1 All-optical clock division at 40 GHz using a semiconductor amplifier nonlinear interferometer R. J. Manning, I. D. Phillips, A. D. Ellis, A. E. Kelly, A. J. Poustie, K.J. Blow BT Laboratories, Martlesham Heath, Ipswich, Suffolk IP5 3RE UK Tel: Fax: Abstract We demonstrate all-optical clock division of a 40 GHz pulse train, using a semiconductor nonlinear interferometer with feedback. Blocks of pulses are also output if operating conditions are chosen appropriately. We also observed the dynamical evolution of the clock divided train.
2 Introduction Semiconductor optical amplifiers (SOAs) have become widely used in optical communications research for high speed switching, in particular demultiplexing [ 1, 2 ]. The so-called TOAD topology [ 3 ] has become widely used in this context, and the use of SOAs in a loop mirrorand has also formed the basis of demonstrations of more sophisticated all-optical logic functionality, such as an all-optical memory [ 4 ], a shift register with inverter [ 5 ], and a half adder [ 6 ]. We have recently reported spontaneous alloptical clock division at 10GHz and 20 GHz using a TOAD arrangement with optical feedback, and have also shown that this unit can also act as an all-optical shift register with inverter at the same repetition rates[ 7 ], where the output consists of alternate blocks of ones (pulses), and zeros. Here we show that the spontaneous clock division phenomenon is scaleable to 40 GHz repetition rates, using a derivative of the TOAD topology, namely a UNI (Ultrafast Nonlinear Interferometer) [ 8 ]. We also experimentally follow the evolution of the clock divided output with number of circulations in the loop. Experimental A schematic of the experimental arrangement is shown in Figure 1. A similar arrangement has been used recently to demonstrate a 10 Gbit/s all-optical memory [ 9 ]. The pulse source used was a 10 GHz external cavity modelocked semiconductor laser (ECMLL), which produced 3.5 ps pulses (FWHM) at a wavelength of 1550nm. These pulses were passively multiplexed in optical fibre to a give a 40 GHz pulse stream that was amplified using an Erbium doped fibre amplifier (EDFA) and input to the UNI as probe pulses. The unswitched probe pulses from the UNI passed through ~1 km of
3 dispersion shifted fibre (having a dispersion zero at a wavelength of ~1550nm), an acoustic-optic modulator (AOM) (usually transmitting) and an EDFA. After amplification, these pulses were fed back to the UNI as switching pulses. The UNI is essentially a Mach-Zehnder interferometer, which is made using a single fibre for both arms by exploiting polarisation diversity. In these experiments, the UNI was used in a counter-propagating configuration similar to that described in references [ 10, 11 ]. Probe pulses input to the UNI were launched at 45 to the axes of polarisation maintaining (PM) fibre and split into orthogonally polarised pulse pairs. The pulse pair, separated by 15ps after propagation through 7 m of fibre due to polarisation mode dispersion (PMD), and was input to a polarisation insensitive SOA with a mean power of ~ 3 dbm. After the SOA the pulse pair was launched at - 45 into a second 7m length of PM fibre, so the pulse pair suffered a delay of the opposite sense and therefore recombined. The resultant pulse passed through a fibre polariser P. Switching pulses were input via the 3dB coupler with a mean power of ~ 7dBm and were counter-propagating to the probe pulse pair. The switching pulse causes a change in gain and hence refractive index of the SOA and this affects the relative phase of the pulse pair. If the phase difference is π then a polarisation rotation of 90 occurs at the polariser of the UNI, and full switching occurs. The SOA used here had a gain recovery time of 80 ps at a current bias of 400 ma, and an alpha parameter of ~9 at 1550nm [ 12 ]. The SOA was 1 mm long, which meant that only one pulse pair was present in the SOA at any one time. This was important, since it meant one pulse pair was affected by only one switching pulse. Results and Discussion
4 Depending on the arrival time of the switching pulses with respect to the probe pulses, we were able to observe either block behaviour or clock division [Error: Reference source not found,10]. The driving electronics for the AOM allowed us to start a sequence and allow it to run for up to 185 ms, or ~35,000 circulations. We were able to observe eye diagrams for everyany individual round trip of the system, and hence time resolve the clock division evolution. Figure 2 shows the output pulse train after ~20 circulations, where clock division is virtually complete. Figure 3 is a 3D plot showing the input 40GHz pulse train and its observed evolution as it circulated in the shift register. The clock division can be seen to be complete after ~100µs (i.e. 20 circulations). The evolution shows that clock division occurs over a fairly small number of circulations. Existing theory [ 13 ] would suggest that, for an input pulse train of ~10 5 pulses, such as we have here for a feedback path of ~1km, the number of circulations required for clock division to evolve would be of the same order. However, further modelling shows that if there is even a small amplitude modulation (~4%) on one channel, the number of circulations required is drastically reduced. The 40 GHz input used does have some small amplitude modulation, and it is believed that this is responsible for the very rapid evolution we observe here. Further experiments using a 40 GHz fibre ring laser source were performed in which the 1 km length of fibre was removed. Here there was no amplitude modulation on the channels and clock division was again observed (the evolution was not monitored). We were also able to observe blocks of 40 GHz pulses as output, as first shown by Hall et al. , by delaying the arrival time of the switching pulses by half a bit period [Error: Reference source not found]. The block output is shown in Figure 4. The block period of
5 ~300ns corresponded to the time taken to traverse the feedback path, which was mostly made up of fibre for the EDFA. Conclusions We have shown that a UNI combined with optical feedback can be used as an all-optical clock divider or an all-optical shift register at 40 GHz. The principles of operation described in reference  still apply at these higher repetition rates.
