Scientists from Far Eastern Federal University (FEFU), in collaboration with colleagues from Harbin Institute of Technology (HIT), have developed a method for applying colored marks to steel using a laser controlled by a neural network. Unlike conventional coatings, such a mark is formed directly in the structure of the metal surface in the form of nano-grooves and cannot be erased.
The technology allows applying images of different colors to stainless steel due to light diffraction. In other words, the laser forms such a fine regular pattern on the surface that the metal begins to split light differently and acquires the desired shade.
With multiple laser pulses, the smooth surface is covered with many identical nano-grooves. The distance between them determines the wavelength and, accordingly, the color. Their inclination sets the direction and intensity of diffraction.
The main difficulty was that it was not possible to precisely control the distance between the grooves and their inclination simultaneously for a long time. These parameters depend on the angle at which the laser hits the metal and its polarization. Any error could result in only a dull ripple instead of a bright pattern.
To solve the problem, the scientists assembled a multi-coordinate laser system with five axes of movement. It allows changing the beam's angle of incidence from minus 30°C to plus 30°C, and the polarization angle from 0° to 180°. However, even this was not enough for the required accuracy due to side optical effects.
Then the researchers connected two neural networks. The first optimized the main parameters of laser processing: pulse energy, number of pulses, and step between scanning lines. This allowed obtaining more even grooves.
The second neural network was trained on 130 experimental datasets. It established a connection between the angle of incidence and polarization of the laser, on the one hand, and the resulting period and orientation of nanostructures, on the other.
As a result, the spread in groove orientation was reduced to 6.5 degrees. With a random selection of parameters, it averaged 47 degrees, which resulted in dull colors.
Scientists from Far Eastern Federal University, in collaboration with colleagues from Harbin Institute of Technology, have developed a method for applying colored marks to steel using a laser controlled by a neural network. Unlike conventional coatings, such a mark is formed directly in the structure of the metal surface in the form of nano-grooves and cannot be erased.
The technology allows applying images of different colors to stainless steel due to light diffraction. In other words, the laser forms such a fine regular pattern on the surface that the metal begins to split light differently and acquires the desired shade.
With multiple laser pulses, the smooth surface is covered with many identical nano-grooves. The distance between them determines the wavelength and, accordingly, the color. Their inclination sets the direction and intensity of diffraction.
The main difficulty was that it was not possible to precisely control the distance between the grooves and their inclination simultaneously for a long time. These parameters depend on the angle at which the laser hits the metal and its polarization. Any error could result in only a dull ripple instead of a bright pattern.
To solve the problem, the scientists assembled a multi-coordinate laser system with five axes of movement. It allows changing the beam's angle of incidence from minus 30°C to plus 30°C, and the polarization angle from 0° to 180°. However, even this was not enough for the required accuracy due to side optical effects.
Then the researchers connected two neural networks. The first optimized the main parameters of laser processing: pulse energy, number of pulses, and step between scanning lines. This allowed obtaining more even grooves.
The second neural network was trained on 130 experimental datasets. It established a connection between the angle of incidence and polarization of the laser, on the one hand, and the resulting period and orientation of nanostructures, on the other.
As a result, the spread in groove orientation was reduced to 6.5 degrees. With a random selection of parameters, it averaged 47 degrees, which resulted in dull colors.
Development tested in practice
The scientists created a multi-colored image on steel in the form of a swirl of five colored areas. For each zone, the neural network calculated the necessary angles.
Electron microscopy confirmed the accuracy of the result: the structure period differed from the calculated one by no more than 20 nm, and the orientation by 0.6 degrees. The resulting colors almost matched the planned ones.
According to Alexander Kuchmizhak, an employee of FEFU and IAPU FEB RAS, this approach can be applied not only to stainless steel but also to other metals and semiconductors. The five-axis scheme will also allow creating similar structures not only on flat surfaces but also on complex-shaped parts in the future.
The technology is proposed for use in protective holographic marks, optical elements, and biocompatible surfaces. The research results are published in the journal ACS Applied Materials & Interfaces.
The scientists created a multi-colored image on steel in the form of a swirl of five colored areas. For each zone, the neural network calculated the necessary angles.
Electron microscopy confirmed the accuracy of the result: the structure period differed from the calculated one by no more than 20 nm, and the orientation by 0.6 degrees. The resulting colors almost matched the planned ones.
According to Alexander Kuchmizhak, an employee of FEFU and IAPU FEB RAS, this approach can be applied not only to stainless steel but also to other metals and semiconductors. The five-axis scheme will also allow creating similar structures not only on flat surfaces but also on complex-shaped parts in the future.
The technology is proposed for use in protective holographic marks, optical elements, and biocompatible surfaces. The research results are published in the journal ACS Applied Materials & Interfaces.
Read more on the topic:
- Russian scientists create world's first current-driven laser: a step towards quantum chips
- Mayak Plant Introduces Modern Plotter for the First Time: Laser Becomes a Thing of the Past
- This laser won't let the doctor make a mistake: new technology from NRNU MEPhI automatically protects tissues during stone removal