Femtosecond-laser sharp shaping of millimeter-scale geometries with vertical sidewalls

As femtosecond(fs)laser machining advances from micro/nanoscale to macroscale,approaches capable of machining macroscale geometries that sustain micro/nanoscale precisions are in great demand.In this research,an fs laser sharp shaping approach was developed to address two key challenges in macroscale machining(i.e.defects on edges and tapered sidewalls).The evolution of edge sharpness(edge transition width)and sidewall tapers were systematically investigated through which the dilemma of simultaneously achieving sharp edges and vertical sidewalls were addressed.Through decreasing the angle of incidence(AOI)from 0◦to−5◦,the edge transition width could be reduced to below 10µm but at the cost of increased sidewall tapers.Furthermore,by analyzing lateral and vertical ablation behaviors,a parameter-compensation strategy was developed by gradually decreasing the scanning diameters along depth and using optimal laser powers to produce non-tapered sidewalls.The fs laser ablation behaviors were precisely controlled and coordinated to optimize the parameter compensations in general manufacturing applications.The AOI control together with the parameter compensation provides a versatile solution to simultaneously achieve vertical sidewalls as well as sharp edges of entrances and exits for geometries of different shapes and dimensions.Both mm-scale diameters and depths were realized with dimensional precisions below 10µm and surface roughness below 1µm.This research establishes a novel strategy to finely control the fs laser machining process,enabling the fs laser applications in macroscale machining with micro/nanoscale precisions.

Error assessment of laser cutting predictions by semi-supervised learning

Experimentation data of perspex glass sheet cutting, using CO2 laser, with missing values were modelled with semi-supervised artificial neural networks. Factorial design of experiment was selected for the verification of orthogonal array based model prediction. It shows improvement in modelling of edge quality and kerf width by applying semi-supervised learning algorithm, based on novel error assessment on simulations. The results are expected to depict better prediction on average by utilizing the systematic randomized techniques to initialize the neural network weights and increase the number of initialization. Missing values handling is difficult with statistical tools and supervised learning techniques; on the other hand, semi-supervised learning generates better results with the smallest datasets even with missing values.