Options
2012
Journal Article
Titel
Thermal simulation of pulsed direct laser interference patterning of metallic substrates using the smoothed particle hydrodynamics approach
Alternative
Thermische Simulation der direkten Pulslaser-Interferenztexturierung metallischer Substrate mit Hilfe des hydrodynamischen Ansatzes mit Partikelglättung
Abstract
A two-dimensional thermal model for the calculation of temperature distributions, including heat conduction and phase change, for single-pulsed direct laser interference patterning (DLIP) was described. The numerical model was solved by the meshfree smoothed particle hydrodynamics (SPH) method. To obtain a good computational accuracy, the chosen particle distribution must permit several particles within the very short absorption length of the material. The observed influence of the smoothing length on the model accuracy leads to the conclusion that a 21 particle neighbourhood in the particle approximation is suitable. This corresponds to a kernel smoothing length of the 2.4-fold particle spacing in case of an equidistant rectangular particle distribution. Numerical tests provide evidence that a particle discretisation by a local refinement approach is comparable to a fine equidistant discretisation in terms of accuracy. However, employing the locally refined particle distribution leads to a strong reduction of both computation time and memory capacity requirement. The obtained results are in good agreement with results computed using a previous finite element model. An advantage of the SPH thermal model is that the enthalpy discontinuity at the phase transition point is considered. In contrast, existing FEM models of single-pulsed DLIP spread the enthalpy of phase transition over a broader temperature interval, and therefore differences in computed temperature distributions become particularly apparent close to the vaporisation point. Experimental investigations of DLIP were performed on stainless steel substrates. The computations of the adapted model correspond very well to the molten pool dimensions measured and estimated on the basis of the experiments. It is uncertain, though, whether the evaporation noticed in the computational model actually takes place as convection in the molten bath is able to reduce the temperature of the melt considerably. To assess the extent of evaporation and hence the support of the melt displacement by the evaporation-induced recoil pressure the fluid flow in the molten bath has to be accounted for in the model. A noticeable benefit of the meshfree SPH method in contrast to conventional numerical methods is that deformations, tangling and even disintegration of the computational domain do not pose a serious problem. Therefore, the developed model is extendable to investigate the molten bath convection and evaporation during the DLIP process. Considering moderate fluence, the convection is of thermo-capillary nature, that is, strong thermal gradients at the surface of the substrate cause surface tension gradients that induce shear stresses leading to convection. At high fluence, convection is dominated by the evaporation-induced recoil pressure. To understand the mechanisms involved in single-pulsed DLIP of metallic substrates, the effects of molten bath convection and even ablation by material evaporation have to be assessed. Thus, future investigations on modelling DLIP using SPH will include the discussion of fluid flow and evaporation.