Résumé
The monolithic integration of III-V lasers on Si has been explored for the past decades. It would allow the low-cost mass production of devices on large silicon wafers, providing high scalability[1] and reducing III-V material waste. However, several challenges need to be overcome in order to grow high quality III-V layers. The crystallographic polarity difference between the materials leads to the formation of antiphase boundaries[2]. These consist of III-III or V-V bonds that introduce a local charge and thus form efficient vertical conduction paths that prevent devices from operating. However, they can be eliminated either by using large miscut substrates[2] or by careful surface preparation of the Si substrate[3], [4]. In addition, the difference in lattice parameters and thermal expansion coefficients leads to the formation of a high density of dislocations[5], a linear defect created above the critical thickness to relieve the strain in the layers. The dangling bonds at the dislocation core introduce deep levels into the bandgap[6] that form scattering centers and act as Shockley-Read-Hall centers, significantly reducing the electron mobility[7] and the minority carrier lifetime[8]. For example, quantum well laser diodes grown on Si show a significant increase in threshold current, at least three times higher than on the native substrate[9], and the typical laser lifetime is estimated to be less than a week. These results illustrate the direct effect of dislocations on the degradation of the device performance. Several strategies have been developed to overcome this problem, the most common of which is to combine TDD reduction using filter layers and annealing cycles with the use of quantum dots. This allows performance similar to that of lasers grown on their native substrate[10], but this technique requires the growth of thick and complex buffer layers. One way to overcome this limitation is to use active zone designs that are dislocation tolerant, such as interband cascade lasers[11] and quantum cascade lasers[12]. In this talk, I will discuss the effect of dislocations on laser performance, and present some of our work with the new active zone design.