Predicting algorithm for laser processing parameters using spectral-domain optical coherence tomography

Abstract

Laser processing is widely used in the industrial field, nonetheless, there is a tremendous amount of time required for prototyping products using multiple parameters such as laser power and number of laser pulses, which is necessary to be adjusted based on targeted material, shape, and environment. Optical coherence tomography (OCT) using near-infrared light, which has the ability to measure depth of laser processed holes or grooves, is a promising imaging technique for reducing time for trial-and-error. We explored the possibilities of developing a predicting algorithm for laser processing parameters using a spectral-domain OCT (SD-OCT) with a center wavelength of 900 nm. Sapphire, copper, silica, aluminum, and brass were processed by our pulse laser with a wavelength of 1030 nm, pulse width of 290 femtoseconds (fs), and repetition rate of 100 kHz. Different holes with laser powers from 5 to 20 W and pulse cumulative numbers from 10 to 100 counts were fabricated. Diameters and depths of holes were measured by a SD-OCT as well as a laser scanning microscope as a reference. The measured results using a SD-OCT and a laser microscope were compared, and the averaged difference was less than 19.0 %. Next, a predicting plane, relating laser processing parameters consisting of laser power and cumulative laser pulses to the measured parameters consisting of diameter and depth, was studied. The predicting relational plane obtained by a Sapphire sample predicted the hole shapes of copper, silica, aluminum, and brass within 30.1% of coefficients of variation (<i>cv</i>). In addition, the SD-OCT system imaged clear microcracks in sapphire samples that were not captured by the laser microscope. These results indicate that an integrated system of the OCT and a laser processing machine has promise for predicting laser processing parameters and reducing time for trial-and-error.