference and Exhibition on Optical Communication (ECOC2011), paper We.10.P1.37.
13. H. Brahami, M. Bougioukos, M. Menif, A. Maziotid, C. Stamatiadis, Ch. Kouloumentas, D. Apostolopoulos, H. Avramopoulos, and D. Erasme, “Experimental Demonstration of an All-Optical Packet Forwarding Gate Based on a Single SOA-MZI at 40Gb/s,” in Conference on Optical Fiber Communication Conference2011(OFC2011), Technical digest (CD) (Optical Society of America 2011),paper OMK5.
14. S. Shimizu and H. Uenohara, “A proposal of a novel gain profile model of multi-quantum-well semiconductor optical amplifiers,” Jpn. J. Appl. Phys. 49(3), 030204 (2010). [CrossRef]
