²ÝÁñÉçÇø

Figure 2 shows the simulation flow used to analyze these devices. The structure is segmented into a number of slices that are all kept at the same voltage. For each voltage simulated, a mode is calculated for the first slice and then an electrical simulation is performed using that mode profile. The gain at the given voltage produces a complex index profile that is passed back to the BeamPROP BPM solver, which then propagates the field (which may grow or decay depending on the gain) for the length of the slice. This field is then passed to the electrical simulation for the next slice, and so on until the end of the device. At that point, the field is reflected according to the specified facet reflectivities and the process repeats in the backward direction. These bidirectional loops are repeated, while adjusting the photon number, until the field at each facet has converged. The solution for that voltage bias is then known. The photon number determines the optical power emitted and the total current is simply the aggregated current from all the slices. The L-I-V curves can then be constructed from the solutions for a sequence of voltages.  

Simulation results for the far field are shown in Figure 4. The angular far fields along x and y are shown as contour plots on the left of the figure, while the cross-cuts along x are shown in the right. The results for the device without beam spoilers are shown on the top of the figure, while the results for the device with beam spoilers are shown on the bottom. It is clear that the beam quality is significantly enhanced (more Gaussian) when beam spoilers are included in the design.