Optimizing Plasma Coating with Hyperspectral Snapshot Imaging
Plasma coating describes an industrial production process in which components of all kinds are coated with thin layers of different materials. To achieve high-quality results, it is essential to obtain a homogeneous layer based on a consistent composition. The information on the compounds of a plasma arc can potentially be retrieved using hyperspectral imaging.
However, plasma coating is a high-speed process that demands near real-time analysis, making it impossible for push-broom sensors to be used here. Hyperspectral snapshot cameras, by contrast, can contribute to such an optimization.
In a pre-study conducted by Cubert, the goal was to detect a spatial pattern in the plasma, which is characterized by different chemical compounds during vaporization. These compounds were to be visualized with the help of a hyperspectral snapshot camera.
Based on this information, the plasma coating process could be optimized, allowing automated reactions such as electronic readjustment of the entire process.
Setup
A hyperspectral FireflEYE S185 SE was placed with a fire-resistant blanket in a measurement chamber, targeting the electrodes used for the plasma coating at a diagonal angle. The optics were adjusted with an ND filter and a pinhole aperture to protect the optics of the hyperspectral camera.
The electrodes (diameter 1-1.5 mm) are coated steel wires consisting of several chemical substances, mainly various steel alloys. They are powered with strong electrical pulses, generating the plasma.
Collapse of the Plasma Arc
In a first test run, the hyperspectral snapshot camera was used to record the collapse of the plasma arc. For this, the integration time of the camera was set to 200 ms, allowing the transition from plasma, which is extremely bright, to the comparatively darker thermal annealing of the metal wires to be captured. The figure visualizes this process.
Four exemplary pseudo-RGB images (650, 500, 450 nm), taken at timestamps 0 ms, 333 ms, 667 ms, and 996 ms, show how the plasma arc collapses from the first to the last image. Below the RGB images, selected spectra (raw data) are shown, representing the spectral information of marked pixels (rectangles inside the RGBs).
The first image, showing active plasma, is overexposed as expected, but the areas in the other images that show the sweltering wires are also overexposed.
In the second image, the last plasma is driven out; thereafter, the electrodes cool down (images three and four). It is remarkable that a considerable proportion of the metal wire is vaporized, even after the electrodes have already been deactivated.
Detecting Different Compounds in the Plasma Coating Process
For the actual purpose of identifying different compounds within the plasma coating process, the hyperspectral camera was used to record a video with an integration time of 20 µs. In the following four exemplary measurements, all taken within one second (T1@415 ms, T2@500 ms, T3@725 ms, T4@864 ms), are analyzed.
The following figure again shows pseudo-RGB images (650, 500, 450 nm) of this measurement series. Besides the very bright plasma beam, both vaporized (bright) and liquid metal (dark) are visible. The spectra below the images show the spectral information of the green and red rectangles inside the RGBs.
Analysis
The analysis shows that during the welding process, the intensity of the spectra varies, as expected. Among all measurements, designated maximum peaks were identified, which vary in their intensity relative to each other but remain stable in their wavelength position. This implies the presence of individual spectral signatures of different chemical compounds that are vaporized within the plasma. To visualize this behavior, a common normalized difference index is used, which considers two selected bands with I = (b1–b2) / (b1+b2)
The following figure shows the relationship between the wavelengths 515 nm and 551 nm. The values of the pixels inside the images range from -0.33 to 0.3. A spatial pattern within the plasma clearly becomes visible.
When applying further wavelength peaks to the normalized index (in the following example 531 nm and 590 nm), as can be seen in the following figure, the pattern shows a different behavior than in the previous figure.
Conclusion
This indicates that by using a hyperspectral snapshot camera, it is possible to recognize and identify different chemical compounds within the plasma during the welding process at a later stage.
About the Author
Dr. Matthias Locherer has been the Sales Director at Cubert GmbH since 2017. With a PhD in Earth Observation from Ludwig Maximilian University of Munich, he brings extensive expertise in remote sensing, spectral imaging, and data analysis. Matthias has contributed to numerous research projects and publications, particularly in the multispectral monitoring of biophysical and biochemical parameters using hyperspectral satellite missions. His deep knowledge of optical measurement techniques and physical modeling makes him a key driver in advancing innovative hyperspectral technologies at Cubert.