Significant enhancment of the accuracy of impurity determination in vacuums using classification one-point...

Laser-induced breakdown spectroscopy (LIBS) is a highly promising technique for the in-situ, real-time diagnosis of impurity deposits on the inner walls of tokamak devices. The deposited impurity on plasma-facing materials (PFCs) pose a significant risk to the steady-state operation of the tokamak. Under vacuum conditions, an accurate quantitative analysis of the thin co-deposition layers is a technical challenge. In this study, 30 co-deposited layer samples of tungsten (10.0-92.3 a.t.%), molybdenum (2.0-77.8 a.t.%), iron (2.9-12.1 a.t.%) and copper (1.2-18.7 a.t.%) were prepared to simulate the co-deposition layers found on PFCs in Experimental Advanced Superconducting Tokamak (EAST). A variation of the CF-LIBS algorithm, the so-called One Point Calibrated LIBS (OPC-LIBS), was employed to analyze these co-deposited layer samples under conditions of 5 × 10^-5 mbar. It was found that the matrix matching degree among the measured samples and the selection of standard samples play a decisive role in the quantitative analysis capability of OPC-LIBS. In actual situations, the composition of the co-deposited impurity layers at different locations in the Tokamak will be quite different. We addressed this challenge by developing the Classified OPC-LIBS (COPC-LIBS) model, an enhanced version of OPC-LIBS with pre-classification to offset matrix effects in LIBS analysis. For tungsten in the co-deposition layers, the root mean square (RMSE) calculated by the CF-LIBS method was 14.7, the OPC-LIBS method was 11.5, and the newly invented COPC-LIBS was reduced to only 5.1. The COPC-LIBS method is a highly efficient technique that can precisely measure the distribution of co-deposited layers on the surface of inner wall materials. The diagnostic data obtained from this method will provide valuable insights into the interaction between plasma and wall materials during the operation of fusion devices.