Artificial Intelligence to improve HPGe detector performance and reliability

Student: Thomas Wonderley
Supervisors: Andy Boston (UoL), Christopher Hurt (Mirion)
Institution: University of Liverpool

High-Purity Germanium (HPGe) detectors represent the pinnacle of spectroscopic performance and provide the most sensitive and precise measurement of gamma radiation. For this reason, HPGe detectors play a critical role in a range of applications such as scientific research (including nuclear physics and space science), environmental sample measurements (e.g. food, water, soil, air) for public safety, national security applications, including the International Monitoring Network that checks for prohibited weapons testing, and as part of measurement systems supporting nuclear decommissioning and site remediation (e.g. Fukushima). Therefore, these systems are mission-critical and are required to operate successfully on-demand.

HPGe detectors are complex instruments pushing the limits of material science, surface chemistry, high performance instrumentation and nuclear physics. Many production parameters have a significant influence on final sensor performance. Identifying, analysing, and controlling these variables are essential for managing production throughput with stable quality while meeting on-time customer delivery and performance expectations.

The student will investigate using artificial intelligence techniques such as machine learning to gain more predictability regarding the viability of crystals during the manufacturing process, to improve detector performance in the field and increase the lifetime of the detector by monitoring key features through the full detector lifecycle. The project will require experimental measurements, advanced data analysis and will involve working with academic and industrial stakeholders.