FA: Development of a Python-based tool for the spatially resolved evaluation of X-ray images for determining concentration maps of solutes in ammonothermal experiments with optical cells

4. Dezember 2025

Manan Rupesh Nandwana –

In this project work the main focus is on developing a Python-based Image processing tool for analyzing X-ray images. The goal is simple to generate spatially resolved concentration maps of solutes during ammonothermal synthesis experiments. Growing high-quality gallium nitride (GaN) crystals is typically done using the ammonothermal method, which in-volves extreme heat and pressure. Studying the dissolution, transport, and growth inside the optical cells demands precise monitoring. The earlier observation methods gave us little to go on in those closed systems, as it is now very much needed to study different processes like concentration of solutes inside the opaque autoclave system. Now, though, by combin-ing X-ray absorption imaging with digital image processing, we can directly visualize these working mechanics and in real-time.
The tool itself takes 16-bit grayscale X-ray images and gets to work. It automatically pins down regions of interest, figures out the pixel intensity distributions, and then uses the Beer–Lambert law to turn that grayscale brightness into hard numbers. Specifically, quan-titative parameters like solute concentration and density variation. An Image Processing protocol was established using standard, powerful libraries like NumPy, OpenCV, Pillow, Matplotlib, and Pandas for efficient numerical computation, image handling, and clear graphical plotting. The process is systematic: the tool reads a sequence of images, identifies key geometric regions using techniques like the Hough Circle Transform, and computes average grayscale values. These values directly represent changes in material density or composition as the experiment progresses.
The output graphs are essential as they clearly show how grayscale intensity changes over time or with solute concentration. This establishes a clear connection between the underlying physical and chemicall processes and the visual variations.
By this method one can accurately track how quickly things are dissolving and crystals are growing inside the autoclave.This tool helps gain more knowledge that the whole framework is modular, reproducible, for different datasets or changing experiment settings. That guar-antees the data is both solid and reliable.
The raw X-ray images captured were then successfully turned into real, measurable scientific data by combining imaging, computation, and physical modeling. This change alone helps us understand how the solute acts in that supercritical ammonia, which directly translates into better ammonothermal reactor designs and the ability to dial in the perfect syn-thesis parameters for growing superior GaN crystals.

—