6 10 GHz ECMLL t=3.5 ps λ= 1550 nm AOM1 3dB Lightwave Converter Sampling scope GHz Mux ~1 km dispersion shifted fibre Variable Delay Isolator 7 m PM Fibre PMD=15ps SOA 7 m PM Fibre PMD=15ps EDFA Filter 45 Splice 3dB P Ultrafast Nonlinear Interferometer (UNI) Variable Delay
7 Pulse Amp Time(ps)
8 1 0.8 Pulse Amp Circulation Time(ps) time (us)
11 1 References  R.J. Manning, Ellis, A.D., Poustie, A.J. and Blow, K.J.: Semiconductor laser amplifiers for ultrafast all-optical signal processing, J. Opt. Soc. Am. B, , pp  Hess, R., Caraccia-Gross, M., Vogt, W., Gamoer, E., Besse, P.A., Duelk, M., Gini, E., Melchior, H., Mikkelsen, B., Vaa, M., Jepsen, K.E., Stubjkaer, K.E., and Bouchoule, S.: All-Optical Demultiplexing of 80 to 10 Gb/s Signals with Monolithic Integrated High Performance Mach- Zehnder Interferometer, IEEE Phot. Technol. Lett, , pp  Sokoloff, J.P., Prucnal, P.R., Glesk, I. and Kane, M.: A terahertz optical asymmetric demultiplexer (TOAD), IEEE Photonics Technol. Lett., , pp  Poustie, A.J., Blow, K.J., Manning, R.J. : All-optical regenerative memory for long term data storage, Opt. Commun., 1997, 140, pp  Poustie, A.J., Manning, R.J., and Blow, K.J., : All-optical circulating shift register using a semiconductor optical amplifier in a fibre loop mirror, Electron. Lett., 1996, 32, pp  Poustie, A.J., Blow, K.J., Kelly, A.E., Manning, R.J., : All-optical binary half adder, Opt. Commun., 1998, 156, pp  R.J. Manning, A.E. Kelly, K.J. Blow, A.J. Poustie, D. Nesset, : Semiconductor optical amplifier based nonlinear optical loop mirror with feedback: Two modes of operation at high switching rates, Opt. Commun., 1998, 157, pp  Patel, N.S., Rauschenbach, K.A., Hall, K.L., 40 Gbit/s demultiplexing using an ultrafast nonlinear interferometer, IEEE Phot. Tech. Lett., 1996, 8, pp  Manning, R.J., Phillips, I.D., Ellis, A.D., Kelly, A.E., Poustie, A.J., Blow, K.J.: 10 Gbit/s alloptical regenerative memory using a single SOA based logic gate, Electron. Lett., 1999, 35, pp158-
12  Hall, K.L., Donnelly, J.P., Groves, S.H., Fennelly, C.I., Bailey, R.A., and Napoleone, A.: 40 Gbit/s all-optical circulating shift register with an inverter, Opt. Lett., 1997, 22, pp  Phillips, I.D., Ellis, A.D., Thiele, H.J., Manning, R.J., and Kelly, A.E.: 40 Gbit/s all-optical data regeneration and demultiplexing with long pattern lengths using a semiconductor nonlinear interferometer, Electron. Lett., 1998, 34, pp  Manning, R.J., Kelly, A.E., Poustie, A.J., and Blow, K.J.: Wavelength dependence of switching contrast ratio of semiconductor optical amplifier-based nonlinear loop mirror, Electron. Lett., 1998, 34, pp  Blow, K.J., Manning, R.J., Poustie, A.J., Nonlinear optical loop mirrors with feedback and a slow nonlinearity' Opt. Commun., 1997, 134, pp43-48