ICTON 2006 157 We.A2.3 1-4244-0236-0/06/$20.00 ©2006 IEEE Polarisation Dependent Dynamics in Bulk Semiconductor Optical Amplifiers S. Philippe 1* , A.L. Bradley 1 , F. Surre 2 , P. Landais 2 , Member, IEEE, B. Kennedy 3 , M. Martinez-Rosas 4 1 School of Physics, Trinity College Dublin, Dublin, Ireland 2 School of Electronic Engineering, Dublin City University, Dublin, Ireland 3 Departamento de Ingenieria Electrica, Universidad de Santiago, Chile 4 Universidad Autonoma de Baja California, Ensenada, Mexico * Tel: +353 1 6082677, Fax: +353 1 6711759, Email: philipps@tcd.ie ABSTRACT A free space contra-propagation set-up is implemented and a dynamic pump-probe study of the eigenmodes of a SOA in the picosecond regime is undertaken. The time-resolved probe transmission of the TE and TM modes of the device is measured for the co and cross-polarised cases as a function of pump pulse energy. The relative contributions of interband and intraband processes to the gain compression are studied, as well as the coupling between the TE and TM modes. The high gain compression due to intraband effects observed in the TETE case is of particular interest for ultrafast all-optical switching. Keywords: semiconductor optical amplifier, gain, polarisation, pump-probe, optical switching. 1. INTRODUCTION Semiconductor optical amplifiers (SOAs) exhibit non-linearities that make them suitable for high-speed signal processing applications. They are polarisation sensitive, with different gain and refractive indices in the TE and TM modes [1]. Depending on the chosen configuration, polarisation can either be taken advantage of, using cross-gain and cross-phase modulation for all-optical wavelength conversion [2-5] or have to be minimized. They are however always present and should be well understood in order to optimize any SOA-based all-optical wavelength conversion scheme. A number of applications based on non-linear polarisation rotation have been presented, however fiberised set-ups are commonly used, making it difficult to set and maintain the state of polarisation of the signals. Conventional pump-probe experiments use a co-propagation configuration, where the pump and probe signals travels collinearly through the SOA and are separated at the output using a polarizer. In that case the pump and probe output states of polarisation have to be linear and orthogonal to each other, practically restricting the input states of polarisation to the eigenmodes of the device, pump TE probe TM and pump TM probe TE. No restrictions on the signals polarisation apply when a contra-propagation configuration is used and the probe and pump signal are separated by a beamsplitter. We propose to use a free space contra-propagation configuration and present a polarisation dependent pump-probe study of a bulk SOA in the picosecond regime. In this paper we focus on the behaviour of the SOA when light is injected along the eigenmodes of the device, with comparison of co and cross-polarised cases. 2. EXPERIMENT 2.1. Experimental set-up A free space contra-propagation configuration is used, allowing total control and preservation of the state of polarisation of the injected and collected signals, as shown on Fig. 1. The 1580 nm pulsed input is produced using a fibre based femtosecond laser, with a repetition rate of 82 MHz. The pulse width after dispersion in the fiberised output is 2.5 ps and a grating filter with a 1nm spectral width is used. The pulses are coherent and one is delayed with respect to the other using a variable delay stage. The polarisation of each beam is controlled independently with a quarter wave plate (QWP) and a half wave plate (HWP). Two beamsplitters allow us to monitor the input and output of each beam. The light is coupled in and out of the SOA by two anti reflection- coated aspheric lenses mounted on micro-control 3D translation stages. The device under test is a commercially available 1.5 mm long bulk InGaAsP/InP travelling wave SOA biased at 350 mA and temperature regulated at 20 °C by means of a Peltier cooler. An interference filter is used to isolate the probe signal from the broad ASE at 1580 nm and the signal is detected using a lock-in technique.