A simulative approach to tailoring the energy deposition during single pulse laser annealing or doping

Abstract

Some high-efficiency solar cell concepts, which are thought to replace standard solar cells in future, require large shallow doped zones of both polarities on one side of the cell (for example interdigitated back-contact silicon solar cells IBC). These zones are difficult and expensive to diffuse by furnace diffusion, and thus, other production methods are searched for, one being laser doping. Laser doping has already been researched on as a replacement for furnace diffusion of large-area emitters in the late 1970ies and early 1980ies and was revived for the doping of selective emitters in the '00s.<br />However, the requirements for doping large areas for high efficiency solar cell concepts have not been fulfilled yet, making further improvements necessary. The creation of e.g. vacancy-related defects induced during laser doping has been commonly associated with the high crystallisation velocity after the end of the pulse, resulting in low cell efficiencies. However, only the recent introduction of new laser devices allows temporal pulse shaping that makes an exact control of the crystallisation behaviour possible. Additionally, these laser devices enable new laser doping strategies by customizing the energy deposition per time, either by varying the pulse shape or by varying the pulse length as additional parameters. In the past usually only the pulse energy per area could be changed and the measured results were directly linked to the change of the laser parameters. In this way, the actual processes taking place during the laser irradiation were not considered. Contrary, in this thesis the cause-effect relationships between laser parameter and experimental results are split into a two-stage approach.<br />Firstly, the cause-effect relationships between the laser parameters and process characteristics are established. "Process characteristics" include the melt depth, the doping depth, the melt duration, the maximum temperature, the solidification velocity and the ablation depth. In the second step - not scope of this thesis - the connection between these simulation results and directly measurable doping results can be drawn.<br />The term "measurable doping results" refers to e.g. sheet resistance, defect concentration, the minority carrier life time, diode characteristics or the cell efficiency. The centrepiece of this procedure is that different laser pulses triggering similar processes lead to similar measurable results. This changes the view on laser doping from a "pulse picture" to a "process picture" by thinking in terms of the variation of the process characteristics instead of the laser parameters.<br />Therefore, the influence of energy deposition per time on the process characteristics is simulated with an equilibrium continuum phase-change model. The results demonstrate that the pulse length and the onset of evaporation significantly influence the doping process. Using long pulses with low intensity results in rectangular doping profiles, avoids evaporating the highly doped surface layers, lowers the crystallisation velocity especially for shallow melt spots and reduces thermal gradients in the bulk. In the next step the gained knowledge is used to tailor the energy deposition per time, i.e. to shape the pulse in a way to achieve desired process characteristics. Thereby a novel and more flexible method has been developed to slow down the crystallisation velocity, which is easy to implement in an industrial production environment.<br />Finally, a set of pulses was tailored so that only one process characteristic parameter is varied at a time. This approach enables detailed research on the significance of certain process characteristics on the experimental results. The feasibility of pulse-shaping to induce a desired doping process is proven and the tools for research on the effects of the process on the measured doping results are described.